Академический Документы
Профессиональный Документы
Культура Документы
Have you lost track of developments in generative linguistics, fnding yourself unsure about
the distinctive features of Minimalism? Would you like to know more about recent advances in
the genetics of language, or about right hemisphere linguistic operation? Has your interest in
narrative drawn you to question the relation between stories and grammars? Te Cambridge
Encyclopedia of the Language Sciences addresses these issues, along with hundreds of others.
It includes basic entries for those unfamiliar with a given topic and more specifc entries for
those seeking more specialized knowledge. It incorporates both well-established fndings and
cutting-edge research as well as classical approaches and new theoretical innovations. Te vol-
ume is aimed at readers who have an interest in some aspect of language science but wish to
learn more about the broad range of ideas, fndings, practices, and prospects that constitute
this rapidly expanding feld, a feld arguably at the center of current research on the human
mind and human society.
Patrick Colm Hogan is a professor in the Department of English and the Program in Cognitive
Science at the University of Connecticut. He is the author of ten books, including Cognitive
Science, Literature, and the Arts: A Guide for Humanists and Te Mind and Its Stories: Narrative
Universals and Human Emotion (Cambridge University Press, 2003).
For both range and depth of exposition and commentary on the diverse disciplinary angles
that exist on the nature of language, there is no single volume to match this fne work of
reference.
Akeel Bilgrami, Columbia University
Te Cambridge Encyclopedia of the Language Sciences is a very welcome addition to the feld
of language sciences. Its comprehensiveness is praiseworthy, as is the quality of its entries and
discussions.
Seymour Chatman, University of California, Berkeley
Tis ambitious and comprehensive work, and the very high quality of the editors and con-
tributors, ensure that it will be a valuable contribution to the understanding of language and
its uses, for both professionals and a more general audience.
Noam Chomsky, Massachusetts Institute of Technology
Advance Praise for
The Cambridge Encyclopedia of the Language Sciences
THE CAMBRIDGE ENCYCLOPEDIA OF
THE LANGUAGE SCIENCES
Edited by
PATRICK COLM HOGAN
University of Connecticut
CAMBRI DGE UNI VERSI TY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore,
So Paulo, Delhi, Dubai, Tokyo, Mexico City
Cambridge University Press
32 Avenue of the Americas, New York, NY 10013-2473, USA
www.cambridge.org
Information on this title: www.cambridge.org/9780521866897
Cambridge University Press 2011
Tis publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2011
Printed in the United States of America
A catalog record for this publication is available from the British Library.
Library of Congress Cataloging in Publication data
Te Cambridge encyclopedia of the language sciences / edited by Patrick Colm Hogan.
p. cm.
Includes bibliographical references and index.
ISBN 978-0-521-86689-7 (hardback)
1. Linguistics Encyclopedias. I. Hogan, Patrick Colm. II. Title.
P 29. C 28 2009
410.3dc22 2008041978
ISBN 978-0-521-86689-7 Hardback
Cambridge University Press has no responsibility for the persistence or
accuracy of URL s for external or third-party Internet Web sites referred to in
this publication and does not guarantee that any content on such Web sites is,
or will remain, accurate or appropriate.
GENERAL EDITOR
Patrick Colm Hogan
University of Connecticut, Storrs
ADVISORY EDITORIAL BOARD
Florian Coulmas Barbara Lust
German Institute of Japanese Studies Cornell University
and Duisberg-Essen University
William Croft Lee Osterhout
University of New Mexico University of Washington
Lyle Jenkins James Pustejovsky
Biolinguistics Institute Brandeis University
CONSULTING EDITORIAL BOARD
Mark Baker Howard Lasnik
Rutgers University University of Maryland
Deborah Cameron Loraine Obler
University of Oxford City University of New York
Nigel Fabb William OGrady
University of Strathclyde University of Hawaii
Carol Ann Fowler Susan Pintzuk
Haskins Laboratories University of York
and University of Connecticut
Ronald Geluykens Eleanor Rosch
University of Oldenburg University of California, Berkeley
Margaret Harris Jay Rueckl
Oxford Brookes University University of Connecticut
Zoltn Kvecses Mark Turner
Etvs Lornd University Case Western Reserve University
To the memory of B. N. Pandit (19162007)
philosopher, Sanskritist, father-in-law
Purua-artha-nyn gun pratiprasava kaivalya
sva-rpa-pratih v citi-aktir-iti
Patajali
ix
List of Entries page xi
A Note on Cross-References and the Alphabetization of the Entries xv
Preface: On the Very Idea of Language Sciences xvii
Acknowledgments xxiii
1 Language Structure in Its Human Context: New Directions
for the Language Sciences in the Twenty-First Century 1
William Croft
2 The Psychology of Linguistic Form 12
Lee Osterhout, Richard A. Wright, and Mark D. Allen
3 The Structure of Meaning 23
James Pustejovsky
4 Social Practices of Speech and Writing 35
Florian Coulmas
5 Explaining Language: Neuroscience, Genetics, and Evolution 46
Lyle Jenkins
6 Acquisition of Language 56
Barbara Lust
7 Elaborating Speech and Writing: Verbal Art 65
Patrick Colm Hogan
ENTRIES 77
List of Contributors 941
Index 953
CONTENTS
xi
ENTRIES
A
Abduction
Absolute and Statistical Universals
Accessibility Hierarchy
Acoustic Phonetics
Adaptation
Ad Hoc Categories
Adjacency Pair
Age Groups
Aging and Language
Agreement
Agreement Maximization
Alliteration
Ambiguity
Amygdala
Analogy
Analogy: Synchronic and Diachronic
Analyticity
Anaphora
Animacy
Animal Communication and Human
Language
Aphasia
Areal Distinctness and Literature
Art, Languages of
Articulatory Phonetics
Artifcial Languages
Aspect
Auditory Processing
Autism and Language
Autonomy of Syntax
B
Babbling
Basal Ganglia
Basic Level Concepts
Bilingual Education
Bilingualism, Neurobiology of
Bilingualism and Multilingualism
Binding
Biolinguistics
Birdsong and Human Language
Blended Space
Blindness and Language
Bounding
Brain and Language
Brocas Area
C
Cartesian Linguistics
Case
Categorial Grammar
Categorization
Causative Constructions
C-Command
Cerebellum
Charity, Principle of
Childrens Grammatical Errors
Chirographic Culture
Clitics and Cliticization
Codeswitching
Cognitive Architecture
Cognitive Grammar
Cognitive Linguistics and Language
Learning
Cognitive Linguistics, Language Science,
and Metatheory
Cognitive Poetics
Coherence, Discourse
Coherence, Logical
Colonialism and Language
Color Classifcation
Communication
Communication, Prelinguistic
Communicative Action
Communicative Intention
Comparative Method
Competence
Competence and Performance, Literary
Compositionality
Computational Linguistics
Concepts
Conceptual Blending
Conceptual Development and Change
Conceptual Metaphor
Conduit Metaphor
Connectionism and Grammar
Connectionism, Language Science, and
Meaning
Connectionist Models, Language
Structure, and Representation
Consciousness and Language
Consistency, Truth, and Paradox
Constituent Structure
Constraints in Language Acquisition
Construction Grammars
Contact, Language
Context and Co-Text
Control Structures
Conversational Implicature
Conversational Repair
Conversation Analysis
Cooperative Principle
Core and Periphery
Corpus Callosum
Corpus Linguistics
Creativity in Language Use
Creoles
Critical Discourse Analysis
Critical Periods
Culture and Language
Cycle, Te
D
Deconstruction
Defnite Descriptions
Deixis
Descriptive, Observational, and
Explanatory Adequacy
Dhvani and Rasa
xii
List of Entries
Dialect
Dialogism and Heteroglossia
Difusion
Digital Media
Diglossia
Discourse Analysis (Foucaultian)
Discourse Analysis (Linguistic)
Discrete Infnity
Disorders of Reading and Writing
Division of Linguistic Labor
Dyslexia
E
Ellipsis
Embodiment
Emergentism
Emergent Structure
Emotion and Language
Emotion, Speech, and Writing
Emotion Words
Emplotment
Encoding
nonc/Statement (Foucault)
Essentialism and Meaning
Ethics and Language
Ethnolinguistic Identity
Event Structure and Grammar
Evidentiality
Evolutionary Psychology
Exemplar
Exemplar Teory
Extinction of Languages
F
Family Resemblance
Feature Analysis
Felicity Conditions
Field (Bourdieu)
Film and Language
Filters
Focus
Foregrounding
Forensic Linguistics
Formal Semantics
Forms of Life
Frame Semantics
Framing Efects
Frontal Lobe
Functional Linguistics
G
Games and Language
Gender and Language
Gender Marking
Generative Grammar
Generative Poetics
Generative Semantics
Generic- and Specifc-Level Metaphors
Genes and Language
Gesture
Government and Binding
Grammaticality
Grammaticality Judgments
Grammaticalization
Grooming, Gossip, and Language
H
Habitus, Linguistic
Head-Driven Phrase Structure Grammar
Hippocampus
Historical Linguistics
Historical Reconstruction
Holophrastic Stage, Te
Homologies and Transformation Sets
I
Icon , Index , and Symbol
Ideal Speech Situation
Identity, Language and
Ideology and Language
Idioms
Idle Talk and Authenticity
Ijtihd (Interpretive Efort)
I-Language and E-Language
Illocutionary Force and Sentence Types
Image Schema
Implicational Universals
Indeterminacy of Translation
Indexicals
Inequality, Linguistic and
Communicative
Infantile Responses to Language
Information Structure in Discourse
Information Teory
Innateness and Innatism
Integrational Linguistics
Intension and Extension
Intentionality
Internal Reconstruction
Interpretation and Explanation
Interpretive Community
Intertextuality
Intonation
Irony
L
Language, Natural and Symbolic
Language Acquisition Device (LAD)
Language Change, Universals of
Language Families
Language-Game
Language-Learning Environment
Language of Tought
Language Policy
Laws of Language
Learnability
Left Hemisphere Language Processing
Legal Interpretation
Lexical Acquisition
Lexical-Functional Grammar
Lexical Learning Hypothesis
Lexical Processing, Neurobiology of
Lexical Relations
Lexical Semantics
Lexicography
Linguistic Relativism
Literacy
Literariness
Literary Character and Character Types
Literary Universals
Literature, Empirical Study of
Logic and Language
Logical Form
Logical Positivism
M
Mapping
Markedness
Market, Linguistic
Marxism and Language
Meaning and Belief
Meaning and Stipulation
Meaning Externalism and Internalism
Media of Communication
Memes and Language
Memory and Language
Mental Models and Language
Mental Space
Merge
Metalanguage
Metaphor
Metaphor, Acquisition of
Metaphor, Information Transfer in
Metaphor, Neural Substrates of
Metaphor, Universals of
Meter
Methodological Solipsism
Methodology
Metonymy
Minimalism
Mirror Systems, Imitation, and
Language
Modality
Modern World-System, Language and
the
Modularity
Montague Grammar
Mood
Morpheme
Morphological Change
Morphological Typology
Morphology
Morphology, Acquisition of
Morphology, Evolution and
xiii
List of Entries
Morphology, Neurobiology of
Morphology, Universals of
Motif
Movement
Music, Language and
N
Narrative, Grammar and
Narrative, Neurobiology of
Narrative, Scientifc Approaches to
Narrative Universals
Narratives of Personal Experience
Narratology
Nationalism and Language
Natural Kind Terms
Necessary and Suf cient Conditions
Negation and Negative Polarity
Network Teory
Neurochemistry and Language
Neuroimaging
Number
O
Occipital Lobe
Optimality Teory
Oral Composition
Oral Culture
Ordinary Language Philosophy
Origins of Language
Overregularizations
P
Parable
Paralanguage
Parameters
Parietal Lobe
Parsing, Human
Parsing, Machine
Passing Teories
Performance
Performative and Constative
Perisylvian Cortex
Perlocution
Person
Philology and Hermeneutics
Phoneme
Phonetics
Phonetics and Phonology, Neurobiology
of
Phonological Awareness
Phonology
Phonology, Acquisition of
Phonology, Evolution of
Phonology, Universals of
Phrase Structure
Pidgins
Pitch
Poetic Form, Universals of
Poetic Language, Neurobiology of
Poetic Metaphor
Poetics
Point of View
Politeness
Politics of Language
Possible Worlds Semantics
Possible Worlds Semantics and Fiction
Pragmatic Competence, Acquisition of
Pragmatics
Pragmatics, Evolution and
Pragmatics, Neuroscience of
Pragmatics, Universals in
Pragmatism and Language
Predicate and Argument
Preference Rules
Prestige
Presupposition
Primate Vocal Communication
Priming, Semantic
Principles and Parameters Teory
Principles and Parameters Teory and
Language Acquisition
Print Culture
Private Language
Projectibility of Predicates
Projection (Blending Teory)
Projection Principle
Proposition
Propositional Attitudes
Prototypes
Proverbs
Psychoanalysis and Language
Psycholinguistics
Psychonarratology
Psychophysics of Speech
Q
Qualia Roles
Quantifcation
Quantitative Linguistics
R
Radical Interpretation
Reading
Realization Structure
Rectifcation of Names ( Zheng Ming )
Recursion , Iteration , and
Metarepresentation
Reference and Extension
Reference Tracking
Register
Regularization
Relevance Teory
Religion and Language
Representations
Rhetoric and Persuasion
Rhyme and Assonance
Rhythm
Right Hemisphere Language Processing
Role and Reference Grammar
Rule-Following
S
Schema
Scripts
Second Language Acquisition
Self-Concept
Self-Organizing Systems
Semantic Change
Semantic Fields
Semantic Memory
Semantic Primitives (Primes)
Semantics
Semantics, Acquisition of
Semantics, Evolution and
Semantics, Neurobiology of
Semantics, Universals of
Semantics-Phonology Interface
Semantics-Pragmatics Interaction
Semiotics
Sense and Reference
Sentence
Sentence Meaning
Sexuality and Language
Signed Languages, Neurobiology of
Sign Language, Acquisition of
Sign Languages
Sinewave Synthesis
Situation Semantics
Socially Distributed Cognition
Sociolinguistics
Source and Target
Specifc Language Impairment
Speech-Acts
Speech Anatomy, Evolution of
Speech-Language Pathology
Speech Perception
Speech Perception in Infants
Speech Production
Spelling
Spreading Activation
Standardization
Standard Teory and Extended Standard
Teory
Stereotypes
Story and Discourse
Story Grammar
Story Schemas, Scripts , and Prototypes
Stress
Structuralism
Stylistics
Stylometrics
Subjacency Principle
Suggestion Structure
Syllable
xiv
List of Entries
Synchrony and Diachrony
Syntactic Change
Syntax
Syntax, Acquisition of
Syntax, Evolution of
Syntax, Neurobiology of
Syntax, Universals of
Syntax-Phonology Interface
T
Teaching Language
Teaching Reading
Teaching Writing
Temporal Lobe
Tense
Text
Text Linguistics
Talamus
Tematic Roles
Teory of Mind and Language Acquisition
Tone
Topicalization
Topic and Comment
Traces
Transformational Grammar
Translation
Truth
Truth Conditional Semantics
Two-Word Stage
Typology
U
Underlying Structure and Surface
Structure
Universal Grammar
Universal Pragmatics
Universals, Nongenetic
Usage-Based Teory
Use and Mention
V
Vagueness
Verbal Art, Evolution and
Verbal Art, Neuropsychology of
Verbal Display
Verbal Humor
Verbal Humor, Development of
Verbal Humor, Neurobiology of
Verbal Reasoning
Verbal Reasoning, Development of
Verifability Criterion
Verse Line
Voice
Voice Interaction Design
W
Wernickes Area
Word Classes (Parts of Speech)
Word Meaning
Word Order
Word Recognition, Auditory
Word Recognition, Visual
Words
Working Memory and Language
Processing
Writing, Origin and History of
Writing and Reading, Acquisition of
Writing and Reading, Neurobiology of
Writing Systems
X
X-Bar Teory
xv
A NOTE ON CROSS-REFERENCES AND THE ALPHABETIZATION
OF THE ENTRIES
Cross-references are signaled by small capitals (boldface when implicit). Tey are designed
to indicate the general relevance of the cross-referenced entry and do not necessarily imply
that the entries support one another. Note that the phrasing of the cross-references does not
always match the entry headings precisely. In order to minimize the disruption of reading,
entries often use shortened forms of the entry headings for cross-references. For example, this
process involves parietal structures points to the entry Parietal Lobe. In some cases, a
cross-reference may refer to a set of entries. For example, architectures of this sort are found
in connectionism alerts the reader to the presence of entries on connectionism generally,
rather than to a single entry. Finally, a cross-reference may present a heading in a diferent
word order. For example, the target entry for here we see another universal of phonol-
ogy would be listed as Phonology, Universals of.
In general, entries with multiword headings are alphabetized under their main language
term. Tus, the entry for Universals of Phonology is listed as Phonology, Universals of. Te
main exceptions to this involve the words language and linguistic or linguistics , where another
term in the heading seemed more informative or distinctive in the context of language sciences
(e.g., Linguistic Market is listed as Market, Linguistic).
xvii
a complex of principles that roughly defne scientifc method.
Tese principles do not apply in the same way to chemical
interactions and group relations and that is one reason why
narrow demarcation criteria fail. However, they are the same
general principles across diferent domains. Very simply,
scientifc method involves inter alia the following practices:
1) the systematic study of empirically ascertainable patterns
in a given area of research; 2) the formulation of general prin-
ciples that govern those patterns; 3) the attempt to uncover
cases where these principles do not govern observed pat-
terns; 4) the attempt to eliminate gaps, vagueness, ambiguity,
and the like from ones principles and from the sequences of
principles and data that produce particular explanations; and
5) the attempt to increase the simplicity of ones principles
and particular explanations. Discourses are scientifc to the
extent that they routinely involve these and related practices.
Note that none of this requires, for example, strict falsifca-
tion or detailed prediction. For example, social phenomena
are most often too complex to allow for signifcant prediction,
in part because one cannot gather all the relevant data before-
hand. Tis does not mean that they are closed to systematic
explanations after the fact, as more data become available .
Of course, following such methodological guidelines is not
all there is to the actual practice of science. Tere are always
multiple options for formulating general principles that ft
the current data. Te evaluation of simplicity is never entirely
straightforward. Teories almost invariably encounter anom-
alous data in some areas and fail to examine other areas.
Moreover, in many cases, the very status of the data is unclear.
Despite all this, we hierarchize theories. We teach some and do
not teach others. Agencies fund some and do not fund others.
Te very nature of the enterprise indicates that even in ideal cir-
cumstances, this cannot be purely meritocratic. Moreover, real
circumstances are far from ideal. Tus, in the real world, adher-
ence to methodological principles may be very limited (see, for
example, Faust 1984, Mahoney 1977, and Peters and Ceci 1982).
Tis is where social constructionism enters. It seems undeni-
able that relations of institutional power, the political economy
of professions, and ideologies of nation or gender guide what is
institutionalized, valued, funded, and so forth .
A title referring to language sciences tacitly raises at least three
questions. First, what is a science? Second, what is language?
Finally, what is a language science? I cannot propose answers
to these questions in a short preface. Moreover, it would not be
appropriate to give answers here. Te questions form a sort of
background to the essays and entries in the following pages,
essays and entries that often difer in their (explicit or implicit)
answers. However, a preface of this sort can and should
indicate the general ideas about science and language that
governed the development of Te Cambridge Encyclopedia of
the Language Sciences.
WHAT IS SCIENCE?
Philosophers of science have often been concerned to defne a
demarcation criterion, separating science from nonscience. I
have not found any single criterion, or any combination of cri-
teria, compelling in the sense that I have not found any argu-
ment that, to my satisfaction, successfully provides necessary
and sufficient conditions for what constitutes a science.
In many ways, ones acceptance of a demarcation criterion is
guided by what one already considers to be a science. More
exactly, ones formulation of a demarcation criterion tends to
be a function of what one takes to be a paradigmatic science or,
in some cases, an exemplary case of scientifc practice.
Te advocates of strict demarcation criteria meet their mir-
ror opposites in writers who assert the social construction of
science, writers who maintain that the diference between
science and nonscience is simply the diference between dis-
tinct positions within institutions, distinct relations to power.
Suppose we say that one discipline or theory is a science and
another is not. Tis is just to say that the former is treated as
science, while the latter is not. Te former is given authority in
academic departments, in relevant institutions (e.g., banks, in
the case of economics), and so on.
Again, this is not the place for a treatise on the philosophy
of science. Here it is enough to note that I believe both sides
are partially correct and partially incorrect. First, as already
noted, I do not believe that there is a strict, defnitive demar-
cation criterion for science. However, I do believe that there is
PREFACE: ON THE VERY IDEA OF LANGUAGE SCIENCES
xviii
Preface
most important architectures of the latter sort are found in
connectionism .
As a wide range of writers have stressed, the distinctive fea-
ture of mind our second level of analysis is intentional-
ity. However, intentionality, as subjective and experiential, is
often not well suited for scientifc study. Many theorists have
therefore sought to systematize and objectify our understand-
ing of mind. Most cognitive treatments of the mental level
have their roots in folk psychology, a minimal, common-
sense objectifcation of intention in terms of beliefs and aims.
But these cognitive treatments draw on empirical research
and principles of scientifc method to develop models of the
human mind that are sometimes very far removed from folk
psychology. Specifcally, they most often replace belief by
mental representations and algorithmically specifable
operations on those representations. We may therefore refer to
these models as representational. Representationalism serves
to make intention more tractable through a mentalistic archi-
tecture that is precisely articulated in its structures, processes,
and contents.
Finally, our treatment of societies may be loosely divided
into the more intentional or mental pole of individual inter-
action and the less subjective, more broadly statistical pole of
group dynamics. (See Figure 1 .)
Tese divisions apply to language no less than they apply
to other areas of human science. We draw on our represen-
tational account of syntax to understand certain brain pro-
cesses in the perisylvian cortex. Conversely, we explain
In putting together a volume on science, then, I have tried
to incorporate the insights of both the more positive views
of science and the more social constructionist views. Put in
a way that may seem paradoxical, I have tried to include all
approaches that ft the loose criteria for science just mentioned.
I believe that these loose criteria apply not only to paradigmatic
sciences themselves but also to many social critiques of sci-
ence that stress social construction. I have therefore included a
wide range of what are sometimes called the human sciences.
Indeed, the volume could be understood as encompassing the
language-relevant part of the human sciences which leads to
our second question .
WHAT IS LANGUAGE?
Like science, ones defnition of language depends to a great
extent on just what the word calls to mind. Ones view of lan-
guage is likely to vary if one has in mind syntax or seman-
tics, hearers or speakers, dialogues or diaries, brain damage
or propaganda, storytelling or acoustic phonetics. A frst
impulse may be to see one view of language as correct and the
others as false. And, of course, some views are false. However,
I believe that our understanding of language can and, indeed,
should sustain a great deal of pluralism.
In many ways, my own paradigm for human sciences is cog-
nitive science. Cognitive science brings together work from a
remarkable array of disciplines literally, from Anthropology
to Zoology. Moreover, it sustains a range of cognitive archi-
tectures, as well as a range of theories within those architec-
tures. Tus, it is almost by its very nature pluralistic. Of course,
some writers wish to constrain this pluralism, insisting that one
architecture is right and the others are wrong. Certainly, one
can argue that particular architectures are wrong. However,
perhaps the most noteworthy aspect of cognitive science is
that it sustains a series of types of cognitive architecture. In
Cognitive Science, Literature, and the Arts (2003), I argued that
these types capture patterns at diferent levels of analysis. Tus,
all are scientifcally valuable.
More exactly, we may distinguish three levels of cognitive
analysis: bodies, minds, and groups or societies. Tese lev-
els stand in a hierarchical relation such that bodies are more
explanatorily basic than minds, and minds are more explana-
torily basic than groups or societies. Lower levels provide
causally necessary principles for higher levels. Minds do not
operate without brains. People without minds do not interact
in groups. In other words, lower levels explain higher levels.
However, higher-level patterns provide interpretive principles
for understanding lower levels (see interpretation and
explanation). We explain the (mental) feeling of fear by ref-
erence to the (bodily) operation of the amygdala. But, at the
same time, we understand amygdala activity as fear because
we interpret that activity in terms of the mental level.
In the analysis of cognition, the most basic, bodily cogni-
tive architecture is provided by neurobiology. However, due
to the intricate particularity of neurobiology, we often draw
on more abstract associative models at this level. Tese mod-
els serve to make the isolation and explanation of patterns
less computationally complex and individually variable. Te
Body
Associative Networks
Brains
Society
Mental Representations
Intentions
Mind
Group Dynamics
Individual Interactions
Figure 1. Levels of cognitive analysis. Between the levels, black
arrows represent the direction of explanation, while hollow arrows repre-
sent the direction of interpretation. Within the levels, the superior items
are more computationally tractable or algorithmically specifable models
of the inferior items, either singly (in the case of brains and intentions)
or collectively (in the case of individual interactions). Tractability may be
produced by simplifcation (as in the case of bodily architectures), by
systematic objectifcation (as in the case of mental architectures), or by
statistical abstraction (as in the case of social analysis).
xix
Preface
essays. However, it was not the only factor. In language sciences,
and indeed in human sciences generally, we need to add two
further considerations. Te preceding analysis implicitly treats
language patterns as if they are comparable to any patterns
isolated in the natural sciences. However, there are two difer-
ences between patterns in language and, say, the patterns iso-
lated by physicists. First, language patterns are mutable. Tey
are mutable in three ways at the level of groups, at the level of
individual minds or brains, and at the level of common genetic
inheritance. Insofar as language patterns change at the level
of groups, this mutability is comprehended by group dynam-
ics and related processes (most obviously in historical lin-
guistics). But mental and neurobiological theories do not
necessarily treat the other two sorts of mutability, for such the-
ories tend to focus on steady states of language. We therefore
account for changes in the individual mind or brain by refer-
ence to development or acquisition (see phonology, acqui-
sition of; syntax, acquisition of; and so on). We account
for changes in common genetic properties through the evolu-
tion of language (see phonology, evolution of; syntax,
evolution of; and so on) .
Te second diference between patterns in language and pat-
terns isolated by physicists is related to the frst. Just as we may
be insuf cient in language, we may be more than suf cient. In
other words, there is a diference between ordinary usage and
skilled usage. Rocks do not fall well or badly. Tey simply fall,
and they do so at the same rate. People, however, speak well
or badly, efectively or inefectively, in a manner that is clichd
or unusually creative (see creativity in language use).
Te point is most obvious in verbal art which leads us to the
most sweet and pleasing sciences of poetry, as Cervantes put
it (1950, 426) .
In keeping with the preceding analysis, then, the main top-
ics in language science are treated initially in a series of seven
overview essays. Te frst essay provides a general introduc-
tion to the study of language. Its purpose is to orient readers
toward the feld as a whole. Te second and third essays turn to
the mental level of language since this is the most widely ana-
lyzed. Due to the amount of work in this area, and due to the
diversity of approaches, the treatment of this level is divided
into two chapters. Te frst addresses formal aspects of lan-
guage syntax, phonology, and so forth. Te second takes up
meaning. Te fourth and ffth chapters address the other two
levels of language society (at the top) and the brain (at the
bottom). Te latter also addresses the topics of genetics and
evolution, integrating these with the treatment of the brain. Te
sixth chapter takes up language acquisition. Tus, it turns from
the evolution of the general language capacities of the human
brain to the development of the particular language compe-
tence of individual human minds. Finally, the seventh chap-
ter considers the nonordinary use of language in verbal art.
Te subsequent entries specify, elaborate, augment, and
revise the ideas of these essays. Here, of course, the editor of
a volume on language sciences faces the problem of just what
entries should be included. In other words, if language sciences
encompass the language-related part of neuroscience, social
science, and so forth, just what is that language-related part?
What does it include, and what does it exclude? One might defne
the impairment of (mental) syntactic capacities by reference
to (bodily) lesions in that area. For our purposes, the crucial
part of this analysis is its implication that language includes
all three levels and that the sciences of language should
therefore encompass brains, associative networks, intentions,
mental representations, individual interactions, and group
dynamics. Tis takes us to our third question .
WHAT IS A SCIENCE OF LANGUAGE?
Te preceding sections converge on a broad, pluralistic but
not indiscriminate account of what constitutes a language
science. Specifcally, a language science is the application
of general principles of scientifc method to language phe-
nomena at any level. At the level of brain, we have neuro-
linguistics (see brain and language). At the level of
associative networks, we have connectionism. Intentionalism
leads us to certain forms of ordinary language philos-
ophy. Representational architectures are particularly well
developed, including Noam Chomskys various theories (see,
for example, minimalism), cognitive linguistics, and
other approaches. Personal interaction and group dynamics
are taken up in pragmatics, linguistic discourse analy-
sis, and sociolinguistics. Just as language encompasses
patterns at all these levels, language science necessarily
includes systematic accounts of language at all these levels.
Again, the levels of language are interrelated without being
reducible. Similarly, the various sciences are interrelated
systematically interrelated through upward explanation
and downward understanding or interpretation without
being reducible.
Te preceding points should serve to clarify something that
is obvious, but rather vague, in ordinary speech: Language
science is not the same as language. Language science is a
systematic treatment of language that seeks to provide both
explanation and understanding. Tus, an encyclopedia of
the language sciences does not present the same material as
an encyclopedia of language. It presents the current state of
theoretical explanation and understanding (along with some
historical background that is important for contextualizing
current theories). It does not present the current state of knowl-
edge about particular features of particular languages except
insofar as these features enter into research programs that aim
toward broader explanatory accounts or principles of more
general understanding. Tus, the phonology of Urdu, the
morphology of Quechua, the metrical principles of English
verse lines, and the story and discourse structure of
Chinese narratives enter into the following essays and entries
only insofar as they enter into larger theoretical concerns .
Of course, to say this is only to mark out a general area for an
encyclopedia of the language sciences. It does not determine
precisely what essays and/or entries should make up such a
work. Tis leads to our fnal concern.
THE STRUCTURE OF THE VOLUME
Te preceding view of language science guided the formulation
of topics for the entries and the organization of the introductory
Preface
xx
be important things to say, not only about the neurobiology
of syntax, but also about the neurobiology of pragmat-
ics and the neuropsychology of verbal art. Te frst has
certainly been more fully researched than the second or third.
But that is only more reason to stress the importance of the sec-
ond and third, to bring together what research has been done,
and to point to areas where this research could be productively
extended.
While it is possible to be systematic with research areas, one
cannot be systematic with theories. Teories are more idiosyn-
cratic. Tey difer from one another along many axes and can-
not be generated as a set of logical possibilities. I have sought to
represent theories that have achieved some level of acceptance
in scientifc communities. Given limitations of space, decisions
on this score have often been dif cult particularly because
social constructionist and related analyses show that accep-
tance in scientifc communities is by no means a straightfor-
ward function of objective scientifc value.
Tis leads us once again to the issue of the validity of theo-
ries. It should come as no surprise that my view of the issue in
efect combines a pluralistic realism with a roughly Lakatosian
advocacy of research programs and a Feyerabend-like practical
anarchism (Feyerabend 1975; Lakatos 1970). Specifcally, I take
it that some theories are true and others are not. However, I do
not believe that only one theory is true. Diferent theories may
organize the world in diferent ways. Tere is no correct way of
organizing the world (though some ways will be more useful
than others for particular purposes). On the other hand, once
the world is organized in a certain way, then certain accounts of
the world are correct and certain accounts are incorrect. To take
a simple example, we may divide the color spectrum in difer-
ent ways (see color classification). No division is correct
or incorrect. But once we have a division, there are facts about
the color of particular objects. (Tis view is related to Donald
Davidsons (1984) argument that truth is not relative to a con-
ceptual scheme, though it is, of course, relative to the mean-
ing of ones words. It also may have some similarity to Hilary
Putnams (1981) internal realism, depending on how that is
interpreted.)
Moreover, even for one organization of the world, we can
never defnitively say that a given theory is or is not true. Note
that this means we cannot even strictly falsify a theory. We can
refer to the ongoing success of a research program and that is
important. Yet I do not share Imre Lakatoss (1970) optimism
about research programs. To some extent, research programs
appear to succeed insofar as they have powerful institutional
support, often for not very good intellectual reasons. Here, then,
I agree with Paul Feyerabend (1975) that orthodoxy in theori-
zation is wrong. It is wrong not only in explicitly or implicitly
identifying institutional support with validity. Tus, it is wrong
not only for social constructionist reasons. It is wrong also for,
so to speak, positivist reasons. It is wrong in diminishing the
likelihood of intellectual progress, the likelihood of increasing
the validity of our theories, which is to say the scope of explana-
tion and understanding produced by these theories.
Whether or not this very brief sketch points toward a
good philosophy of science, it does, I believe, point toward
a good philosophy for an encyclopedia of science perhaps
this part very narrowly as including only phenomena that are
necessarily bound up with oral speech, sign languages,
or writing. More theoretically, one might defne this part as
comprising neurobiological, mental, or social phenomena that
occur only in connection with distinctive properties of speech,
signing, or writing.
Certainly, an encyclopedia treating language will focus
on phenomena that are inseparable from speech, sign lan-
guages, and/or writing and on such distinctive aspects of natu-
ral language as syntax. However, here, too, I believe it would
be a mistake to confne language sciences within a narrowly
defned domain. Terefore, I have adopted a looser criterion.
Te volume centrally addresses distinctive properties of natu-
ral language. However, it takes up a wider range of phenomena
that are closely connected with the architectural or, even more
importantly, the functional features of speech, sign languages,
and writing.
Tere are several cognitive operations for which speech,
signing, and writing appear to have direct functional conse-
quences . One is referential the specifcation, compilation,
and interrelation of intentional objects (see the entries on
reference). Here I have in mind phenomena ranging from
the division of the color spectrum to the elaboration of causal
relations . A second area is mnemonic the facilitation and par-
tial organization of memory (see, for example, encoding) . A
third is inferential the derivation of logical implications . A
fourth is imaginative the expansion and partial structuring
of simulation. One could think of the frst and second func-
tions as bearing on direct, experiential knowledge of present
or past objects and events. Te third and fourth functions bear,
rather, on indirect knowledge of actual or possible objects and
events. Two other functions are connected with action rather
than with knowledge . Te frst is motivational the extension
or elaboration of the possibilities for emotional response (see
emotion and language). A fnal area is interpersonal the
communication of referential intents, memories, inferences,
simulations, and motivations .
In determining what should be included in the volume,
I have taken these functions into account, along with archi-
tectural considerations. Tus I see issues of interpretation
and emplotment (one of the key ways in which we organize
causal relations) as no less important than phonology or syn-
tactic structure. Of course, we have more frmly established
and systematic knowledge in some areas than in others. Tus
some entries will necessarily be more tentative, and make refer-
ence to a broader variety of opinion or a more limited research
base. But that is not a reason to leave such entries aside. Again,
the purpose of an encyclopedia of language science is not to
present a compilation of well-established particular facts, but
rather to present our current state of knowledge with respect to
explanation and understanding.
In keeping with this, when generating the entries (e.g.,
Phonology, Syntax, Neurobiology of Phonology,
Neurobiology of Syntax, Acquisition of Phonology, and so
on), I have tried to be as systematic as possible. Tus the vol-
ume includes some topics that have been under-researched
and under-theorized. For example, if neurobiology does in fact
provide a causal substrate for higher levels, then there should
Preface
xxi
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bakhtin , M. M . 1981. Te Dialogic Imagination: Four Essays . Ed.
Michael Holquist . Trans. Caryl Emerson and Michael Holquist.
Austin : University of Texas Press .
Cervantes Saavedra , Miguel de . 1950. Te Adventures of Don Quixote .
Trans. J. M. Cohen . New York : Penguin.
Davidson , Donald . 1984. On the very idea of a conceptual scheme.
In Inquiries into Truth and Interpretation , 18398. Oxford : Oxford
University Press .
Faust , David . 1984. Te Limits of Scientifc Reasoning . Minneapolis :
University of Minnesota Press .
Feyerabend , Paul . 1975. Against Method: Outline of an Anarchistic
Teory of Knowledge . London : Verso.
Hogan , Patrick Colm . 2003. Cognitive Science, Literature, and the Arts:
A Guide for Humanists . New York : Routledge .
Lakatos , Imre . 1970. Falsifcation and the methodology of scientifc
research programmes. In Criticism and the Growth of Knowledge ,
ed. Imre Lakatos and Alan Musgrave , 91195. Cambridge : Cambridge
University Press .
Mahoney , Michael . 1977. Publication prejudices: An experimental
study of confrmatory bias in the peer review system . Cognitive
Terapy and Research 1 : 161 75.
Peters , Douglas , and Stephen Ceci . 1982. Peer-review practices of psy-
chological journals: Te fate of published articles, submitted again .
Behavioral and Brain Sciences 5.2 : 187 95.
Putnam , Hilary . 1981. Reason, Truth, and History . Cambridge :
Cambridge University Press.
particularly language science. I have tried to follow this phi-
losophy throughout the volume. Specifcally, I have sought to
present a range of theoretical ideas (as well as more theory-
independent topics), placing them together in such a way as
to encourage a mutual sharpening of ideas and insights. To
borrow M. M. Bakhtins terms (1981), I have not set out to pro-
vide a monological source of authoritative discourse. Rather, I
have sought to present a heteroglot volume with which read-
ers may interact dialogically (see dialogism and hetero-
glossia) hopefully, to produce more intellectually adequate
theories later. Toward this end, I have encouraged authors to
be open about their own judgments and attitudes. Tere is a
common view that a piece of writing is biased if the speaker
frankly advocates one point of view. But, in fact, the opposite is
the case. A piece of writing is biased if a speaker acts as though
he or she is simply reporting undisputed facts, when in fact he
or she is articulating a partisan argument. Being open, dialogi-
cal, and multivocal does not mean being bland. Indeed, insight
is more likely to be produced through the tension among ideas
and hypotheses that are clearly delineated in their diferences.
Tis is no less true in the language sciences than elsewhere.
Indeed, that is one reason why this volume treats language sci-
ences, not the science of language.
Patrick Colm Hogan
xxiii
First of all, I must thank Phil Laughlin, who (inspired by the exemplary MIT Encyclopedia of
the Cognitive Sciences ) suggested the project initially and invited me to make a more formal
proposal. Without Phils initial idea and subsequent encouragement, this volume would not
now exist. Eric Schwartz took over from Phil, then Simina Calin took over from Eric; both were
supportive and helpful, as were the several editorial assistants, most recently April Potenciano,
Christie Polchowski, and Jeanie Lee. Regina Paleski and Mark Fox ably shepherded this com-
plex project through the production process; Phyllis Berk worked with devotion on copy edit-
ing the manuscript; and Robert Swanson took on the tough job of indexing.
Te members of the editorial board kindly provided comments on the list of entries and
suggested possible authors. Tey also served as second readers for most of the entries. I am
indebted to them all. It is dif cult and unrewarding work, but extremely valuable. Some entries
were evaluated by specialists not on the editorial board. I am deeply grateful to the following
scholars who agreed to read and comment on entries: J. Abutalebi, E. Ahlsn, A. Aikhenvald,
S. Anderson, A. Atkin, S. Barker, J. Beall, D. Beaver, H. Bejoint, H. Ben-Yami, A. Berger,
D. Bickerton, A. Bilgrami, S. Blackmore, J. Blommaert, C. Bowern, E. Brenowitz, J. Bybee,
J. Carroll, T. Deacon, M. DeGraf, J.-L. Dessalles, A. Edgar, C. Elgin, R. Ferrer i Cancho, J. Field,
H. Filip, D. Finkelstein, J. Forrester, R. Gibbs, R. Gibson, R. Giora, R. Gleave, K. Gluer-Pagin,
M. Goral, M. Hashimoto, J. Heath, D. Herman, R. Hilpinen, J. Hintikka, K. Hofman, K. Holyoak,
P. Hutchinson, J. Hyun, P. Indefrey, M. Israel, K. Johnson, M. Johnson, J. Kane, P. Kay, A. Kibort,
S. Kiran, C. Kitzinger, W. Labov, B. Laford, C. Larkosh, A. Libert, P. Livingston, K. Ludwig,
M. Lynch, J. Magnuson, G. Marcus, R. May, J. McGilvray, A. Mehler, S. Mills, D. Moyal-Sharrock,
K. Oatley, B. OConnor, L. Pandit, B. Partee, J. Pennebaker, P. Portner, C. Potts, J. Robinson,
S. Rosen, S. Ross, J. Saul, R. Schleifer, M. Shibatani, R. Skousen, S. Small, W. Snyder, M. Solms,
F. Staal, P. Stockwell, L. Talmy, H. Truckenbrodt, J. P. Van Bendeghem, W. van Peer, S. Wheeler,
and L. Zhang. Tanks to M. Cutter for help with the illustrations.
For some time, a graduate assistant, Karen Renner, took care of many secretarial duties.
Tis work was supported by the English Department at the University of Connecticut, with
some added funding from the University of Connecticut Research Foundation. Te support of
the English Department was due to the kindness and commitment of our department head,
Bob Tilton without his help, this project would not have been possible.
Work for the entry on Ad Hoc Categories was supported by National Science Foundation
Grant BCS-0212134 and DARPA Contract FA865005-C-7256 to Lawrence W. Barsalou.
Te entry on Dyslexia was prepared with support from a British Academy Research
Readership to Margaret J. Snowling.
Preparation of the manuscript for Speech Production was supported by NIDCD A-93 and
NIDCD grant DC-03782, both to Haskins Laboratories.
Research for Paisley Livingstons entries benefted from fnancial support from the Research
and Postgraduate Studies Committee of Lingnan University, Hong Kong.
ACKNOWLEDGMENTS
1
the sum of individual actions performed by separate persons;
in particular, each individual involved must take into consid-
eration the other individuals beliefs, intentions, and actions in
a way that can be described as cooperative. A shared coopera-
tive activity between two individuals can be defned in terms
of a set of attitudes held by the cooperating individuals and as
a way of carrying out the individual action (Bratman 1992 ). Te
attitudes are as follows:
(a) Each individual participant intends to perform the
joint action. Tat is, each participants intention is not
directed simply toward his/her individual action but
toward the joint action that is carried out by both partici-
pants together.
(b) Each participant intends to perform the joint action
in accordance with and because of each ones meshing
subplans. Tat is, each participants individual actions are
intended to mesh with the other participants actions in
order to successfully achieve the joint action.
(c) Neither participant is coercing the other.
(d) Each participant has a commitment to mutual sup-
port. Tat is, each one will help the other to carry out the
subplans; each participant is thus responsible for more than
just execution of his/her own subplan.
(e) All of (a)(d) are common ground, or shared knowl-
edge between the individuals. Te concept of common
ground plays a central role in understanding the function
of language in social interaction; it is discussed more fully
toward the end of this essay.
Finally, in addition to these mental attitudes on the part
of the participants, there must be mutual responsiveness in
action. Tat is, the participants will coordinate their individual
actions as they are executed in order to ensure that they mesh
with each other and, hence, that the joint action will be suc-
cessfully carried out (to the best of their abilities). Coordination
is essential in carrying out joint actions successfully, and this is
where language plays a central role in joint actions.
Te social cognitive abilities necessary for shared cooper-
ative activity appear to be unique to humans, providing what
Michael Tomasello (2008) calls the social cognitive infrastruc-
ture necessary for the evolution of the capacity for modern
human language. Other species than humans have a capacity
for imitative learning of complex vocalizations (see animal
communication and human language ). Tis has not
been suf cient to lead to the evolution of human-like language
among these species. Nonhuman primates have the ability to
plan actions and to recognize regularities in behavior of other
creatures, enough to manipulate their behavior. Tese abilities
are preconditions for executing complex actions such as joint
actions, but they are not suf cient for doing so .
Research on primate behavior in natural and experimen-
tal settings suggest that some primates even have the ability
to recognize conspecifcs as beings with intentional states like
themselves in some circumstances (Tomasello, 2008; this abil-
ity develops in humans only at around nine months of age).
Nevertheless, it has not been demonstrated that nonhuman pri-
mates have the ability to engage in shared cooperative activity
1
LANGUAGE STRUCTURE IN ITS HUMAN
CONTEXT: NEW DIRECTIONS FOR THE LANGUAGE
SCIENCES IN THE TWENTY-FIRST CENTURY
William Croft
Te science of language in the twenty-frst century is likely to
expand its scope compared to that of the twentieth century.
Te twentieth-century science of language focused its atten-
tion on the analysis of language structure: the sound system
of a language ( phonology ) and its grammatical system
( mo rphology and syntax ). Te analysis of linguistic struc-
ture, or form, is central to the science of language. After the
middle of the twentieth century, however, greater attention
was placed on the relationship between language form and its
psychological and social context.
Te analysis of linguistic structure will remain central to
the science of language. However, understanding language
in context will undoubtedly be a crucial feature of language
science in the twenty-frst century. Tis essay focuses on the
basic principles that have emerged in research on language in
its social and cognitive context, the ways that this context con-
strains language structure and use, and the new directions in
research implied by the integration of language structure and
context. Tis essay is necessarily selective in the topics cov-
ered, and the selection represents a particular way to integrate
language form and its context. It also brings together theories
that have originated in philosophy, psychology, and sociology,
as well as diferent branches of linguistics. Such efort is neces-
sary in order to treat language as a unitary phenomenon, and
also to relate central questions of linguistic analysis to other
scientifc domains. Language structure cannot be fully under-
stood without situating it with respect to current theories of
joint action, social cognition, conceptualization of experience,
memory and learning, cultural transmission and evolution,
shared knowledge and practice in communities, and demo-
graphic processes in human history.
WHY TALK? THE PRAGMATICS OF LANGUAGE
Why do we talk? Why does language exist? It is only by answer-
ing these questions that we can understand how language fts
in its context. Te answer is that language plays an essential
role in social interaction, fundamentally at the level of joint
action between two or more individuals (Clark 1996 ; Tomasello,
2008). What makes a joint action joint is that it is more than just
The Cambridge Encyclopedia of the Language Sciences
2
butterfy ) or a grammatical construction (such as the Modifer-
Head construction for English noun phrases) emerges as a con-
vention when it becomes a regularly used means for solving the
recurrent coordination problem of referring to a specifc expe-
rience that is to be communicated .
Linguistic convention actually operates at two levels: the
grammatical level of words and constructions, at which the
speakers intentions are formulated; and the phonological level
of the articulation and perception of the sounds that make up
the grammatical units. Tis is the phenomenon described as
duality of patterning in language (Hockett 1960 ). One could
imagine in principle that linguistic convention possessed only
one level: perceivable sounds (or gestures or written images,
depending on the medium), corresponding to part (i) in the
defnition of convention, that directly conveyed the speakers
intention (the recurrent coordination problem) as a whole, cor-
responding to part (v) in the defnition of convention. Tese
exist in interjections with specifc functions such as Hello and
Tanks . However, most linguistic expressions are complex,
consisting of discrete, meaningful units. Complex linguistic
expressions evolved for two reasons: First, the number of difer-
ent speaker intentions to be communicated grew to be indef-
nitely large; and second, a speakers intended message came to
be broken down into recurrent conceptual parts that could be
recombined to produce the indefnite variety of messages .
Again, one could imagine that each conventional linguis-
tic unit consisted of a unique sound (gesture, image). But lan-
guages have distinct meaningful units that are made up of
diferent sequences of the same sounds: bat, sat, Sam, same,
tame, time, etc. Tis system has evolved for the same two rea-
sons: the increasing number of meaningful units (even the
recurring ones) necessary to convey the indefnitely large num-
ber of speaker intentions, and an ability to break down a sound
signal (or gesture, or image) into parts that can be recombined
as a sequence of sounds (or gestures or images). Tus, the dual-
ity of patterning characteristic of human language has evolved
to accommodate the huge number of speaker intentions that
people want to convey, and to exploit the facts that intentions
can be broken down into recombinable conceptual units and
that the medium of expression can be broken down into recom-
binable units as well .
Language is therefore a joint action that operates simulta-
neously at four levels (Clark 1996 ). Te higher-numbered levels
are dependent on the lower-numbered levels; the individual
actions of the interlocutors are given in italics:
(4) proposing and taking up a joint project (joint action);
(3) signaling and recognizing the communicative intention;
(2) formulating and identifying the proposition;
(1) producing and attending to the utterance.
Te highest level corresponds to the illocutionary
force in speech-act theory (Austin 1962 ); the next level to
Gricean meaning, or the informative act (Clark 1992 ); the next
level to the propositional act (Searle 1969 ); and the lowest level to
the utterance act (Austin 1962 ; Searle 1969 ). Each level enables
the level(s) above it, and succeeds only if the level(s) below has
been successfully achieved (e.g., one cannot recognize the
as already defned. Tomasello (ibid.) suggests that in particu-
lar, helpfulness, Michael Bratmans condition (d), may be criti-
cal to the evolution of the ability to carry out joint actions .
Te fnal condition for joint action is that the individual
actions must be coordinated in accordance with the shared
attitudes of the participants. Any joint action poses coordi-
nation problems between the participants (Lewis 1969 ). Any
means that is used to solve the coordination problem on a
particular occasion is a coordination device. Tere are vari-
ous coordination devices that human beings use to solve the
coordination problems of joint actions, of which the simplest is
joint attention to jointly salient properties of the environment
(Tomasello 1999 , 2003 ). But by far the most efective coordina-
tion device is for the participants to communicate with each
other: By communicating their mental states, the participants
greatly facilitate the execution of any joint action.
communication is itself a joint action, however. Te
speaker and hearer must converge on a recognition of the
speakers intention by the hearer (see communicative
intention ; see also cooperative principle ). Tis is H.
Paul Grices defnition of meaning ([ 1948 ] 1989 ), or Herbert
Clarks informative act (Clark 1992 ; see the next section). And
this joint action poses coordination problems of its own. Te
essential problem for the joint action of communication is that
the participants cannot read each others minds. Language is
the primary coordination device used to solve the coordina-
tion problem of communication, which is in turn used to solve
the coordination problem for joint actions in general. Indeed,
that is the ultimate purpose of language: to solve the coordina-
tion problem for joint actions, ranging from the mundane to the
monumental (Clark 1999 ). Tis fact is essential for understand-
ing the structure of discourse and the linguistic expressions
used in it, as Clark ( 1992 , 1996 ) has shown for many aspects of
conversational interaction, and it also accounts for many fun-
damental properties of linguistic structure .
LANGUAGE, COMMUNICATION, AND CONVENTION
A language can be provisionally described as a conventional
system for communication (this defnition is modifed later in
this section). David Lewis ( 1969 ) and Clark ( 1996 , Chapter 5)
defne convention as follows:
(i) A regularity in behavior
(ii) that is partly arbitrary (that is, we could have equally
chosen an alternative regularity of behavior),
(iii) that is common ground in the community,
(iv) as a coordination device
(v) for a recurrent coordination problem.
In other words, conventions can emerge when members of
the community have shared knowledge that a certain repeated
behavior can act among them as a coordination device for a
recurrent coordination problem. Tis defnition of convention
is general: It applies to conventions such as shaking hands (or
kissing on the cheek) for greeting, or driving on the right (left)
side of the road. Te defnition also applies straightforwardly to
language: A string of sounds (i.e., a word or morpheme, such as
Language Structure in Its Human Context
3
diferent recurrent coordination problems. For example, patient
is ambiguous between the linguistic semantic role ( Te patient
in sentence 25 is Roland ) and a role in the domain of medicine
( Te patient in room 25 is Roland ). Linguistic convention alone
cannot tell which meaning is intended by the speaker. Only
joint salience, provided in the example sentences by the mean-
ings of the other words and the broader context of conversa-
tion, will successfully solve the coordination problem of what
is meant by patient .
Indexicality and ambiguity are so pervasive in language that
no utterance can be successfully conveyed without recourse to
nonconventional coordination devices. But convention itself is
also incomplete. Tis is because every situation being commu-
nicated is unique and can be construed as the recurrence of
diferent coordination problems. Te simplest example of this
phenomenon is that diferent words can be used to describe
the current situation, each representing a diferent construal
of the current situation in comparison to prior situations. For
example, one can refer to an individual as the prime minister,
Tony Blair, the Labour Party leader, my friend, that guy, he , etc.;
each expression construes reference to the current person as
the recurrence of a diferent coordination problem .
Te need to use nonconventional coordination devices as
well as linguistic convention in communication is not generally
a problem for successful joint actions by cooperative human
beings. However, in some contexts, successful coordination
is quite dif cult. For example, scholarly discourse on abstract
theoretical concepts often leads to alternative construals of
what is intended by particular scholars. What do we take Plato
to have meant? Tis changes over time and across persons.
Alternative construals, not always accurately described as
misunderstandings, occur in more everyday circumstances
as well, as readers can verify for themselves.
In addition, human beings are not always cooperative. Te
complexity of language as joint action here leaves open many
possible means of language abuse. For example, lying abuses
linguistic convention in its role of helping coordinate a shared
cooperative activity, namely, coming to a shared belief. Other
types of language abuse exploit nonconventional coordination
devices. For example, in one lawsuit, the courts ordered a gov-
ernment agency to destroy certain documents , intending the
term to denote their information content; the agency destroyed
the documents , that is, the physical objects, after making cop-
ies of them (Bolinger 1980 ). Here, the ambiguity of documents
requires recourse to joint salience, but the agency abused this
nonconventional coordination device (the lawsuit was about
privacy of information). Finally, the fact that a current situation
can be construed as an instance of diferent recurrent coordi-
nation problems leads to alternative framings of the situation,
such as referring to an entity as a fetus or an unborn baby . Tese
alternative framings bias the conceptualization of the current
situation in ways that invite certain inferences and courses of
action, rather than others .
THE LINGUISTIC SYSTEM IN CONTEXT
In the preceding sections, language is described as a conven-
tional system for communication, and the role of convention
communicative intention if one did not pay attention to the
utterance produced).
THE INCOMPLETENESS OF CONVENTION
Te model of language as joint action describes the social cog-
nitive system that must have evolved in the human species for
modern human language to have emerged. It describes what
appears to be a stable system that led to the emergence of highly
complex cooperative activity among humans, namely, what is
called society or culture. But it is not a complete picture of the
nature of language in social interaction.
Linguistic convention can function as a coordination device
for communication because there are recurrent coordination
problems in communication: People have repeatedly wished
to convey similar intentions formulated in similar concepts.
Convention, linguistic or otherwise, is a regularity of behav-
ior that emerges in a community or society. But convention
must emerge from previous successful communication events
where a convention did not previously exist. In other words,
there must be a precedent: You and I use a coordination device
because we used it before (or observed it used before), and it
worked. Following a precedent is a coordination device, but it
is not (yet) convention; it is based not on regular behavior that
is mutually known in the community but only on previous suc-
cessful uses that we are aware of (Lewis 1969 ).
Following a precedent cannot be the ultimate root of con-
vention either. It always requires a successfully coordinated
communicative act as a precedent. Te ultimate coordination
device is joint salience: Each participant can assume that in a
particular situation, certain features are salient to both partici-
pants (Lewis 1969 ). Joint salience is possible because humans
have the social cognitive capacity for joint attention to their
environment (Tomasello 2003 ). Joint attention forms a basis for
common ground, as discussed later in this article .
Linguistic convention, however, is not perfect; it does not
trump or replace the nonconventional coordination devices of
precedent and joint salience in the act of communication. Tis
is partly because of the kind of conventions found in language,
and partly because of the nature of convention itself.
Linguistic conventions are incomplete because of the phe-
nomena of indexicality and ambiguity (Clark 1996 ). A lin-
guistic convention such as hat or fnd represents a type, but on
a particular occasion of use, we often intend to convey a partic-
ular token of the category. Tus, I found the hat communicates
a particular taking event involving a specifc hat. In order to
identify which fnding event and which hat, the interlocutors
must rely on joint salience in the context, facilitated in part by
the past tense of fnd and the article the combined with hat , to
coordinate successfully on the right fnding event and the right
hat. Linguistic shifters, such as the pronoun I , more explicitly
require joint salience, namely, who is the speaker in the con-
text. Proper names denote tokens, but even a proper name such
as William Croft may be (and is) used for more than one indi-
vidual, for example, the contemporary linguist and the English
Baroque musical composer .
Most words are also highly ambiguous; that is, the same
regularity of behavior is used as a coordination device to solve
The Cambridge Encyclopedia of the Language Sciences
4
by phonological categories, and syntagmatic contrasts by the
phonological structure of words and larger prosodic units.
Te syntagmaticparadigmatic distinction is the most basic
way to describe the fact that the linguistic system allows a (re-)
combination of meaningful units in diferent ways. Te adap-
tive motivation for the emergence of such a communication
system was described previously: Te number of intentions to
be communicated is so great that a set of simple (atomic) sym-
bols will not suf ce, but experience is such that it can be broken
down into recurrent parts for which conventional linguistic
expressions can develop. Te same motivations gave rise to the
syntagmaticparadigmatic distinction in phonology as well .
Paradigmatic principles of structure in grammar and pho-
nology are represented in terms of linguistic categories, phono-
logical and grammatical. Tese abstract linguistic categories
can be mapped onto the substantive categories of the actual
phonetic realization (for phonology) and of utterance meaning
(for grammar). Linguistic typology (Comrie 1989 ; Croft 2003 ),
which takes a cross-linguistic perspective on grammatical
analysis, has demonstrated that the ways in which phonologi-
cal categories are mapped onto phonetic space, and grammati-
cal or lexical categories are mapped onto conceptual space, are
not unlimited. For example, phonetic similarities and concep-
tual similarities constrain the structure of phonological and
grammatical categories, respectively .
Syntagmatic principles of structure are represented in var-
ious ways, but all such representations refect another basic
principle, the hierarchical organization of the structure of
utterances. Sentences are organized in a hierarchical struc-
ture, representing groupings of words at diferent levels. So
Te cat sat on the mat is not just a string of roles that contrast
syntagmatically, as in [Determiner Noun Copula Preposition
Determiner Noun]. Instead, it is a set of nested groupings of
words: [[Determiner Noun] [Copula] [Preposition [Determiner
Noun]]]. Te nested groupings are frequently represented in a
variety of ways, such as the syntactic trees of phrase (constitu-
ent) structure analysis. Tey can also be represented as depen-
dency diagrams (for example, the determiner is related to the
noun as its modifer, which in turn is related to the copula as
its subject), and representations combining constituency and
dependency also exist .
Te structure of a construction often appears to be motivated,
though not entirely predicted, by the structure of the meaning
that it is intended to convey. For example, the syntactic group-
ings in [[ Te cat ] is [ on [ the mat ]]] are motivated semantically;
the in the cat modifes cat semantically as well as syntactically
(indicating that the cats identity is known to both speaker and
hearer). Te (partial) motivation of syntactic structure by its
meaning is captured by general principles in diferent theories.
Tese principles can be described as variants of the broader
principle of diagrammatic iconicity (Peirce 1932 ): roughly, that
the abstract structure of the linguistic expression parallels the
abstract structure of the meaning intended, to a great extent.
It is dif cult to evaluate the structure of meaning indepen-
dently of the structure of linguistic form . However, diferent
speech communities settle on a similar range of constructions
to express the same complex meaning the regularities dis-
covered in linguistic typology (see, for example, the studies
in language and of language in communication was discussed.
In this section, the linguistic system is described in broad out-
line. Linguistic structure has been intensively studied over the
past century ever since Ferdinand de Saussure inaugurated
the modern analysis of linguistic structure, Structuralism
(Saussure [ 1916 ] 1966 ). Tis section focuses on those aspects of
linguistic structure that are generally agreed upon and shows
the extent to which they emerge from the principles that have
been presented in the preceding section.
Te most fundamental structuralist principle is the central-
ity of the linguistic sign or symbol, that is, the notion that lan-
guage pairs form and meaning, and that particular linguistic
forms convey particular meanings. Tis principle fts directly
with the defnition of convention. Te regularity in behavior in
part (i) of the defnition of convention is the expression of a lin-
guistic form by a speaker; the recurrent coordination problem
in part (v) of the defnition is the communication of a meaning
between the interlocutors .
Also central to the structural analysis of language is the
arbitrariness of the linguistic sign. Tat is, arbitrariness exists
in the particular form and meaning that are paired. Tis con-
forms with part (ii) of the defnition of convention, namely,
that the convention is partly arbitrary. Arbitrariness is usu-
ally defned in structuralist analysis as the principle that one
cannot entirely predict the form used from the meaning that is
intended. From a communicative point of view, arbitrariness
means that another choice could have served approximately
equally well. For example, the choice of producing the string
of sounds butterfy for a particular meaning could have been
replaced with the choice of producing the string of sounds
Schmetterling a choice made by members of the German
speech community. Two diferent choices are communica-
tively equivalent in that neither choice is preferred for the
meaning intended and that is usually because the choice of
one expression over the other is arbitrary in the structuralist
sense .
Another principle that can be traced back to Saussure is
the distinction between the paradigmatic and syntagmatic
contrast of linguistic units. In a complex (multiword or multi-
morpheme) grammatical construction, such as Te cat sat on
the mat, each word enters into two diferent types of contrast.
For example, the frst word the contrasts with the word cat in
that the s role in the construction (determiner) contrasts with
cat s role (head noun). Tis is a syntagmatic contrast. But the
also contrasts with another possible fller of the same role in
the construction, such as a in A cat sat on the mat ; and cat con-
trasts with hamster, parakeet , etc. in the same way. Tese are
paradigmatic contrasts.
More recent grammatical theories represent paradigmatic
contrast in terms of a set of elements belonging to a grammati-
cal category. Tus, the and a belong to the category determiner ,
and cat, hamster, parakeet, etc. belong to the category noun .
Syntagmatic contrasts are represented by contrasting roles in
the syntactic structure or constructions used in the utterance.
For example, the determiner category is functioning as a modi-
fer of the noun category in a noun phrase construction. Finally,
the syntagmaticparadigmatic distinction also applies to pho-
nology (sound structure): Paradigmatic contrast is represented
Language Structure in Its Human Context
5
content words , such as nouns and verbs. In this way, the speaker
has transformed the unique whole of the original experience
into parts that can be expressed by language .
Tis is not the end of the verbalization process. Content
words denote only general categories of parts of the experi-
ence to be verbalized. In order to communicate the original
experience, the speaker must tie down the categories to the
unique instances of objects, events, and so forth in the expe-
rience, and the speaker must assemble the parts into a struc-
ture representing the original whole that the speaker intends
to verbalize. Tat is to say, corresponding to the categorizing
step in verbalizing the parts of the experience, there is a par-
ticularizing step that indicates the unique parts; and corre-
sponding to the steps of propositionalizing and subchunking
are integrative steps of structuring and cohering, respectively
(Croft 2007). Tese latter three steps give rise to grammar in the
sense of grammatical constructions, infections, and particles,
and the semantic commonalities among grammatical catego-
ries across languages.
Te particularizing step takes a category (a type) and selects
an instance (token) or set of tokens, and also identifes it by sit-
uating it in space and time. For object concepts, selecting can
be accomplished via the infectional category of number, and
via the grammatical categories of number and quantifcation
( three books, an ounce of gold ). For action concepts, selecting
is done via grammatical aspect, which helps to individuate
events in time ( ate vs. was eating ), and via agreement with sub-
ject and/or object, since events are also individuated by the
participants in them ( I read the paper and She read the maga-
zine describe diferent reading events). Objects and events can
be situated in space via deictic expressions and other sorts of
locative expressions ( this book, the book on the table ). Events
and some types of objects can be situated in time via tense and
temporal expressions ( I ate two hours ago; ex-mayor ). Events
and objects can also be situated relative to the mental states
of the interlocutors: Te article in the book indicates that the
particular object is known to both speaker and hearer, and the
modal auxiliary in She should come indicates that the event
exists not in the real world but in the attitude of obligation in
the mind of the speaker .
Te structuring step takes participants and the predicated
event in a clause and puts them together, reassembling the
predicate and the argument(s) into the subchunk from which
they were derived by propositionalizing. Grammatically this is
a complex area. It includes the expression of grammatical rela-
tions in what is called the argument structure of a predicate, so
that She put the clock on the mantle indicates which referent is
the agent (the subject), which the thing moved (the object), and
which the destination of the motion (the prepositional phrase).
But it also includes alternative formulations of the same event,
such as Te clock was put on the mantle (the passive voice con-
struction) and It was the mantle where she put the clock (a cleft
construction). Te alternative constructions function to pres-
ent the information in the proposition in diferent ways to the
hearer, depending on the way the discourse is unfolding; they
are referred to as information structure or discourse function.
Finally, the cohering step takes the clauses (subchunks) and
reassembles them into a whole that evokes the original whole
published in Typological Studies in Language and the Oxford
Studies in Typology and Linguistic Teory). Tis fact suggests
that there are regularities in the meaning to be conveyed that
are then refected in the grammatical constructions used to
express them .
GRAMMAR AND THE VERBALIZATION OF EXPERIENCE
Te preceding sections have described the general context of
language use and the basic principles of language structure.
Te grammars of particular languages conform to the basic
principles of language structure in the preceding section. But
the grammars of particular languages, while diverse in many
ways, are similar to a much greater degree than would be pre-
dicted from the general principles in the preceding section, or
even the context of language use described in the earlier sec-
tions. For example, all languages have structures like clauses
in which some concept (prototypically an action concept,
usually labeled a verb) is predicated on one or more concepts
that are referred to (prototypically an object or person, usually
labeled a noun). Te noun-like expressions are in turn orga-
nized into phrases with modifers. Clauses are related to each
other by varying degrees of grammatical integration. Certain
semantic categories are repeatedly expressed across languages
as grammatical infections or function words (e.g., articles,
prepositions, auxiliaries) that combine with the major lexical
categories of words in sentences.
Tese universal patterns in grammar are attributable to the
way that experience is verbalized by human beings. Te fun-
damental problem of verbalization is that each experience that
a speaker wishes to verbalize is a unique whole. But a linguis-
tic utterance is unlike an experience: An utterance is broken
down into parts, and these parts are not unique; they have
been used before in other utterances. (Tis latter point is the
fact of convention; a particular linguistic form is used regularly
and repeatedly for a recurrent coordination problem.)
Te process by which the unique whole of experience is
turned into a linguistic utterance made up of reusable parts has
been described by Wallace Chafe ( 1977 ). Te frst step is that
the speaker subchunks the experience into smaller parts, each
also a unique Gestalt similar in this way to the original experi-
ence. Te subchunking process may be iterated (in later work,
Chafe emphasizes how consciousness shifts from one chunk
to another in the experience to be verbalized). A subchunk of
the experience is then propositionalized; this is the second
step. Propositionalizing involves breaking up an experience by
extracting certain entities that are (at least prototypically) per-
sistent, existing across subchunks. Tese entities are the refer-
ents that function as arguments of the predicate; the predicate
is what is left of the subchunk after the arguments have been
separated. Propositionalizing therefore breaks down the expe-
rience into parts arguments and the predicate that are not of
the same type as the original experience (i.e., not a Gestalt) .
Once the whole has been broken down into these parts,
the parts must be categorized, that is, assigned a category that
relates the parts of the current experience to similar parts of
prior experiences. Categorizing is the third step in the verbali-
zation process. Tese categories are what are expressed by
The Cambridge Encyclopedia of the Language Sciences
6
frst and then recent proposals to apply it to grammar (syntax
and semantics ) are examined.
One of the major results of instrumental phonetics is the
discovery that phonetic variation in speech is ubiquitous.
Variation in the realization of phonemes is found not just
across speakers but also in the speech of a single speaker. Tere
are at least two reasons why such variation in the speech signal
would exist. Presumably, the level of neuromuscular control
over articulatory gestures needed for identical (invariant) pro-
ductions of a phoneme is beyond a speakers ability. At least as
important, the variation in the speech signal does not prevent
successful communication (or not enough of the time to lead
to the evolution of even fner neuromuscular control abilities
in humans) .
Tere is evidence, moreover, that the mental representa-
tion of phonological categories includes the representation of
individual tokens of sounds and the words that contain them.
Speakers retain knowledge of fne-grained phonetic detail
(Bybee 2001 ; Pierrehumbert 2003 ). Also, there are many fre-
quency efects on phonological patterns (Bybee 2001 ). For
example, higher-frequency forms tend to have more reduced
phonetic realizations of phonemes than lower-frequency
forms.
Finally, human beings are extremely good pattern detectors
from infancy on into adulthood. Infants are able to develop
sensitivity to subtle statistical patterns of the phonetic signals
they are exposed to. Tis type of learning, which occurs with-
out actively attending to the stimulus or an intention to learn
is called implicit learning (Vihman and Gathercole, unpub-
lished manuscript). It contrasts with explicit learning , which
takes place under attention from the learner particularly
joint attention between an infant learning language and an
adult and is involved in the formation of categories and sym-
bolic processing. Tere is neuroscientifc evidence that implicit
learning is associated with the neocortex and explicit learning
with the hippocampus (ibid.) .
A number of researchers have proposed a usage-based or
exemplar model of phonological representation to account
for these patterns (Bybee 2001 ; Pierrehumbert 2003 ). In this
model, phonological categories are not represented by specifc
phonetic values for the phoneme in the language, but by a clus-
ter of remembered tokens that form a density distribution over
a space of phonetic parameters. Te phonetic space represents
the phonetic similarities of tokens of the phonological category.
Tis model includes properties of implicit learning (the clus-
ter of individual tokens) and explicit learning (the labeling of
the density distribution as representing tokens of, say, /e/ and
not /i/). Consolidation of token memories also takes place
individual tokens decay in memory, highly similar tokens are
merged, and the distribution of tokens can be restructured
but new tokens are constantly being incorporated into the rep-
resentation and infuencing it.
Marilyn Vihman and S. Kunnari ( 2006 ) propose three types
of learning for an exemplar model. First, there is an initial
implicit learning of statistical regularities of the sensory input.
Second, explicit learning of linguistic categories, such as the
words that are templates containing the sound segments, takes
place. Finally, a second layer of implicit learning of statistical
experience for the hearer. Tis step can be accomplished by
various clause-linking devices, including subordination of var-
ious kinds, coordination, and other clause-linking construc-
tions found in the worlds languages. Coherence of clauses
in discourse is also brought about by discourse particles and
reference tracking , that is, grammatical devices, such as
pronouns or ellipsis , which show that an event is related to
another event via a shared participant ( Harry flled out the form
and _ mailed it to the customs of ce ).
Te three steps of particularizing, structuring, and cohering
result in a grammatical structure that evokes a reconstituted
version of the original unique whole. Tese six steps in verbal-
ization are not necessarily processed sequentially or indepen-
dently. Te steps in the verbalization process are dependent on
the grammatical resources available in the language, which
constrain the possibilities available to the speaker. For exam-
ple, when a speaker takes a subchunk and extracts participants
from it, there must be a construction available in the language
to relate the participants to the predicate, as with put in the
earlier example. Tus, subchunking must be coordinated with
propositionalizing and structuring. Also, the steps may not be
overtly expressed by grammatical infections or particles. For
example, Te book fell does not overtly express the singular
number of book , or that the event is situated in the real world
rather than a nonreal mental space of the speaker .
Finally, the reconstituted experience evoked by the linguis-
tic utterance is not the same as the unique whole with which
the speaker began. Te cognitive processes humans use in ver-
balization do not simply carry out one or more of the six steps
described. Tey also conceptualize the experience in diferent
ways, depending on the speakers choices. Tese choices range
from the subtle diference between describing something as
leaves or foliage , or the more dramatic framing diferences
between fetus and unborn baby referred to previously . Tere
are a wide range of conceptualization processes or construal
operations that have been identifed in language (see, e.g.,
Langacker 1987 ; Talmy 2000 ). Te construal operations can be
accounted for by processes familiar from cognitive psychol-
ogy: attention, comparison, perspective, and Gestalt (Croft
and Cruse 2004 , Chapter 4). Tese psychological processes are
part of the meaning of all linguistic units: words, infections,
and constructions. As a consequence, every utterance presents
a complex conceptualization of the original experience that
the speaker intends to verbalize for the hearer. Te conven-
tionalized conceptualizations embodied in the grammatical
resources of a language represent cultural traditions of ways to
verbalize experience in the speech community .
VARIATION AND THE USAGE - BASED MODEL
One of the results of recent research on language structure
and language use is the focus on the ubiquity of variation in
language use, that is, in the verbalization of experience and its
phonetic realization. Te ubiquity of variation in language use
has led to new models of the representation of linguistic knowl-
edge in the mind that incorporate variation as an essential
characteristic of language. Tese models are more developed in
phonetics and phonology. Te phonological model is described
Language Structure in Its Human Context
7
linguistic comparison of the meanings expressed by grammat-
ical categories and constructions. Te typological approach to
grammar has constructed conceptual spaces for a number of
semantic domains using techniques such as the semantic map
model (see Haspelmath 2003 for a survey of recent studies) and
multidimensional scaling (Croft and Poole 2008 ).
To sum up, the usage-based/exemplar model can be applied
to both phonological patterns in words and grammatical struc-
tures in constructions. A speakers knowledge of language
is the result of the interplay between two learning processes.
One learning process is the tallying of statistical regularities of
tokens of words and constructions with a particular phonetic
realization, performing a particular communicative act in a
specifc social interaction. Te other is the organization of these
tokens into categories and the formation of generalizations that
allow the reuse or replication of these grammatical structures
to solve future coordination problems in communication .
VARIATION AND CHANGE: AN EVOLUTIONARY APPROACH
Te view of language described in the preceding sections roots
both language structure and a speakers linguistic knowledge
in the individual acts of linguistic behavior that a speaker has
engaged in and will engage in. It is a dynamic view of language
in that linguistic behavior is essentially historical: a tempo-
ral series of utterances, each one linked to prior utterances as
repeated behavior to solve recurrent coordination problems in
social interaction. Each member of a speech community has a
history of his or her participation in linguistic events, either as
speaker or hearer. Tis history is remembered in the exemplar-
based representation of that members linguistic knowledge,
but also consolidated and organized in such a way that each
unique experience is broken down and categorized in ways
that allow for reuse of words and constructions in future com-
munication events.
Each time a speaker produces an utterance, he or she rep-
licates tokens of linguistic structures sounds, words, and
constructions based on the remembering of prior tokens of
linguistic structures, following the principles of convention
and verbalization described earlier. However, the replication
process is never perfect: Variation is generated all of the time,
as described in the preceding section. Te variation generated
in the process of language use can be called frst-order varia-
tion. Variation in replication is the starting point for language
change. Language change is an instance of change by replica-
tion (rather than inherent change); change by replication is the
domain of an evolutionary model of change (Hull 1988 ; Croft
2000 ).
Change by replication is a two-step process. Te frst step is
the generation of variation in replication. Tis requires a repli-
cator and a process of replication by which copies are produced
that preserve much of the structure of the original. In bio-
logical evolution, the canonical replicator is the gene, and the
process of replication takes place in meiosis (which in sexual
organisms occurs in sexual reproduction). Copies of the gene
are produced, preserving much of the structure of the original
gene. Variation is generated by random mutation processes
and by recombination in sexual reproduction .
regularities gives rise to probability distributions for each lin-
guistic phonological and lexical category. Te result of this
last layer of learning is the exemplar or usage-based model
described by Janet Pierrehumbert and Joan Bybee .
Te application of the usage-based/exemplar model to
grammar is more complex. Most research in this area has
compared the range of uses of a particular word or grammati-
cal construction. However, this does not represent the process
of language production (that is, verbalization), analogous to
the phonetic variation found in the production of phonemes.
Studies of parallel verbalizations of particular scenes demon-
strate that variation in the verbalization of the same scene by
speakers in similar circumstances is ubiquitous, much like the
phonetic realization of phonological categories (Croft 2010).
Tere is also substantial evidence for frequency efects in
grammar. For example, English has a grammatical category of
auxiliary verb that has distinctive syntax in negation ( I ca nt
sing vs. I didnt sing ), questions ( Can he sing? vs. Did he sing ? ).
Tese syntactic patterns are actually a relic of an earlier stage
of English when word order was freer; it has survived in the
auxiliaries of modern English because of their higher token
frequency (Bybee and Tompson 1997 ), as well as their seman-
tic coherence. Frequency plays a central role in the historical
process of Grammaticalization (Hopper and Traugott 2003 ), in
which certain constructions develop a grammatical function
(more precisely, they are recruited to serve the particularizing,
structuring, and cohering steps of the verbalization process).
Part of the grammaticalization process is that the construction
increases in frequency; it therefore undergoes grammatical
and phonological changes, such as fxation of word order, loss
of syntactic fexibility, and phonetic reduction (Bybee 2003 ). A
well-known example is the recruitment of the go + Infnitive
construction for the future tense: She is going (to Sears) to buy
a food processor becomes future Shes going to buy a food pro-
cessor , with no possibility of inserting a phrase between go and
the infnitive, and is fnally reduced to Shes gonna buy a food
processor .
Finally, early syntactic acquisition is driven by implicit
learning of patterns in the linguistic input (Tomasello 2003 ).
Te process of syntactic acquisition is very gradual and induc-
tive, involving an interplay between detection of statistical reg-
ularities and the formation of categories that permit productive
extension of grammatical constructions. Children occasionally
produce overregularization errors, and these are also sensitive
to frequency (more frequent forms are more likely to be pro-
duced correctly, and less frequent forms are more likely to be
subject to regularization) .
A usage-based model of grammatical form and meaning
is gradually emerging from this research. An exemplar model
of grammatical knowledge would treat linguistic meanings as
possessing a frequency distribution of tokens of remembered
constructions used for that meaning. Tose constructions
would be organized in a multidimensional syntactic space
organized by structural similarity (e.g., Croft 2001 , Chapter 8)
and whose dimensions are organized by the function played
by the construction in the verbalization process. Te mean-
ings of constructions are themselves organized in a conceptual
space whose structure can be inferred empirically via cross-
The Cambridge Encyclopedia of the Language Sciences
8
conventions of the speech community. Each speakers system-
atic knowledge of his or her language is diferent, because of
diferences in the range of language use to which each speaker
is exposed .
SPEECH COMMUNITIES AND COMMON GROUND
Language in this revised sense is the product of a speech com-
munity: the utterances produced by communicative interac-
tions among speakers. A speech community is defned by its
social interactions involving language: Members of the speech
community communicate with one another, and the commu-
nity is defned by communicative isolation from other com-
munities. Communicative isolation is relative, of course, and
in fact the structure of human speech communities is far more
complex than the structure of biological populations.
Two related phenomena serve to defne communities: com-
mon ground and shared practice. Common ground plays an
essential role in defning joint action and convention, both cen-
tral to understanding the nature of language. Common ground
consists of knowledge, beliefs, and attitudes presumed by two or
more individuals to be shared between them. Common ground
can be divided into two types: personal common ground and
communal common ground (Clark 1996 , Chapter 4). Personal
common ground is shared directly in face-to-face interaction
by the persons. Personal common ground has two bases. Te
frst is the perceptual basis: We share knowledge of what is in
our shared perceptual feld. Te perceptual basis is provided
by virtue of joint attention and salience, as mentioned earlier.
A shared basis for common ground has the following prop-
erties: Te shared basis provides information to the persons
involved that the shared basis holds; the shared basis indicates
to each person that it provides information to every person that
the shared basis holds; and the shared basis indicates the prop-
osition in the common ground (Clark 1996 , 94). A basis for com-
mon ground varies in how well it is justifed; hence, we may not
always be certain of what is common ground or not.
Te second basis for personal common ground is a dis-
course basis. When I report on situations I have experienced
to you in conversation, and vice versa, these become part of
our personal common ground. Although we did not expe-
rience them perceptually together, we did experience the
reporting of them linguistically together. Te discourse basis
thus involves joint attention (to the linguistic signal), as well as
the common ground of a shared language. Te discourse basis
and the perceptual basis both require direct interaction by the
interlocutors. Tey correspond to social networks, which are
instrumental in language maintenance and change (Milroy
1987 ).
Te other type of common ground is communal common
ground. Communal common ground is shared by virtue of
common community membership. A person can establish
common ground with a stranger if they both belong to a com-
mon community (e.g., Americans, linguists, etc.). Some com-
munities are quite specialized while other communities are
very broad and even all-encompassing, such as the community
of human beings in this world, which gives rise to the possibil-
ity of communication in the frst place.
In language, replication occurs in language use. Te repli-
cators are tokens of linguistic structures in utterances (called
linguemes in Croft 2000 ). Tese tokens are instances of lin-
guistic behavior. Te process of language change is therefore
an example of cultural transmission, governed by principles
of evolutionary change. Te replication process in language
change is governed by the principle of convention. As we have
seen in the preceding section, variation is generated in the pro-
cess of verbalization, including the recombination of linguistic
forms. Tis represents innovation in language change. First-
order variation is the source of language change. Experiments
in phonological perception and production indicate that sound
change is drawn from a pool of synchronic variation (the title
of Ohala 1989 ). Indeterminacy in the interpretation of a com-
plex acoustic signal can lead to reanalysis of the phonological
categories in that signal. Likewise, it appears that grammati-
cal change is also drawn from a pool of synchronic variation,
namely, variation in verbalization. Tere is an indeterminacy
in the understanding of the meaning of a word or construction
because we cannot read each others minds, our knowledge of
linguistic conventions difers because we have been exposed
to diferent exemplars, and every situation is unique and can
be construed in diferent ways. Tis indeterminacy gives rise
to variation in verbalization (Croft 2010), and can lead to the
reanalysis of the mapping of function into grammatical form
(Croft 2000 ) .
Te second step of the evolutionary process is the selection
of variants. Selection requires an entity other than the repli-
cator, namely, the interactor. Te interactor interacts with its
environment in such a way that this interaction causes repli-
cation to be diferential (Hull 1988 ). In biological evolution, the
canonical interactor is the organism. Te organism interacts
with its environment. In natural selection, some organisms
survive to reproduce and therefore replicate their genes while
others do not; this process causes diferential replication.
In language, selection occurs in language use as well. Te
interactor is the speaker. Te speaker has variant linguistic
forms available and chooses one over others based on his or
her environment. In language, the most important environ-
mental interaction is the social relationship between speaker
and hearer and the social context of the speech event. Tis is, of
course, the realm of sociolinguistics (see, e.g., Labov 2001 , and
the following section). Selection goes under the name of propa-
gation in language change.
Selection (propagation) is a function of the social value that
variants acquire in language use. First-order variation does
not have a social value. Socially conditioned variation is sec-
ond-order variation. Once a variant is propagated in a speech
community, it can lead to third-order variation, that is, varia-
tion in linguistic conventions across dialects and languages.
Linguistic diversity is the result of language change .
Te evolutionary model requires a revision to the defni-
tion of language ofered near the beginning of this essay. In the
evolutionary model, a language is a population of utterances,
the result of the employment of linguistic conventions in a
speech community. Te linguistic system is the result of the
ways in which speakers have consolidated the uses of language
in which they have participated into their knowledge of the
Language Structure in Its Human Context
9
leading to distinct dialects and eventually to mutually unintel-
ligible languages.
Tis ubiquitous demographic process is refected in the fam-
ily trees of languages that have been constructed by linguists
working on genetic classifcation. Tese family trees allow for
the possibility of reconstructing not just protolanguages but
also the underlying social processes that are traced in them.
Even sociolinguistic situations that obscure family trees leave
linguistic evidence of other social processes. Extensive bor-
rowing indicates a period of intensive social contact. Dif culty
in separating branches of a linguistic family tree indicates an
expansion through a new area but continued low-level con-
tact between the former dialects. Tese can be seen in the dia-
lect continua found in much of Europe, where the Romance,
Germanic, and Slavic peoples expanded over a mostly continu-
ous terrain (Chambers and Trudgill 1998 ). Shared typological
(structural) traits may be due to intimate contact between lan-
guages with continued language maintenance, or to a major
language shift by a social group, resulting in a large proportion
of non-native speakers at one point in a languages history.
Te spread of human beings across the globe led to the cre-
ation of a huge number of distinct societies whose languages
diverged. Te number of distinct languages that have survived
until the beginning of the twenty-frst century is about 6,000.
Most linguists generally accept the hypothesis that modern
human language evolved just once in human history, proba-
bly no later than 70,000 to 100,000 years ago. So in principle,
all modern human languages may have originated in a single
common ancestor. Tracing back the actual lineages of con-
temporary languages deep into human prehistory appears to
be extremely dif cult, if not impossible. Nevertheless, there is
no doubt that contemporary linguistic diversity is extremely
ancient in human history. What we can discover about linguis-
tic history by the comparison of existing languages can poten-
tially shed important light on human history and prehistory .
Tere are linguistic descriptions of a small propor-
tion of existing human languages, though descriptive work
has increased and the overall quality of descriptions has
improved dramatically, thanks to advances in linguistic sci-
ence throughout the twentieth century. It would be safe to say
that the diversity of linguistic structure, and how that struc-
ture is manifested in phonetic reality on the one hand and in
the expression of meaning on the other, is truly remarkable
and often unexpected. Many proposed universals of language
have had to be revised or even abandoned as a consequence,
although systematic analysis of existing linguistic descriptions
by typologists have revealed many other language universals
that appear to be valid. Linguistic diversity has revealed alter-
native ways of conceptualizing experience in other societies, as
well as alternative methods of learning and alternative means
for communication for the accomplishment of joint actions .
But the single most important fact about the diversity of
human language is that it is severely endangered. Of the 6,000
diferent languages extant today, 5,000 are spoken by fewer
than 100,000 people. Te median number of speakers for a
language is only 6,000 (Crystal 2000 ). Many languages are no
longer spoken by children in the community, and therefore
will go extinct in another generation. Te loss for the science
Clark argues that the basis of communal common ground
is shared expertise. tienne Wenger, on the other hand, defnes
communities of practice in terms of shared practice: Individuals
engage in joint actions together, and this gives them common
ground and creates a community (Wenger 1998 ). Wengers
defnition of a community of practice, therefore, requires face-
to-face interaction, like personal common ground. However,
shared practice can be passed on as new members enter the
community and share practice with remaining current mem-
bers. Tis is cultural transmission and can lead to individuals
being members of the same community through a history of
shared practice, even if they do not interact directly with every
other member of the community .
Since communities are defned by shared practice, and
human beings engage in a great variety of joint actions with
diferent groups of people, the community structure of human
society is very complex. Every society is made up of multiple
communities. Each person in the society is a member of mul-
tiple communities, depending on the range of shared activities
he or she engages in. Te diferent communities have only par-
tially overlapping memberships.
As a consequence, the structure of a language is equally
complex. A linguistic structure a pronunciation, a word, a
construction is associated with a particular community, or
set of communities, in a society. A pronunciation is recog-
nized as an accent characteristic of a particular community.
Words will have diferent meanings in diferent communi-
ties (e.g., subject is a grammatical relation for linguists but
a person in an experiment for psychologists). Te same con-
cept will have diferent forms in diferent communities (e.g.,
Zinfandel for the general layperson, Zin to a wine afcionado).
Tus, a linguistic convention is not just a symbol a pairing
of form and meaning but includes a third part, the commu-
nity in which it is conventional. Tis is part (iii) of the defni-
tion of convention given in an earlier section. Finally, each
individual has a linguistic repertoire that refects his or her
knowledge and exposure to the communities in which he or
she acts .
Te choice of a linguistic form on the part of a speaker is
an act of identifcation with the community that uses it. Tis
is the chief mechanism for selection (propagation) in language
change: Ihe propagation of variants refects the dynamics of
social change. More recent work in sociolinguistics has argued
that linguistic acts of social identity are not always pas-
sive: Individuals institute linguistic conventions to construct
an identity as well as to adopt one (Eckert 2000 ) .
LANGUAGE DIVERSITY AND ITS ENDANGERMENT
Variation in language can lead to language change if it is prop-
agated through a speech community. Social processes over
human history have led to the enormous linguistic diversity
we fnd today a diversity that newer social processes also
threaten. Te basic social process giving rise to linguistic
diversity is the expansion and separation of populations into
distinct societies. As groups of people divide for whatever rea-
son, they become communicatively isolated, and the common
language that they once spoke changes in diferent directions,
The Cambridge Encyclopedia of the Language Sciences
10
Clark , Herbert H. 1992 . Arenas of Language Use . Chicago and
Stanford : University of Chicago Press and the Center for the Study of
Language and Information.
. 1996 . Using Language . Cambridge : Cambridge University Press .
Clark , Herbert H. . 1999 . On the origins of conversation. Verbum
21 : 14761.
Comrie , Bernard. 1989 . Language Universals and Linguistic Typology .
2d ed. Chicago : University of Chicago Press .
Croft , William. 2000 . Explaining language change: An evolutionary
approach . Harlow, Essex : Longman .
. 2001 . Radical Construction Grammar: Syntactic Teory in
Typological Perspective . Oxford : Oxford University Press .
. 2003 . Typology and Universals . 2d ed. Cambridge : Cambridge
University Press .
. 2007 . Te origins of grammar in the verbalization of experi-
ence. Cognitive Linguistics 18 : 33982.
. 2010. Te origins of grammaticalization in the verbalization of
experience. Linguistics 48: 148.
Croft , William , and D. Alan Cruse. 2004 . Cognitive Linguistics .
Cambridge : Cambridge University Press .
Croft , William , and Keith T. Poole . 2008 . Inferring universals from
grammatical variation: Multidimensional scaling for typological
analysis. Teoretical Linguistics 34 : 137 .
Crystal , David. 2000 . Language Death . Cambridge : Cambridge
University Press .
Eckert , Penelope . 2000 . Linguistic Variation as Social Practice: Te
Linguistic Construction of Identity in Belten High . Oxford : Blackwell .
Grice , H. Paul. [1948] 1989 . Meaning. In Studies in the Way of Words ,
21323. Cambridge : Harvard University Press .
Haspelmath , Martin. 2003 . Te geometry of grammatical mean-
ing: Semantic maps and cross-linguistic comparison. In Te New
Psychology of Language . Vol. 2. Ed. Michael Tomasello , 21142.
Mahwah, NJ : Lawrence Erlbaum Associates .
Hockett , Charles F. 1960 . Te origin of speech. Scientifc American
203 : 8896.
Hopper , Paul , and Elizabeth Traugott . 2003 . Grammaticalization. 2d
ed. Cambridge : Cambridge University Press .
Hull , David L. 1988 . Science as a Process: An Evolutionary Account of the
Social and Conceptual Development of Science . Chicago : University
of Chicago Press .
Labov , William . 2001 . Principles of Linguistic Change. Vol. 2. Social
Factors . Oxford : Blackwell .
Langacker , Ronald W. 1987 . Foundations of Cognitive Grammar. Vol. 1.
Teoretical Prerequisites . Stanford : Stanford University Press .
Lewis , David . 1969 . Convention . Cambridge, MA : MIT Press .
Milroy , Lesley. 1987 . Language and Social Networks. 2d ed. Oxford : Basil
Blackwell .
Ohala , John. 1989 . Sound change is drawn from a pool of syn-
chronic variation. In Language Change: Contributions to the Study
of its Causes , ed. Leiv Egil Breivik and Ernst Hkon Jahr , 17398.
Berlin : Mouton de Gruyter .
Peirce , Charles Sanders . 1932 . Ground, object and interpretant. In
Collected Papers of Charles Sanders Peirce . Vol. 2: Elements of Logic,
ed. Charles Hartshorne and Paul Weiss , 13455. Cambridge : Harvard
University Press .
Pierrehumbert , Janet B. 2003 . Probabilistic phonology: discrimination
and robustness. In Probabilistic Linguistics , ed. Rens Bod , Jennifer
Hay , and Stefanie Jannedy , 177228. Cambridge, MA : MIT Press .
Saussure , Ferdinand de. [1916] 1966 . Cours de linguistique gnrale . Ed.
Ch. Bally and A. Sechehaye . ( Course in General Linguistics . Trans.
Wade Baskin. New York : McGraw-Hill .)
Searle , John R. 1969 . Speech Acts: An Essay in the Philosophy of Language .
Cambridge : Cambridge University Press .
of language, and more generally for our understanding of
human history, human thought, and human social behav-
ior, is immense. But the loss is at least as great for the speak-
ers themselves. Language use is a mode of social identity, not
just in terms of identifying with a speech community but as
the vehicle of cultural transmission. Te loss of languages, like
other linguistic phenomena, is a refection of social processes.
Te most common social processes leading to language loss are
disruption, dislocation, or destruction of the society (language
loss rarely occurs via genocide of its speakers). Te enormous
consequences of language loss has led to a shift in linguistic
feldwork from mere language description and documentation
to language revitalization in collaboration with members of
the speech community. But reversing language shift ultimately
requires a change in the social status of the speech community
in the local and global socioeconomic system .
SUMMARY
Te scientifc study of language in its pragmatic, cognitive,
and social context beginning in the latter half of the twenti-
eth century is converging on a new perspective on language
in the twenty-frst century. Linguistic conventions coordinate
communication, which in turn coordinates joint actions. Te
fragility of social interaction by individuals leads to creativity,
variation, and dynamism in the verbalization and vocalization
of language. Individual linguistic knowledge (the linguistic
system) refects the remembered history of language use and
mediates processes of language change. Te continually chang-
ing structure of society, defned by common ground emerging
from shared practices (joint actions), guides the evolution of
linguistic conventions throughout its history. Human history
in turn has spawned tremendous linguistic diversity, which
refects the diversity of human social and cognitive capacity.
But the unchecked operation of contemporary social forces is
leading to the destruction of speech communities and the mass
extinction of human languages today .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Austin , J. L. 1962 . How to Do Tings with Words . Cambridge : Harvard
University Press .
Bolinger , Dwight. 1980 . Language, the Loaded Weapon . London :
Longmans .
Bratman , Michael . 1992 . Shared cooperative activity. Philosophical
Review 101 : 32741.
Bybee , Joan L. 2001 . Phonology and Language Use . Cambridge
Cambridge University Press .
. 2003 . Mechanisms of change in grammaticalization: Te role
of frequency. In Handbook of Historical Linguistics , ed. Brian Joseph
and Richard Janda , 60223. Oxford : Blackwell .
Bybee , Joan L ., and Sandra A. Tompson . 1997 . Tree frequency efects
in syntax. In Proceedings of the 23rd Annual Meeting of the Berkeley
Linguistics Society , ed. Matthew L. Juge and Jeri O. Moxley , 37888.
Berkeley : Berkeley Linguistics Society .
Chafe , Wallace . 1977 . Te recall and verbalization of past experi-
ence. In Current Issues in Linguistic Teory , ed. Peter Cole , 21546.
Bloomington : Indiana University Press .
Chambers , J. K. , and Peter Trudgill. 1998 . Dialectology. 2d ed.
Cambridge : Cambridge University Press .
Language Structure in Its Human Context
11
Vihman , Marilyn M. , and V. M. Gathercole . Language Development .
Unpublished manuscript.
Vihman , Marilyn M. , and S. Kunnari . 2006 . Te sources of pho-
nological knowledge: A crosslinguistic perspective. Recherches
Linguistiques de Vincennes 35 : 13364.
Wenger , tienne . 1998 . Communities of Practice: Learning, Meaning
and Identity . Cambridge : Cambridge University Press .
Talmy , Leonard. 2000 . Toward a Cognitive Semantics. Vol. 1. Concept
Structuring Systems Cambridge, MA : MIT Press .
Tomasello , Michael. 1999 . Te Cultural Origins of Human Cognition .
Cambridge : Harvard University Press .
. 2003 . Constructing a Language: A Usage-Based Teory of
Language Acquisition . Cambridge : Harvard University Press .
. 2008 . Te Origins of Human Communication . Cambridge,
MA : MIT Press .
12
with a speakers (or hearers) semantic and contextual knowl-
edge. Here, we review some of what we have learned about the
psychology of linguistic form, as it pertains to sounds, words,
and sentences.
SOUNDS
Sound units. Since the advent of speech research, one of the
most intensively pursued topics in speech science has been
the search for the fundamental sound units of language. Many
researchers have found evidence for phonological units that
are abstract (i.e., generalizations across any number of heard
utterances, rather than memories of specifc utterances) and
componential (constituent elements that operate as part of a
combinatorial system). However, there is other evidence for
less abstract phonological forms that may be stored as whole
words. As a result, two competing hypotheses about phonolog-
ical units have emerged: an abstract componential one versus
a holistic one.
Te more widespread view is the componential one. It pos-
its abstract units that typically relate either to abstract versions
of the articulatory gestures used to produce the speech sounds
(Liberman and Mattingly 1985, Browman and Goldstein 1990),
or to ones derived from descriptive units of phonological the-
ory. Such descriptive units include the feature (see feature
analysis ) , an abstract subphonemic unit of contrast; the pho-
neme , an abstract unit of lexical contrast that is either a conso-
nant or a vowel ; the phone or allophone , a surface variant of the
phoneme ; the syllable , a timing unit that is made up of a vowel
and one or more of its fanking consonants ; the prosodic word ,
the rhythmic structure that relates to patterns of emphasized
syllables ; or various structures that related to tone , the lexi-
cally contrastive use of the voices pitch, and intonation , the
pitch-based tune that relates to the meaning of a sentence (for
reviews, see Frazier 1995; Studdert-Kennedy 1980) .
In the holistic view, the word is the basic unit, whereas
other smaller units are considered to be epiphenomenal (e.g.,
Goldinger, Pisoni, and Logan, 1991). Instance-specifc mem-
ory traces of particular spoken words are often referred to as
episodes . Proponents of this view point out that while abstract
units are convenient for description and relate transparently
to segment-based writing systems, such as those based on
the alphabet, there is evidence that listeners draw on a vari-
ety of highly detailed and instance-specifc aspects of a words
pronunciation in making lexical decisions (for reviews, see
Goldinger and Azuma 2003; Nygaard and Pisoni 1995) .
Some researchers have proposed hybrid models in which
there are two layers of representation: the episodic layer, in which
highly detailed memory traces are stored, and an abstract layer
organized into features or phones (Scharenborg et al. 2005). Te
proponents of hybrid models try to capture the instance-spe-
cifc efects in perception that inspire episodic approaches, as
well as the highly abstracted lexical contrast efects .
PROCESSES. speech production refers to the process by
which the sounds of language are produced. Te process nec-
essarily involves both a planning stage, in which the words and
other linguistic units that make up an utterance are assembled
2
THE PSYCHOLOGY OF LINGUISTIC FORM
Lee Osterhout , Richard A. Wright, and Mark D. Allen
Humans can generate and comprehend a stunning variety of
conceptual messages, ranging from sophisticated types of men-
tal representations, such as ideas, intentions, and propositions,
to more primal messages that satisfy demands of the immedi-
ate environment, such as salutations and warnings. In order
for these messages to be transmitted and received, however,
they must be put into a physical form, such as a sound wave
or a visual marking. As noted by the Swiss linguist Ferdinand
de Saussure ([2002] 2006), the relationship between mental
concepts and physical manifestations of language is almost
always arbitrary. Te words cat , sat , and mat are quite similar
in terms of how they sound but are very dissimilar in meaning;
one would expect otherwise if the relationship between sound
and meaning was principled instead of arbitrary. Although the
relationship between linguistic form and meaning is arbitrary,
it is also highly systematic . For example, changing a phoneme
in a word predictably also changes its meaning (as in the cat ,
sat , and mat example) .
Human language is perhaps unique in the complexity of
its linguistic forms (and, by implication, the system underly-
ing these forms). Human language is compositional; that is,
every sentence is made up of smaller linguistic units that have
been combined in highly constrained ways. A standard view
(Chomsky 1965; Pinker 1999) is that units and rules of combina-
tion exist at the levels of sound (phonemes and phono logy ),
words ( morphemes and morphology) , and sentences (words
and phrases and syntax ). Collectively, these rules comprise a
grammar that defnes the permissible linguistic forms in the
language. Tese forms are systematically related to, but dis-
tinct from, linguistic meaning ( semantics ).
Linguistic theories, however, are based on linguistic
description and observation and therefore have an uncertain
relation to the psychological underpinnings of human lan-
guage. Researchers interested in describing the psychologically
relevant aspects of linguistic form require their own meth-
ods and evidence. Furthermore, psychological theories must
describe not only the relevant linguistic forms but also the pro-
cesses that assemble these forms (during language production)
and disassemble them (during language comprehension). Such
theories should also explain how these forms are associated
The Psychology of Linguistic Form
13
its roots in research that dates back over a century. Prior to
the advent of instrumental and experimental methods in the
late nineteenth century, it was commonly accepted that the
basic units of speech were discrete segments that were alpha-
betic in nature and serially ordered. While it was recognized
that speech sounds varied systematically depending on the
phonetic context, the variants themselves were thought to be
static allophones of an abstract and lexically contrastive sound
unit, that is, a phoneme . Translated into modern terminology,
phonological planning involved two stages: 1) determining the
contextually determined set of discrete surface variants, given
a particular lexical string, and 2) concatenating the resulting
allophones. Te physiological implementation of the concat-
enated string was thought to result in a series of articulatory
steady states, or postures . Te only continuous aspects of sound
production were believed to be brief transitional periods cre-
ated by articulatory transitions from one state to the next. Te
transitional movements were thought to be wholly predictable
and determined by the physiology of a particular speakers
vocal tract. Translated again into modern terminology, percep-
tion (when considered) was thought to be simply the process
of translating the allophones back into their underlying pho-
nemes for lexical access. Te earliest example of the phoneme-
allophone relationship is attributed to Pini , around 500 b.c.e.
whose sophisticated system of phonological rules and relation-
ships infuenced structuralist linguists of the early twentieth
century, as well as generative linguists of the late twentieth
century (for a review, see Anderson 1985; Kiparsky 1979).
Te predominant view at the end of the nineteenth century
was typifed by Alexander M. Bells (1867) descriptive work on
English pronunciation. In it, he presented a set of alphabet-in-
spired symbols whose shapes and orientations were intended
to encode both the articulatory steady states and their result-
ing steady-state sounds. A fundamental assumption in the
endeavor was that all sounds of human language could be
encoded as a sequence of universal articulatory posture com-
plexes whose subcomponents were shared by related sounds.
For example, all labial consonants (p, b, m, f, v, w, etc.) shared
a letter shape and orientation, while all voiced sounds (b, d, g,
v, z, etc.) shared an additional mark to distinguish them from
their voiceless counterparts (p, t, k, f, s, etc.). Bells formaliza-
tion of a set of universal and invariant articulatory constituents,
aligned as an alphabetic string, infuenced other universal
transcription systems such as Henry Sweets (1881) Romic
alphabet , which laid the foundation for the development of the
International Phonetic Alphabet (Passy 1888). It also foreshad-
owed the use of articulatory features, such as those proposed
by Noam Chomsky and Morris Halle (1968) in modern phonol-
ogy, in that each speech sound, and therefore each symbol, was
made up of a set of universal articulatory components.
A second way in which Bells work presaged modern research
was the connection between perception and production.
Implicit in his system of writing was the belief that perception
of speech sounds was the process of extracting the articulations
that produced them. Later perceptual models would incorpo-
rate this relationship in one way or another (Chistovich 1960;
Dudley 1940; Fowler 1986, 1996; Joos 1948; Ladefoged and
McKinney 1963; Liberman and Mattingly 1985; Stetson 1951) .
in some fashion, and an implementation stage, in which the
various parts of the vocal tract, for example the articulators ,
execute a motor plan to generate the acoustic signal. See Carol
A. Fowler (1995) for a detailed review of the stages involved in
speech production. It is worth noting here that even if abstract
phonological units such as features are involved in planning
an utterance, at some point the linguistic string must be imple-
mented as a motor plan and a set of highly coordinated move-
ments. Tis has motivated gestural representations that include
movement plans, rather than static featural ones (Browman
and Goldstein 1990; Fowler 1986, 1996; Saltzman and Munhall
1989; Stetson 1951) .
speech perception is the process by which human lis-
teners identify and interpret the sounds of language. It, too,
necessarily involves at least two stages: 1) the conversion of the
acoustic signal into an electrochemical response at the audi-
tory periphery and 2) the extraction of meaning from the neu-
rophysiological response at the cortical levels. Brian C. J. Moore
(1989) presents a thorough review of the physiological processes
and some of the issues involved in speech perception. A funda-
mental point of interest here is perceptual constancy in the face
of a massively variable signal. Restated as a question, how is
it that a human listener is able to perceive speech sounds and
understand the meaning of an utterance, given the massive
variability created by physiological idiosyncrasies and con-
textual variation? Te various answers to this question involve
positing some sort of perceptual units, be they individual seg-
ments, subsegmental features, coordinated speech gestures, or
higher-level units like syllables, morphemes, or words .
It is worth noting here that the transmission of linguistic
information does not necessarily rely exclusively on the audi-
tory channel; the visible articulators, the lips and to a lesser
degree the tongue and jaw, also transmit information. A listener
presented with both auditory and visual stimuli will integrate
the two signals in the perceptual process (e.g., Massaro 1987).
When the information in the visual signal is unambiguous (as
when the lips are the main articulators), the visual signal may
even dominate the acoustic one (e.g., McGurk and Macdonald
1976). Moreover, writing systems convey linguistic informa-
tion, albeit in a low-dimensional fashion. Most strikingly, sign
languages are fully as powerful as speech-based communica-
tion systems and are restricted to the visual domain. Despite
the diferences between signed and spoken languages in terms
of the articulators and their perceptual modalities, they draw
on the same sorts of linguistic constituents, at least so far as
the higher-level units are concerned: syllable, morpheme,
word, sentence, and prosodic phrase (e.g., Brentari 1998). Some
linguists have also proposed the decomposition of signed lan-
guages into smaller units using manual analogs of phonologi-
cal features, despite the obvious diferences in the articulators
and the transmission media (for a review see Emmory 2002).
Te parallel of signed and spoken language structure despite
the diferences in transmission modalities is often interpreted
as evidence for abstract phonological units at the level of the
mental lexicon (Meier, Cormier, and Quinto-Pozos 2002) .
THE HISTORY OF THE DEBATE: EARLY PHONOLOGICAL UNITS. Te
current debate about how to characterize speech sounds has
The Cambridge Encyclopedia of the Language Sciences
14
rediscovered and improved on the earlier observation that the
speech signal and the articulations that create it are continu-
ous, dynamic, and overlapping. Stetson (1951) can be seen as
responsible for introducing kinematics into research on speech
production. His research introduced the notions of coproduc-
tion , in which articulatory gestures were initiated simulta-
neously, and gestural masking , in which the closure of one
articulatory gesture hides another, giving rise to the auditory
percept of deletion. Stetsons work provided the foundation for
current language models that incorporate articulatory ges-
tures and their movements as the fundamental phonological
units (e.g., Browman and Goldstein 1990; Byrd and Saltzman
2003; Saltzman and Munhall 1989) .
In the perceptual and acoustic domains, the identifcation
of perceptual cues to consonants and vowels raised a series
of questions that remain at the heart of the debate to this day.
Te coextensive and covarying movements that produce the
speech signal result in acoustic information that exhibits a
high degree of overlap and covariance with information about
adjacent units (e.g., Delattre, Liberman, and Cooper 1955). Any
single perceptual cue to a particular speech sound can also be a
cue to another speech sound. For example, the onset of a vowel
immediately following a consonant provides the listener with
cues that identify both the consonant and vowel (Liberman et
al. 1954). At the same time, multiple cues may identify a single
speech sound. For example, the duration of a fricative (e.g., s),
the fricatives noise intensity, and the duration of the preceding
vowel all give information about whether the fricative is voiced
(e.g., z) or voiceless (e.g., s) (Soli 1982). Finally, the cues to
one phone may precede or follow cues to adjacent phones. Te
many-to-one, the one-to-many, and the nonlinear relation-
ships between acoustic cues and their speech sounds pose a
serious problem for perceptual models in which features or
phones are thought to bear a linear relationship to each other.
More recently, researchers studying perceptual learning have
discovered that listeners encode speaker-specifc details and
even utterance-specifc details when they are learning new
speech sounds (Goldinger and Azuma 2003). Te latest set of
fndings poses a problem for models in which linguistic sounds
are stored as abstract units .
In distinctive feature theory, each phoneme is made up
of a matrix of binary features that encodes both the distinc-
tions and the similarities between one class of sounds and
the next in a particular language (Jakobson, Fant, and Halle
1952; Chomsky and Halle, 1968). Te features are thought to be
drawn from a language universal set, and thus allow linguists
to observe similarities across languages in the patterning of
sounds. Moreover, segmenting the speech signal into units
that are hierarchically organized permits a duality of pattern-
ing of sound and meaning that is thought to give language its
communicative power. Tat is, smaller units such as phonemes
may be combined according to language-specifc phonotactic
(sound combination) constraints into morphemes and words,
and words may be organized according to grammatical con-
straints into sentences. Tis means that with a small set of
canonical sound units, together with recursion, the talker
may produce and the hearer may decode and parse a virtually
unbounded number of utterances in the language .
THE HISTORY OF THE DEBATE: EARLY EXPERIMENTAL
RESEARCH. Prior to the introduction of experimental meth-
ods into phonetics, the dominant methodologies were intro-
spection about ones own articulations and careful but
subjective observations of others speech, and the measure-
ment units were letter-based symbols. Tus, the observer and
the observed were inextricably linked while the resolution of
the measurement device was coarse. Tis view was challenged
when a handful of phoneticians and psychologists adopted
the scientifc method and took advantage of newly available
instrumentation, such as the kymograph , in the late 1800s.
Tey discovered that there were no segmental boundaries in
the speech stream and that the pronunciation of a particular
sound varied dramatically from one instance to the next (for
a review of early experimental phonetics, see Khnert and
Nolan 1999 and Minife 1999). In the face of the new instru-
mental evidence, some scholars, like Eduard Sievers (1876), P.-J.
Rousselot (1897), and Edward Wheeler Scripture (1902), pro-
posed that the speech stream, and the articulations that pro-
duced it, were continuous, overlapping, and highly variable,
rather than being discrete, invariant, and linear. For them, the
fundamental sound units were the syllable or even the word or
morpheme. Rousselots research (18971901) revealed several
articulatory patterns that were confrmed by later work (e.g.,
Stetson 1951). For example, he observed that when sounds that
are transcribed as sequential are generated by independent
articulators (such as the lips and tongue tip), they are initi-
ated and produced simultaneously. He also observed that one
articulatory gesture may signifcantly precede the syllable it
is contrastive in, thereby presenting an early challenge to the
notion of sequential ordering in speech.
Laboratory researchers like Raymond H. Stetson (1905,
1951) proposed that spoken language was a series of motor
complexes organized around the syllable. Stetson also frst pro-
posed that perception was the process of perceiving the articu-
latory movements that generate the speech signal. However,
outside of the experimental phonetics laboratory, most speech
researchers, particularly such phonologists as Leonard
Bloomfeld (1933), continued to use phonological units that
remained abstract, invariant, sequential, and letter-like. Tree
events that occurred in the late 1940s and early 1950s changed
this view dramatically. Te frst event was the application to
speech research of modern acoustic tools like the spectrogram
(Potter 1945), sophisticated models of vocal tract acoustics
(e.g., House and Fairbanks 1953), reliable articulatory instru-
mentation, such as high-speed X-ray cinefourography (e.g.,
Delattre and Freeman 1968), and electromyographic studies of
muscle activation (Draper, Ladefoged, and Whitteridge 1959).
Te second was the advent of modern perception research in
which researchers discovered complex relationships between
speech perception and the acoustic patterns present in the sig-
nal (Delattre, Liberman, and Cooper 1955). Te third was the
development of distinctive feature theory in which phonemes
were treated as feature matrices that captured the relation-
ships between sounds (Jakobson 1939; Jakobson, Fant, and
Halle 1952) .
When researchers began to apply modern acoustic and
articulatory tools to the study of speech production, they
The Psychology of Linguistic Form
15
Using the lexical entry cat as an example, imagine a connec-
tionist system in which all the semantic features associated
with cat, such as [whiskers], [domestic pet], and so on (which
are also shared with all other conceptual lexical entities bear-
ing those features, such as <lion>, <dog>, etc.), are directly
associated with the phonological units that comprise its word
form /k/, /ae/, /t/ (which are likewise shared with all other
word forms containing these phonemes) by means of individ-
ual association links that directly tie individual semantic fea-
tures with individual phonological units (Rueckl et al. 1997).
One important consequence of this hypothetical arrangement
is that individual word forms do not exist as free-standing rep-
resentations. Instead, the entire lexical entry is represented as
a vector of weighted links connecting individual phonemes to
individual lexical semantic and syntactic features. It logically
follows from this model, then, that if all or most of the semantic
features of the word cat, for example, were destroyed or oth-
erwise made unavailable to the processor, then the set of pho-
nological forms /k/ /ae/ /t/, having nothing to link to, would
have no means for mental representation, and would therefore
not be available to the language processor. We will present
experimental evidence against this model that, instead, favors
models in which a full phonological word (e.g., /kaet/) is rep-
resented in a localist fashion and is accessible to the language
processor, even when access to its semantic features is partially
or entirely disrupted .
Several of the most prominent theories of morphology and
lexical structure within formal linguistics make explicit claims
about modularity of meaning and form. Ray Jackendof (1997),
for example, presents a theory that has a tripartite structure, in
which words have separate identities at three levels of represen-
tation form, syntax, and meaning and that these three levels
are suf cient to encode the full array of linguistic information
encoded by each word. His model provides further details in
which it is proposed that our ability to store, retrieve, and use
words correctly, as well as our ability to correctly compose
morphemes into complex words, derives from a memorized
inventory of mapping functions that picks out the unique rep-
resentations or feature sets for a word at each level and asso-
ciates these elements with one another in a given linguistic
structure .
While most psycholinguistic models of language process-
ing have not typically addressed the mapping operations
assumed by Jackendof, they do overlap signifcantly in terms
of addressing the psychological reality of his hypothetical tri-
partite structure in the mental lexicon. Although most experi-
mental treatments of the multilevel nature of the lexicon have
been developed within models of language production, as will
be seen, there is an equally compelling body of evidence for
multilevel processing from studies of language comprehension
as well.
Te most infuential lexical processing models over the
last two decades make a distinction between at least two lev-
els: the lemma level, where meaning and syntax are stored,
and the lexeme level, where phonological and orthographic
descriptions are represented. Tese terms and the functions
associated with them were introduced in the context of a com-
putational production model by Gerard Kempen and Pieter
WORDS
In this section, we focus on those representations of form that
encode meaning and other abstract linguistic content at the
most minimally analyzable units of analysis namely, words
and morphemes. As such, we give a brief overview of the study
of lexical morphology, investigations in morphological process-
ing, and theories about the structure of the mental lexicon.
LEXICAL FORM. What is the nature of a representation at the
level of lexical form ? We limit our discussion here largely to
phonological codes, but recognize that a great many of the the-
oretical and processing issues we raise apply to orthographic
codes as well. It is virtually impossible for the brain to store
exact representations for all possible physical manifestations of
linguistic tokens that one might encounter or produce. Instead,
representations of lexical form are better thought of as some-
what abstract structured groupings of phonemes (or graph-
emes ) that are stored as designated units in long-term memory,
either as whole words or as individual morpheme constituents,
and associated with any other sources of conceptual or linguis-
tic content encoded in the lexical entries that these form repre-
sentations map onto. As structured sequences of phonological
segments, then, these hypothesized representational units
of lexical form must be able to account for essentially all the
same meaning-to-form mapping problems and demands that
individual phonological segments themselves encounter dur-
ing on-line performance, due to idiosyncratic variation among
speakers and communicative environments. More specifcally,
representations of morphemes and words at the level of form
must be abstract enough to accommodate signifcant variation
in the actual physical energy profles produced by the motor
systems of individual speakers/writers under various environ-
mental conditions. Likewise, in terms of language production,
units of lexical form must be abstract enough to accommo-
date random variation in the transient shape and status of the
mouth of the language producer .
FORM AND MEANING: INDEPENDENT LEVELS OF LEXICAL REPRE-
SENTATION. Te previous description of words and morphemes
to some degree rests on the assumption that lexical form is rep-
resented independently from other forms of cognitive and lin-
guistic information, such as meaning and lexical synta x (e.g.,
lexical category, nominal class, gender, verbal subcategory,
etc.). Many theories of the lexicon have crucially relied on the
assumption of separable levels of representation within the
lexicon. In some sense, as explained by Allport and Funnell
(1981), this assumption follows naturally from the arbitrari-
ness of mapping between meaning and form, and would thus
appear to be a relatively noncontroversial assumption.
Te skeptical scientist, however, is not inclined to simply
accept assumptions of this sort at face value without consid-
ering alternative possibilities. Imagine, for example, that the
various types of lexical information stored in a lexical entry
are represented within a single data structure of highly inter-
connected, independent, distributed features. Tis sort of
arrangement is easy to imagine within the architecture of a
connectionist model (McClelland and Rumelhart 1986).
The Cambridge Encyclopedia of the Language Sciences
16
one must recognize that in many other cases, such as those
involving irregularly infected words, such as taught , the sys-
tem cannot store a stem and af x at the level of form, as there
are no clear morpheme boundaries to distinguish these con-
stituents, but must instead obligatorily store it as a whole word
at the lexeme level.
Many prominent theories have favored the latter, non-
decompositional hypothesis for all words, including irregu-
lar words like taught , as well as regular compositional words
like wanted (Bybee 1988). Other infuential processing models
propose that complex words are represented as whole-word
units at the lexeme level, but that paradigms of infectionally
related words ( want , wants , wanted ) map onto a common rep-
resentation at the lemma level (Fowler, Napps, and Feldman
1985). In addition to this, another class of models, which has
received perhaps the strongest empirical support, posits full
morphological decomposition at the lexeme level whenever
possible (Allen and Badecker 1999). According to these fully
decompositional models, a complex word like wanted is rep-
resented and accessed in terms of its decomposed constitu-
ents want- and -ed at the level of form, such that the very same
stem want- is used during the recognition of want , wants , and
wanted . According to these models, then, the recognition rou-
tines that are exploited by morphological decomposition at the
level of form resemble those in theoretical approaches to sen-
tence processing, in which meaning is derived compositionally
by accessing independent units of representation of form and
combining the content that these forms access into larger lin-
guistic units, according to algorithms of composition specifed
by the grammar.
While there is compelling empirical support for decompo-
sitional models of morphological processing, researchers are
becoming increasingly aware of important factors that might
limit decomposition. Tese factors are regularity, formal and
semantic transparency, and productivity .
Regularity refers to the reliability of a particular word-for-
mation process. For example, the plural noun kids expresses
noun plurality in a regular, reliable way, while the plural noun
children does not .
Formal transparency refers to the degree to which the mor-
pheme constituents of a complex structure are obvious from
its surface form. For example, morpheme boundaries are fairly
obvious in the transparently infected word wanted , com-
pared to those of the opaquely (and irregularly) infected word
taught .
Although an irregular form like taught is formally opaque,
as defned here, it is nonetheless semantically transparent ,
because its meaning is a straightforward combination of the
semantics of the verb teach and the feature [+Past]. In contrast,
an example of a complex word that is formally transparent yet
semantically opaque is the compound word dumbbell , which
is composed of two recognizable morphemes, but the content
associated with these two surface morphemes do not combine
semantically to form the meaning of the whole word .
Productivity describes the extent to which a word-formation
process can be used to form new words freely. For example, the
suf x -ness is easily used to derive novel nouns from adjectives
(e.g., nerdiness , awesomeness , catchiness ), while the ability to
Huijbers (1983) and receive further refnement with respect to
human psycholinguistic performance in the foundational lexi-
cal production models of Merrill F. Garrett (1975) and Willem
Levelt (1989). Much compelling evidence for a basic lemma/
lexeme distinction has come from analyses of naturally occur-
ring speech errors generated by neurologically unimpaired
subjects, including tip-of-the-tongue phenomena (Meyer and
Bock 1992), as well as from systematic analyses of performance
errors observed in patients with acquired brain lesions . A more
common experimental approach, however, is the pictureword
interference naming paradigm, in which it has been shown
that lemma- and lexeme-level information can be selectively
disrupted during the course of speech production (Schriefers,
Meyer, and Levelt 1990).
In terms of lexical comprehension models, perhaps the
most straightforward sources of evidence for a meaning/form
distinction have come from analyses of the performance of
brain-damaged patients. A particularly compelling case for
the independence of meaning and form might be demon-
strated if an individual with acquired language pathology
were to show an intact ability to access word forms in his/
her lexicon, yet remain unable to access meaning from those
form representations. Tis is precisely the pattern observed in
patients designated as sufering from word meaning deafness .
Tese patients show a highly selective pattern of marked def-
cit in comprehending word meanings, but with perfect or near-
perfect access to word forms. A good example is patient WBN
as described in Mark D. Allen (2005), who showed an entirely
intact ability to access spoken word-form representations. In
an auditory lexical decision task, WBN scored 175/182 (96%)
correct, which shows that he could correctly distinguish real
words from nonwords (e.g., fag vs. fig), presumably relying
on preserved knowledge of stored lexemes to do so. However,
on tasks that required WBN to access meaning from spoken
words, such as picture-to-word matching tasks, he performed
with only 40%60% accuracy (at chance in many cases) .
LEXICAL STRUCTURE: COMPLEX WORDS. A particularly impor-
tant issue in lexical representation and processing concerns
the cognitive structure of complex words , that is, words com-
posed of more than one morpheme. One of the biggest debates
surrounding this issue stems from the fact that in virtually all
languages with complex word structures, lexical information
is encoded in consistent, rule-like structures, as well as in idio-
syncratic, irregular structures. Tis issue can be put more con-
cretely in terms of the role of morphological decomposition in
single-word comprehension theories within psycholinguistics.
Consider the written word wanted , for example. A question for
lexical recognition theories is whether the semantic/syntactic
properties of this word [WANT, Verb, +Past, ] are extracted
and computed in a combinatorial fashion each time wanted
is encountered by accessing the content associated with
the stem want- [WANT, Verb] and combining it with the con-
tent extracted from the af x -ed [+Past] or whether instead,
a single whole-word form wanted is stored at the lexeme level
and associated directly with all of its semantic/syntactic con-
tent. To understand the plausibility that a lexical system could
in principle store whole-word representations such as wanted ,
The Psychology of Linguistic Form
17
claimed to be derived directly, without an intervening level of
syntactic structure.
Te initial evidence of separable syntactic and semantic pro-
cessing streams came from studies of brain-damaged patients
sufering from aphasia , in particular the syndromes known as
Brocas and Wernickes aphasia. Brocas aphasics typically pro-
duce slow, labored speech; their speech is generally coherent
in meaning but very disordered in terms of sentence structure.
Many syntactically important words are omitted (e.g., the , is ),
as are the infectional morphemes involved in morphosyntax
(e.g., -ing , -ed , -s ). Wernickes aphasics, by contrast, typically
produce fuent, grammatical sentences that tend to be inco-
herent. Initially, these disorders were assumed to refect def-
cits in sensorimotor function; Brocas aphasia was claimed to
result from a motoric defcit, whereas Wernickes aphasia was
claimed to refect a sensory defcit. Te standard assumptions
about aphasia changed in the 1970s, when theorists began to
stress the ungrammatical aspects of Brocas aphasics speech;
the term agrammatism became synonymous with Brocas
aphasia. Particularly important in motivating this shift was evi-
dence that some Brocas aphasics have a language-comprehen-
sion problem that mirrors their speech-production problems.
Specifcally, some Brocas aphasics have trouble understanding
syntactically complex sentences (e.g., John was fnally kissed by
Louise ) in which the intended meaning is crucially dependent
on syntactic cues in this case, the grammatical words was
and by (Caramazza and Zurif 1976). Tis evidence seemed to
rule out a purely motor explanation for the disorder; instead,
Brocas aphasia was viewed as fundamentally a problem con-
structing syntactic representations, both for production and
comprehension. By contrast, Wernickes aphasia was assumed
to refect a problem in accessing the meanings of words .
Tese claims about the nature of the aphasic disorders are
still quite infuential. Closer consideration, however, raises
many questions. Pure functional defcits afecting a single lin-
guistically defned function are rare; most patients have a mix-
ture of problems, some of which seem linguistic but others of
which seem to involve motor or sensory processing (Alexander
2006). Many of the Brocas patients who produce agrammatic
speech are relatively good at making explicit grammaticality
judgments (Linebarger, Schwartz, and Safran 1983), suggest-
ing that their knowledge of syntax is largely intact. Similarly, it
is not uncommon for Brocas aphasics to speak agrammatically
but to have relatively normal comprehension, bringing into
question the claim that Brocas aphasia refects damage to an
abstract syntax area used in production and comprehension
(Miceli et al. 1983) . Taken together, then, the available evidence
from the aphasia literature does not provide compelling evi-
dence for distinct syntactic and semantic processing streams.
Another source of evidence comes from neuroimaging
studies of neurologically normal subjects. One useful method
involves the recording of event-related brain potentials (ERPs)
from a persons scalp as he or she reads or listens to sentences.
ERPs refect the summed, simultaneously occurring postsyn-
aptic activity in groups of cortical pyramidal neurons. A par-
ticularly fruitful approach has involved the presentation of
sentences containing linguistic anomalies. If syntactic and
semantic aspects of sentence comprehension are segregated
form novel nouns using the analogous suf x -ity is awkward at
best (?nerdity) if not impossible .
Another phenomenon associated with these lexical proper-
ties is that they tend to cluster together in classes of morpholog-
ically complex word types across a given language, such that
there will often exist a set of highly familiar, frequently used
forms that are irregular, formally opaque, and nonproductive,
and also a large body of forms that are morphologically regular,
formally transparent, and productive. Given the large variety of
complex word types found in human languages with respect to
these dimensions of combinability, as well as the idiosyncratic
nature of the tendency for these dimensions to cluster together
from language to language, it would appear that empirical evi-
dence for morphological decomposition must be established
on a case-by-case basis for each word-formation type within
each language. Tis indeed appears to be the direction that
most researchers have taken .
SENTENCES
On the surface, a sentence is a linear sequence of words. But in
order to extract the intended meaning, the listener or reader
must combine the words in just the right way. Tat much is
obvious. What is not obvious is how we do that in real time,
as we read or listen to a sentence. Of particular relevance to
this essay are the following questions: Is there a representa-
tional level of syntactic form that is distinct from the meaning
of a sentence? And if so, exactly how do we extract the implicit
structure in a spoken or written sentence as we process it? One
can ask similar questions about the process of sentence pro-
duction: When planning a sentence, is there a planning stage
that encodes a specifcally syntactic form? And if so, how do
these representations relate to the sound and meaning of the
intended utterance?
For purely practical reasons, there is far more research on
extracting the syntactic form during sentence comprehen-
sion (a process known as parsing ; see parsing, human ) than
on planning the syntactic form of to-be-spoken sentences.
Nonetheless, research in both areas has led to substantive
advances in our understanding of the psychology of sentence
form.
SYNTAX AND SEMANTICS. A fundamental claim of a genera-
tive grammar is that syntax and semantics are clearly dis-
tinct. A fundamental claim of a cognitive grammar is that
syntax and semantics are so entwined that they cannot be easily
separated. Tis debate among linguists is mirrored by a similar
debate among researchers who study language processing. A
standard assumption underlying much psycholinguistic work
is that a relatively direct mapping exists between the levels of
knowledge posited within generative linguistic theories and
the cognitive and neural processes underlying comprehension
(Bock and Kroch 1989). Distinct language-specifc processes
are thought to interpret a sentence at each level of analysis,
and distinct representations are thought to result from these
computations. But other theorists, most notably those working
in the connectionist framework, deny that this mapping exists
(Elman et al. 1996). Instead, the meaning of the sentence is
The Cambridge Encyclopedia of the Language Sciences
18
(mostly involving the measurement of subjects eye movements
as they read sentences) indicated that readers tended to read
straight through such syntactically simple sentences as Te cat
scratched the ratty old sofa but experienced longer eye fxations
and more eye regressions when they encountered by the rac-
coon in the more complex sentences. When confronted with
syntactic uncertainty, readers seemed to immediately choose
the simplest syntactic representation available (Frazier 1987).
When this analysis turned out to be an erroneous choice (that
is, when the disambiguating material in the sentence required
a more complex structure), longer eye fxations and more
regressions occurred as the readers attempted to reanalyze
the sentence .
A stronger test of the garden-path model, however, requires
the examination of situations in which the semantic cues in
the sentence are clearly consistent with a syntactically com-
plex parsing alternative. A truly modular, syntax-driven
parser would be unafected by the semantic cues in the sen-
tence. Consider, for example, the sentence fragment Te
sofa scratched. Sofas are soft and inanimate and therefore
unlikely to scratch anything. Consequently, the semantic cues
in the fragment favor the more complex relative clause anal-
ysis, in which the sofa is the entity being scratched (as in Te
sofa scratched by the cat was given to Goodwill ). Initial results
seemed to suggest that the semantic cues had no efect on the
initial parse of the sentence; readers seemed to build the syn-
tactically simplest analysis possible, even when it was incon-
sistent with the available semantic information. Such evidence
led to the hypothesis that the language processor is comprised
of a number of autonomously functioning components, each of
which corresponds to a level of linguistic analysis (Ferreira and
Clifton 1986). Te syntactic component was presumed to func-
tion independently of the other components .
Te modular syntax-frst model has been increasingly chal-
lenged, most notably by advocates of constraint-satisfaction
models (Trueswell and Tanenhaus 1994). Tese models pro-
pose that all sources of relevant information (including statisti-
cal, semantic, and real-world information) simultaneously and
rapidly infuence the actions of the parser. Hence, the implau-
sibility of a sofa scratching something is predicted to cause
the parser to initially attempt the syntactically more complex
relative-clause analysis. Consistent with this claim, numer-
ous studies have subsequently demonstrated compelling
infuences of semantics and world knowledge on the parsers
response to syntactic ambiguity (ibid.).
Tere is, however, a fundamental assumption underlying
most of the syntactic ambiguity research (regardless of theo-
retical perspective): that syntax always controls combinatory
processing when the syntactic cues are unambiguous. Recently,
this assumption has also been challenged. Te challenge cen-
ters on the nature of thematic roles , which help to defne
the types of arguments licensed by a particular verb (McRae,
Ferretti, and Amyote 1997; Trueswell and Tanenhaus 1994).
Exactly what is meant by thematic role varies widely, especially
with respect to the amount of semantic and conceptual content
it is assumed to hold (McRae, Ferretti, and Amyote 1997). For
most syntax-frst proponents, a thematic role is limited to a few
syntactically relevant selectional restrictions , such as animacy
into distinct streams of processing, then syntactic and seman-
tic anomalies might afect the comprehension system in dis-
tinct ways. A large body of evidence suggests that syntactic
and semantic anomalies do in fact elicit qualitatively distinct
ERP efects, and that these efects are characterized by distinct
and consistent temporal properties. Semantic anomalies (e.g.,
Te cat will bake the food ) elicit a negative wave that peaks
at about 400 milliseconds after the anomalous word appears
(the N400 efect ) (Kutas and Hillyard 1980) . By contrast, syntac-
tic anomalies (e.g., Te cat will eating the food ) elicit a large
positive wave that onsets at about 500 milliseconds after pre-
sentation of the anomalous word and persists for at least half a
second (the P600 efect [Osterhout and Holcomb 1992]) . In some
studies, syntactic anomalies have also elicited a negativity over
anterior regions of the scalp, with onsets ranging from 100 to
300 milliseconds. Tese results generalize well across types
of anomaly, languages, and various methodological factors.
Te robustness of the efects seems to indicate that the human
brain does in fact honor the distinction between the form and
the meaning of a sentence .
SENTENCE COMPREHENSION. Assuming that sentence pro-
cessing involves distinct syntactic and semantic processing
streams, the question arises as to how these streams interact
during comprehension. A great deal of evidence indicates that
sentence processing is incremental , that is, that each succes-
sive word in a sentence is integrated into the preceding sen-
tence material almost immediately. Such a strategy, however,
introduces a tremendous amount of ambiguity that is,
uncertainty about the intended syntactic and semantic role of a
particular word or phrase. Consider, for example, the sentence
fragment Te cat scratched. Tere are actually two ways to
parse this fragment. One could parse it as a simple active sen-
tence, in which the cat is playing the syntactic role of subject
of the verb scratched and the semantic role of the entity doing
the scratching (as in Te cat scratched the ratty old sofa ). Or one
could parse it as a more complex relative clause structure, in
which the verb scratched is the start of a second, embedded
clause, and the cat is the entity being scratched, rather than
the one doing the scratching (as in Te cat scratched by the rac-
coon was taken to the pet hospital ). Te ambiguity is resolved
once the disambiguating information ( the ratty sofa or by the
raccoon ) is encountered downstream, but that provides little
help for a parser that assigns roles to words as soon as they are
encountered .
How does an incremental sentence-processing system
handle such ambiguities? An early answer to this question
was provided by the garden-path (or modular ) parsing models
developed in the1980s. Te primary claim was that the initial
parse of the sentence is controlled entirely by the syntactic cues
in the sentence (Ferreira and Clifton 1986). As words arrive in
the linguistic input, they are rapidly organized into a structural
analysis by a process that is not infuenced by semantic knowl-
edge. Te output of this syntactic process then guides semantic
interpretation. Tis model can be contrasted with interactive
models, in which a wide variety of information (e.g., seman-
tics and conceptual/world knowledge) infuences the earliest
stages of sentence parsing. Initial results of numerous studies
The Psychology of Linguistic Form
19
by changing the infectional morpheme at the end of the verb
to a passive form (Te mysterious crime was solved ).
Terefore, if meaning drives sentence processing in this situa-
tion, then the verb solving would be perceived to be in the wrong
syntactic form ( -ing instead of ed ), and should therefore elicit a
P600 efect . Kim and Osterhout observed that verbs like solving
elicited a P600 efect, showing that a strong semantic attrac-
tion between a predicate and an argument can determine
how words are combined, even when the semantic attraction
contradicts unambiguous syntactic cues. Conversely, in anom-
alous sentences with an identical structure but with no seman-
tic attraction between the subject noun and the verb (e.g., Te
envelope was devouring ), the critical verb elicited an N400
efect rather than a P600 efect. Tese results demonstrate that
semantics, rather than syntax, can drive word combinations
during sentence comprehension .
SENTENCE PRODUCTION. Generating a sentence requires the
rapid construction of novel combinations of linguistic units,
involves multiple levels of analysis, and is constrained by a vari-
ety of rules (about word order, the formation of complex words,
word pronunciation, etc). Errors are a natural consequence of
these complexities (Dell 1995). Because they tend to be highly
systematic, speech errors have provided much of the data upon
which current models of sentence production are based. For
example, word exchanges tend to obey a syntactic category rule ,
in that the exchanged words are from the same syntactic category
(for example, two nouns have been exchanged in the utterance
Stop hitting your brick against a head wall ). Te systematicity of
speech errors suggests that regularities described in theories of
linguistic form also play a role in the speech-planning process.
Te dominant model of sentence production is based
on speech error data (Dell 1995; Garrett 1975; Levelt 1989).
According to this model, the process of preparing to speak a
sentence involves three stages of planning: conceptualization,
formulation, and articulation, in that order. During the concep-
tualization stage, the speaker decides what thought to express
and how to order the relevant concepts sequentially. Te for-
mulation stage begins with the selection of a syntactic frame to
encode the thought; the frame contains slots that act as place
holders for concepts and, eventually, specifc words. Te phono-
logical string is translated into a string of phonological features,
which then drive the motor plan manifested in articulation.
Tis model, therefore, posits the existence of representa-
tions of syntactic structure that are distinct from the represen-
tations of meaning and sound. Other evidence in support of this
view comes from the phenomenon of syntactic priming : Having
heard or produced a particular syntactic structure, a person
is more likely to produce sentences using the same syntactic
structure (Bock 1986). Syntactic priming occurs independently
of sentence meaning, suggesting that the syntactic frames are
independent forms of representation that are quite distinct
from meaning .
CONCLUSIONS
Collectively, the evidence reviewed in this essay indicates that
psychologically relevant representations of linguistic form
(Chomsky 1965); thematic roles are treated as (largely mean-
ingless) slots to be flled by syntactically appropriate fllers. A
second view is that there is a limited number of thematic roles
(agent, theme, benefactor, and so on), and that a verb selects a
subset of these (Fillmore 1968). Although this approach attri-
butes a richer semantics to thematic roles, the required gener-
alizations across large classes of verbs obscure many subtleties
in the meaning and usage of these verbs .
Both of these conceptions of thematic roles exclude knowl-
edge that people possess concerning who tends to do what to
whom in particular situations. Ken McRae and others have pro-
posed a third view of thematic roles that dramatically expands
their semantic scope: Tematic roles are claimed to be rich,
verb-specifc concepts that refect a persons collective expe-
rience with particular actions and objects (McRae, Ferretti,
and Amyote 1997). Tese rich representations are claimed to
be stored as a set of features that defne gradients of typical-
ity (situation schemas ), and to comprise a large part of each
verbs meaning. One implication is that this rich knowledge
will become immediately available once a verbs meaning has
been retrieved from memory. As a consequence, the plausibil-
ity of a particular word combination need not be evaluated by
means of a potentially complex inferential process, but rather
can be evaluated immediately in the context of the verbs mean-
ing. One might therefore predict that semantic and conceptual
knowledge of events will have profound and immediate efects
on the way in which words are combined during sentence pro-
cessing. McRae and others have provided evidence consistent
with these claims, including semantic infuences on syntactic
ambiguity resolution.
Te most compelling argument against the absolute pri-
macy of syntax, however, would be evidence that semantic and
conceptual knowledge can take control of sentence process-
ing even when opposed by contradicting and unambiguous
syntactic cues. Recent work by Ferranda Ferreira (2003) sug-
gests that this might happen on some occasions. She reported
that when plausible sentences (e.g., Te mouse ate the cheese)
were passivized to form implausible sentences (e.g., Te mouse
was eaten by the cheese ), participants tended to name the wrong
entity as do-er or acted-on, as if coercing the sentences to
be plausible. However, the processing implications of these
results are uncertain, due to the use of postsentence rumina-
tive responses, which do not indicate whether semantic infu-
ences refect the listeners initial responses to the input or some
later aspect of processing.
Researchers have also begun to explore the infuence of
semantic and conceptual knowledge on the on-line process-
ing of syntactically unambiguous sentences. An illustrative
example is a recent ERP study by Albert Kim and Lee Osterhout
(2005). Te stimuli in this study were anomalous sentences that
began with a active structure, for example, Te mysterious crime
was solving. Te syntactic cues in the sentence require that
the noun crime be the Agent of the verb solving . If syntax drives
sentence processing, then the verb solving would be perceived
to be semantically anomalous , as crime is a poor Agent for the
verb solve, and therefore should elicit an N400 efect . However,
although crime is a poor Agent, it is an excellent Teme (as in
solved the crime ). Te Teme role can be accommodated simply
The Cambridge Encyclopedia of the Language Sciences
20
Byrd , Dani, and Elliot Saltzman . 2003 . Te elastic phrase: Modeling
the dynamics of boundary-adjacent lengthening . Journal of
Phonetics 31 : 149 80.
Caplan , David . 1995 . Issues arising in contemporary studies of disor-
ders of syntactic processing in sentence comprehension in agram-
matic patients . Brain and Language 50 : 325 38.
Caramazza , Alfonzo, and Edgar Zurif . 1976 . Dissociations of algorith-
mic and heuristic processes in language comprehension: Evidence
from aphasia . Brain and Language 3 : 572 82.
Chistovich , Ludmilla A 1960 . Classifcation of rapidly repeated speech
sounds . Akustichneskii Zhurnal 6 : 392 98.
Chomsky , Noam. 1957 . Syntactic Structures . Te Hague : Mouton .
. 1965 . Aspects of the Teory of Syntax . Cambridge, MA : MIT
Press .
Chomsky , Noam , and Morris Halle . 1968 . Te Sound Pattern of English .
New York : Harper and Row .
Delattre , Pierre, and Donald Freeman . 1968 . A dialect study of
American Rs by x-ray motion picture . Linguistic s 44 : 29 68.
Delattre , Pierre C. , Avin M. Liberman , and Franklin S. Cooper . 1955 .
Acoustic loci and transitional cues for consonants . Journal of the
Acoustical Society of America 27 : 769 73.
Dell , Gary S . 1995 . Speaking and misspeaking. In An Invitation to
Cognitive Science: Language . Cambridge, MA : MIT Press .
Draper , M. , P. Ladefoged , and D. Whitteridge . 1959 . Respiratory mus-
cles in speech . Journal of Speech and Hearing Research 2 : 16 27.
Dudley , Homer. 1940 . Te carrier nature of speech . Bell System
Technical Journal 14 : 495 515.
Elman , Jefrey L. 1990 . Representation and structure in connection-
ist models. In Cognitive Models of Speech Processing , ed. G. T. M.
Altmann , 22760. Cambridge, MA : MIT Press .
Elman , Jefrey L. , Elizabeth A. Bates , A. Karmilof-Smith , D. Parisi , and
K. Plunkett . 1996 . Rethinking innateness. Cambridge, MA: MIT
Press.
Emmory , Karen. 2002 . Language, Cognition, and the Brain: Insights
from Sign Language Research . Mahwah, NJ : Lawrence Erlbaum
Associates .
Ferreira , Fernanda . 2003 . Te misinterpretation of noncanonical sen-
tences . Cognitive Psychology 47 : 164 203.
Ferreira , Fernanda, and Charles Clifton , Jr. 1986 . Te independence of
syntactic processing . Journal of Memory and Language 25 : 348 68.
Fillmore , Charles . 1968 . Te case for case. In Universals of Linguistic
Teory , ed. E. Bach , 180. New York : Holt, Rinehart, & Winston .
Fowler , Carol A. 1986 . An event approach to the study of speech per-
ception from a direct-realist perspective . Journal of Phonetics
14 : 3 28.
. 1995 . Speech production. In Speech, Language, and Communi-
cation , ed. J. L. Miller and P. D. Eimas , 2961. New York : Academic
Press .
. 1996 . Listeners do hear sounds not tongues . Journal of the
Acoustical Society of America 99 : 1730 41.
Fowler , Carol , Susan Napps , and Laurie Feldman . 1985 . Relations
among regular and irregular morphologically related words in the
lexicon as revealed by repetition priming . Memory and Cognition
13 : 241 55.
Franklin , S. , J. Lambon Ralph , J. Morris , and P. Bailey . 1996 . A distinc-
tive case of word meaning deafness? Cognitive Neuropsychology
13 : 1139 62.
Frazier , L. 1987 . Sentence processing: A tutorial review. In Attention
and Performance XII: Te Psychology of Reading , ed. M. Coltheart ,
330 . Hillsdale, NJ : Lawrence Erlbaum Associates .
. 1995 . Representation in psycholinguistics. In Speech,
Language, and Communication , ed. J. L. Miller and P. D. Eimas , 127.
New York : Academic Press .
exist at all levels of language, from sounds to sentences. At each
level, units of linguistic form are combined in systematic ways
to form larger units of representation. For the most part, these
representations seem to be abstract; that is, they are distinct
from the motor movements, sensory experiences, and episodic
memories associated with particular utterances. However, it
is also clear that more holistic (that is, nondecompositional)
representations of linguistic form, some of which are rooted
in specifc episodic memories, also play a role in language
processing.
It also seems to be true that linguistic forms (e.g., the mor-
phological structure of a word or the syntactic structure of a
sentence) are dissociable from the meanings they convey.
At the same time, semantic and conceptual knowledge can
strongly infuence the processing of linguistic forms, as exem-
plifed by semantic transparency efects on word decomposi-
tion and thematic role efects on sentence parsing.
Tese conclusions represent substantive progress in our
understanding of linguistic form and the role it plays in lan-
guage processing. Nonetheless, answers to some of the most
basic questions remain contentiously debated, such as the pre-
cise nature of the rules of combination, the relative roles of
compositional and holistic representations, and the pervasive-
ness of interactions between meaning and form .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Alexander , M. P. 2006 . Aphasia I: Clinical and anatomical issues.
In Patient-Based Approaches to Cognitive Neuroscience (2d ed.),
ed. M. J. Farah and T. E. Feinberg , 16582. Cambridge, MA : MIT
Press .
Allen , Mark D. 2005 . Te preservation of verb subcategory knowledge
in a spoken language comprehension defcit . Brain and Language
95 : 255 64.
Allen , Mark , and William Badecker . 1999 . Stem homograph inhibition
and stem allomorphy: Representing and processing infected forms
in a multilevel lexical system. Journal of Memory and Language
41 : 105 23.
Allport , D. A. , and E. Funnell . 1981 . Components of the mental lexi-
con . Philosophical Transactions of the Royal Society of London B
295 : 397 410.
Anderson , Stephen R . 1985 . Phonology in the Twentieth Century: Teories
of rules and Teories of Representations . Chicago : Te University of
Chicago Press .
Bell , Alexander M. 1867 . Visible Speech: Te Science of Universal
Alphabetics . London : Simpkin, Marshal .
Bloomfeld , Leonard . 1933 . Language. New York : H. Holt & Co .
Bock , J. Katherine . 1986 . Syntactic persistence in language produc-
tion . Cognitive Psychology 18 : 355 87.
Bock , J. K., and Anthony S. Kroch. 1989 . Te isolability of syntac-
tic processing. In Linguistic Structure in Language Processing , ed..
G. N. Carlson and M. K. Tanenhaus , 15796. Boston : Kluwer
Academic .
Brentari , Dianne . 1998 . A Prosody Model of Sign Language Phonology .
Cambridge, MA : MIT Press .
Browman , Catherine P., and Louis Goldstein . 1990 . Gestural specif-
cation using dynamically-defned articulatory structures . Journal
of Phonetics 18 : 299 320.
Bybee , Joan . 1988 . Morphology as lexical organization. In Teoretical
Morphology: Approaches in Modern Linguistics , ed. M. Hammond
and M. Noonan , 11941. San Diego, CA : Academic Press .
The Psychology of Linguistic Form
21
McClelland , Jamesv , and David Rumelhart . 1986 . Parallel Distributed
Processing: Explorations in the Microstructure of Cognition . Vol. 1.
Cambridge, MA : MIT Press .
McGurk , Harry , and John Macdonald . 1976 . Hearing lips and seeing
voices . Nature 264 : 746 8.
McRae , Ken , Todd R. Ferretti , and Liane Amyote . 1997 . Tematic roles
as verb-specifc concepts . Language and Cognitive Processes 12.2:
137 76.
Meier , Richard P. , Kearsy Cormier , and David Quinto-Pozos . 2002 .
Modularity in Signed and Spoken Languages. Cambridge : Cambridge
University Press .
Meyer , Antje , and Kathryn Bock . 1992 . Tip-of-the-tongue phe-
nomenon: Blocking or partial activation? Memory and Cognition
20 : 715 26.
Miceli , G. , L. Mazzuchi , L. Menn , and H. Goodglass . 1983 . Contrasting
cases of Italian agrammatic aphasia without comprehension disor-
der . Brain and Language 19 : 65 97.
Miller , George A. , and Patricia E. Nicely . 1955 . An analysis of percep-
tual confusions among some English consonants . Journal of the
Acoustical Society of America 27 : 329 35.
Minife , Fred D. 1999 . Te history of physiological phonetics in the
United States. In A Guide to the History of the Phonetic Sciences in
the United States: Issued on the Occasion of the 14th International
Congress of Phonetic Sciences, San Francisco, 17 August 1999 , ed.
J. Ohala , A. Bronstein , M. Bus , L. Grazio , J. Lewis , and W. Weigel .
Berkeley : University of California Press. .
Moore , Brian C. J. 1989 . An Introduction to the Psychology of Hearing . 3d
ed. London : Academic Press.
Nygaard , Lynn C. , and David B. Pisoni . 1995 . Speech
Perception: New directions in research and theory. In Speech,
Language, and Communication , ed. J. Miller and P. Eimas , 6396.
New York : Academic Press .
Osterhout , Lee , and Philip J. Holcomb . 1992 . Event-related brain
potentials elicited by syntactic anomaly . Journal of Memory and
Language 31 : 785 806.
Passy , Paul. 1888 . Our revised alphabet . Te Phonetic Teacher, 57 60.
Pinker , Steven. 1999 . Words and Rules: Te Ingredients of Language .
New York : Basic Books .
Potter , Ralph K. 1945 . Visible patterns of sound . Science 102 : 463 70.
Rousselot , P.-J . 1897 1901. Principes de Phontique Experimentale .
Paris : H. Welter .
Rueckl , Jay , Michelle Mikolinski , Michal Raveh , Caroline Miner , and F.
Mars . 1997 . Morphological priming, fragment completion, and con-
nectionist networks . Journal of Memory and Language 36 : 382 405.
Saltzman , Elliot L. , and Kevin G. Munhall . 1989 . A dynamical
approach to gestural patterning in speech production . Ecological
Psychology 1 : 333 82.
Saussure , Ferdinand de . [2002] 2006 . crits de linguistique gnrale , ed.
Simon Bouquet and Rudolf Engler . Paris: Gallimard. English trans-
lation: Writings in General Linguistics , Oxford : Oxford University
Press .
Scharenborg , O. , D. Norris , L. ten Bosch , and J. M. McQueen . 2005 .
How should a speech recognizer work? Cognitive Science
29 : 867 918.
Schriefers , Herbert , Antje Meyer and Willem Levelt . 1990 . Exploring
the time course of lexical access in language production: Picture-
word interference studies . Journal of Memory and Language
29 : 86 102.
Scripture , Edward Wheeler . 1902 . Te Elements of Experimental
Phonetics . New York : Charles Scribners Sons .
Sievers , Eduard . 1876 . Grundzuge der Lautphysiologie zur Einfuhrung
in das Studium der Lautlehere der Indogermanischen Sprachen .
Leipzig : Breitkopf and Hartel .
Garrett , Merrill F. 1975 . Te analysis of sentence production. In Te
Psychology of Learning and Motivation , ed. G. Bower , 13377. New
York : Academic Press .
Goldinger , Stephen D. , and Tamiko Azuma . 2003 . Puzzle-solving
science: Te quixotic quest for units of speech perception . Journal
of Phonetics 31 : 305 20.
Goldinger , Stephen D. , David B. Pisoni , and John S. Logan . 1991 . On
the nature of talker variability efects on recall of spoken word
lists . Journal of Experimental Psychology: Learning, Memory and
Cognition 17 : 152 62.
Hall , D. A. , and M. J. Riddoch . 1997 . Word meaning deafness: Spelling
words that are not understood . Cognitive Neuropsychology
14 : 1131 64.
Hillis , Argye E. 2000 . Te organization of the lexical system. In
What Defcits Reveal about the Human Mind/Brain: Handbook
of Cognitive Neuropsychology , ed. B. Rapp , 185210. New
York : Psychology Press .
House , Arthur S., and Grant Fairbanks. 1953 . Te infuence of con-
sonant enviroment upon the secondary acoustical characteristics
of vowels . Journal of the Acoustical Society of America 25 : 105 13.
Jackendof , Ray . 1997 . Te Architecture of the Language Faculty .
Cambridge, MA : MIT Press .
Jakobson , Roman . 1939 . Observations sur le classment pho-
nologique des consonnes . Proceedings of the 3rd International
Conference of Phonetic Sciences, 34 41. Ghent.
Jakobson , Roman , Gunnar Fant , and Morris Halle . 1952 . Preliminaries
to Speech Analysis . Cambridge, MA : MIT Press .
Joos , Martin . 1948 . Acoustic Phonetics . Language Monograph 23,
Supplement to Language 24 : 1 36.
Kempen , Gerard , and Pieter Huijbers . 1983 . Te lexicalization
process in sentence production and naming: Indirect election of
words . Cognition 14 : 185 209.
Kim , Albert, and Lee Osterhout . 2005 . Te independence of combi-
natory semantic processing: Evidence from event-related poten-
tials . Journal of Memory and Language 52 : 205 25.
Kiparsky , Paul. 1979 . Panini as a Variationist . Cambridge, MA : MIT
Press .
Khnert , Barbara , and Francis Nolan . 1999 . Te origin of coartic-
ulation. In Coarticulation: Teory, Data and Techniques , ed. B.
Rapp, 730. Cambridge : Cambridge University Press .
Kutas , Marta , and Steven A. Hillyard . 1980 . Reading senseless
sentences: Brain potentials refect semantic anomaly . Science
207 : 203 5.
Ladefoged , P. , and N. McKinney . 1963 . Loudness, sound pressure,
and subglottal pressure in speech . Journal of the Acoustical
Society of America 35 : 454 60.
Lambon , Ralph M. , K. Sage , and A. Ellis . 1996 . Word meaning blind-
ness: A new form of acquired dyslexia . Cognitive Neuropsychology
13 : 617 39.
Levelt , Willem . 1989 . Speaking: From intention to articulation .
Cambridge, MA : MIT Press .
Liberman , Alvin M. , Pierre C. Delattre , Franklin S. Cooper , and
Lou J. Gerstman . 1954 . Te role of consonant vowel transitions
in the perception of the stop and nasal consonants . Journal of
Experimental Psychology 52 : 127 37.
Liberman , Alvin M. , and Ignatius G. Mattingly . 1985 . Te motor
theory of speech perception revised . Cognition 21 : 1 36.
Linebarger , Marcia , Myrna Schwartz , and Eleanor Safran . 1983 .
Sensitivity to grammatical structure in so-called agrammatic
aphasics . Cognition 13 : 361 93.
Massaro , Dominic W . 1987 . Speech Perception by Ear and Eye: A
Paradigm for Psychological Inquiry . Hillsdale, NJ : Lawrence
Erlbaum Associates .
The Cambridge Encyclopedia of the Language Sciences
22
Sweet , Henry. 1881 . Sound notation . Transactions of the Philological
Society: 177 235.
Trueswell , John C. , and Michael K. Tanenhaus . 1994 . Toward a lexi-
calist framework of constraint-based syntactic ambiguity reso-
lution. In Perspectives on Sentence Processing , ed. C. Clifton , L.
Frazier, and K. Rayner , 15580. Hillsdale, NJ : Lawrence Erlbaum
Associates .
Soli , Sig D . 1982 . Structure and duration of vowels together specify frica-
tive voicing . Journal of the Acoustical Society of America 72 : 366 78.
Stetson , Raymond H. 1905 . A motor theory of rhythm and discrete
succession II . Psychological Review 12 : 293 350.
. 1951 . Motor Phonetics: A Study of Speech Movement in Action . 2d
ed. Amsterdam : North-Holland .
Studdert-Kennedy , Michael . 1980 . Speech perception . Language and
Speech 23 : 45 65.
23
(2) a. Te girl laughed and sang .
b. Te girl laughed .
Tis is an example of structural entailment , because it is the
structure itself that allows the inference (i.e., if someone does
both A and B, then someone does A). Tis particular rule is
essentially the classical inference rule of conjunction elimina-
tion from propositional logic; that is,
(3)
A B
A
While this relies on a largely syntactic notion of entailment,
semantics should also explain how (4b) is a legitimate infer-
ence from (4a).
(4) a. Te drought killed the crops.
b. Te crops died .
Such lexical entailments involve an inference that is tied directly
to the meaning of a word, namely, the verb kill ; that is, when
something is killed, it dies. Hence, the role of lexical informa-
tion in the construction of logical forms and the inferences we
can compute from our utterances is an important area of lin-
guistics, and one we return to in Section 3.5 below .
Tere is an important distinction in semantics among prop-
ositions , sentences , and utterances. We can think of an
utterance as a speech-act , situated in time and space, that is,
which happens at a particular time and location. A sentence, on
the other hand, is a expression that is inherently linguistic, and
can be expressed on multiple occasions by multiple utterances.
Te notion of a proposition is more complex and contentious, but
it is that object that is traditionally taken as being true or false,
expressed by the sentence when uttered in a specifc context.
1.1 Historical Remarks
Te study of meaning has occupied philosophers for centuries,
beginning at least with Platos theory of forms and Aristotles
theory of meaning . Locke, Hume, and Reid all pay particular
attention to the meanings of words in composition, but not until
the late nineteenth century do we see a systematic approach to
the study of logical syntax emerge, with the work of Bertrand
Russell and Gottlob Frege. Russell and Frege were not interested
in language as a linguistic phenomenon, but rather as a medium
through which judgments can be formed and expressed. Freges
focus lay in formulating the rules that create meaningful expres-
sions in a compositional manner, while also introducing an
important distinction between an expressions sense and its ref-
erence (cf. sense and reference , reference and exten-
sion ). Russells work on the way in which linguistic expressions
denote introduced the problem of definite descriptions
and referential failure, and what later came to be recognized as
the problem of presupposition (cf. pragmatics ).
Ferdinand de Saussure ([ 1916 ] 1983), working within an
emerging structuralist tradition, developed relational tech-
niques for linguistic analysis, which were elaborated into a
framework of componential analysis for language meaning.
Te idea behind componential analysis is the reduction of a
words meaning into its ultimate contrastive elements. Tese
3
THE STRUCTURE OF MEANING
James Pustejovsky
1 INTRODUCTION
Semantics is the systematic study of meaning in language. As a
discipline, it is directed toward the determination of how humans
reason with language, and more specifcally, discovering the pat-
terns of inference we employ through linguistic expressions.Te
study of semantics has diverse traditions, and the current litera-
ture is quite heterogeneous and divided on approaches to some
of the basic issues facing the feld (cf. semantics ). While most
things in the world have meaning to us, they do not carry mean-
ing in the same way as linguistic expressions do. For example,
they do not have the properties of being true or false, or ambigu-
ous or contradictory. (See Davis and Gillon [2004] for discussion
and development of this argument.) For this and other reasons,
this overview essay addresses the question of how linguistic
expressions carry meaning and what they denote in the world .
Where syntax determines the constituent structure of a sen-
tence along with the assignment of grammatical and thematic
relations, it is the role of semantics to compute the deeper
meaning of the resulting expression. For example, the two sen-
tences in (1) difer in their syntactic structures (through their
voice), but they mean essentially the same thing; that is, their
propositional content is identical.
(1) a. Te child ate a cookie.
b. A cookie was eaten by the child.
Early on, such observations led philosophers and linguists to
distinguish meaning from the pure structural form of a sen-
tence (Saussure [ 1916 ] 1983; Russell 1905 ). Semantic theories in
linguistics assume that some sort of logical form is computed
from the constituent structure associated with a sentence, and
it is this meaning representation that allows us to make cate-
gorical and truth-conditional judgments, such as the equiva-
lence in meaning of the two sentences in (1).
Another role played by semantics is in the computation of
inferences from our utterances, such as entailments, implica-
tures, and presuppositions . For example, consider the var-
ious notions of entailment . From the logical form (LF) of the
sentence in (2a), semantics enables us to infer (2b) as a legiti-
mate inference.
The Cambridge Encyclopedia of the Language Sciences
24
is the notion of compositionality (cf. compositionality ). As
speakers of language, we understand a sentence by under-
standing its parts and how they are put together. Te principle
of compositionality characterizes how smaller units of mean-
ing are put together to form larger, more meaningful expres-
sions in language. Te most famous formulation of this notion
comes from Frege, paraphrased as follows:
Te meaning of an expression is a function of the meanings of
its parts and the way they are syntactically combined. (Partee
1984 )
Tis view has been extremely infuential in semantics
research over the past 40 years. If one assumes a compositional
approach to the study of meaning, then two things immediately
follow: 1) One must specify the specifc meaning of the basic
elements of the language, and 2) one must formulate the rules
of combination for how these elements go together to make
more complex expressions. Te frst aspect includes determin-
ing what words and morphemes mean, that is, lexical seman-
tics , which we address in the next section. Te second aspect
entails defning a calculus for how these elements compose to
form larger expressions, that is, argument selection and modi-
fcation . Needless to say, in both of these areas, there is much
divergence of opinion, but semanticists generally agree on the
basic assumptions inherent in compositionality .
2 LEXICAL MEANING
Semantic interpretation requires access to knowledge about
words. Te lexicon of a grammar must provide a systematic and
ef cient way of encoding the information associated with words
in a language. lexical semantics is the study of what words
mean and how these meanings are structured. Te lexicon is
not merely a collection of words with their semantic forms, but
rather a set of structured objects that participate in larger opera-
tions and compositions, both enabling syntactic environments
and acting as signatures to semantic entailments and implica-
tures in the context of larger discourse .
Tere are four basic questions in modeling the semantic
content and structure of the lexicon: 1) What semantic infor-
mation goes into a lexical entry? 2) How do lexical entries
relate semantically to one another? 3) How is this informa-
tion exploited compositionally by the grammar? 4) How is this
information available to semantic interpretation generally?
Te lexicon and lexical semantics have traditionally been
viewed as the most passive modules of language, acting in the ser-
vice of the more dynamic components of the grammar. Tis view
has its origins in the generative tradition (Chomsky [ 1955 ] 1975)
and has been an integral part of the notion of the lexicon ever
since. While the Aspects -model of selectional features (Chomsky
1965 ) restricted the relation of selection to that between lexical
items, work by McCawley ( 1968 ) and Jackendof ( 1972 ) showed
that selectional restrictions must be available to computations
at the level of derived semantic representation rather than at
deep structure. Subsequent work by Joan Bresnan (1982), Gerald
Gazdar et al. (1985), and C. Pollard and I. Sag ( 1994 ) extend the
range of phenomena that can be handled by the projection and
exploitation of lexically derived information in the grammar.
contrastive elements are structured in a matrix, allowing for
dimensional analysis and generalizations to be made about
lexical sets occupying the cells in the matrix.
Tis technique developed into a general framework for lin-
guistic description called distinctive FEATURE ANALYSIS . Tis is
essentially the inspiration for J. Katz and J. Fodors 1963 the-
ory of lexical semantics within transformational grammar. On
this theory, usually referred to as markerese , a lexical entry in
the language consists of grammatical and semantic markers,
and a special feature called a semantic distinguisher . In the
subsequent discussion by U. Weinreich ( 1972 ) and many oth-
ers, this model was demonstrated to be far too impoverished to
characterize the compositional mechanisms inherent in lan-
guage . In the late 1960s and early 1970s, alternative models of
word meaning emerged (Fillmore 1968 [ frame semantics ];
Lakof [1965] 1970 [generative semantics]; Gruber 1976;
Jackendof 1972 ), which respected the relational structure of
sentence meaning while encoding the named semantic func-
tions in lexical entries. In D. R. Dowty ( 1979 ), a model theoretic
interpretation of the decompositional techniques of G. Lakof,
J. D. McCawley, and J. R. Ross was developed.
In the later twentieth century, montague grammar
(Montague 1973, 1974) was perhaps the most signifcant develop-
ment in the formal analysis of linguistic semantics, as it brought
together a systematic, logically grounded theory of composition-
ality, with a model theoretic interpretation. Subsequent work
enriched this approach with insights from D. Davidson ( 1967 ),
H. P. Grice ( 1969 ), Saul Kripke ([ 1972 ] 1980), David Lewis ( 1976 ),
and other philosophers of language (cf. Partee 1976 ; Davidson
and Harman 1972 ) .
Recently, the role of lexical-syntactic mapping has become
more evident, particularly with the growing concern over projec-
tion from lexical semantic form, the problem of verbal alternations
and polyvalency, and the phenomenon of polysemy. Te work of
R. Jackendof (1983, 1997) on conceptual semantics has come to
the fore, as the feld of lexical semantics has developed into a more
systematic and formal area of study (Pustejovsky and Boguraev
1993 ; Copestake and Briscoe 1995, 1567).
Finally, one of the most signifcant developments in the
study of meaning has been the dynamic turn in how sentences
are interpreted in discourse. Inspired by the work of Irene Heim
( 1982 ) and H. Kamp ( 1981 ), the formal analysis of discourse has
become an active and growing area of research, as seen in the
works of Jeroen Groenendijk and Martin Stokhof ( 1991 ), Kamp and
U. eyle ( 1993 ), and Nicholas Asher and Alex Lascarides ( 2003 ).
In the remainder of this essay, we examine the basic prin-
ciple of how meanings are constructed. First, we introduce
the notion of compositionality in language. Since words are
the building blocks of larger meanings, we explore various
approaches to lexical semantics. Ten, we focus on how units
of meaning are put together compositionally to create proposi-
tions. Finally, we examine the meaning of expressions above
the level of the sentence, within a discourse .
1.2 Compositionality
Because semantics focuses on how linguistic expressions come
to have meaning, one of the most crucial concepts in the feld
The Structure of Meaning
25
and alternations, among other relations (cf. Pustejovsky and
Boguraev 1993 ) .
2.2 Argument Structure
Once the basic semantic types for the lexical items in the
language have been specifed, their subcategorization and
selectional information must be encoded in some form. Te
argument structure for a word can be seen as the simplest spec-
ifcation of its semantics, indicating the number and type of
parameters associated with the lexical item as a predicate. For
example, the verb die can be represented as a predicate taking
one argument, and kill as taking two arguments, while the verb
give takes three arguments.
(5) a. die(x)
b. kill(x,y)
c. give(x,y,z)
What originally began as the simple listing of the parameters
or arguments associated with a predicate has developed into
a sophisticated view of the way arguments are mapped onto
syntactic expressions. E. Williamss ( 1981 ) distinction between
external (the underlined arguments for kill and give) and
internal arguments and J. Grimshaws proposal for a hierarchi-
cally structured representation (cf. Grimshaw 1990 ) provide
us with the basic syntax for one aspect of a words meaning.
Similar remarks hold for the argument list structure in HPSG
( head-driven phrase structure grammar ) and LFG
( lexical-functional grammar ) .
One infuential way of encoding selectional behavior has
been the theory of thematic relations (cf. thematic roles ;
Gruber 1976 ; Jackendof 1972 ). Tematic relations are now gen-
erally defned as partial semantic functions of the event being
denoted by the verb or noun, and behave according to a pre-
defned calculus of role relations (e.g., Carlson 1984; Dowty
1991; Chierchia 1989 ). For example, semantic roles, such as
agent, theme, and goal, can be used to partially determine
the meaning of a predicate when they are associated with the
grammatical arguments to a verb.
(6) a. put < AGENT,THEME,LOCATION >
b. borrow < RECIPIENT,THEME,SOURCE >
Tematic roles can be ordered relative to each other in terms
of an implicational hierarchy. For example, there is consider-
able use of a universal subject hierarchy such as shown in the
following (cf. Fillmore 1968 ; Comrie 1981 ) .
(7) AGENT > RECIPIENT/BENEFACTIVE > THEME/PATIENT >
INSTRUMENT > LOCATION
Many linguists have questioned the general explanatory
coverage of thematic roles, however, and have chosen alter-
native methods for capturing the generalizations they prom-
ised. Dowty ( 1991 ) suggests that theta-role generalizations
are best captured by entailments associated with the predi-
cate itself. A theta-role can then be seen as the set of predicate
entailments that are properties of a particular argument to the
verb. Characteristic entailments might be thought of as proto-
type roles, or proto-roles; this allows for degrees or shades of
Recently, with the convergence of several areas in linguis-
tics ( lexical semantics , computational lexicons, type theories),
several models for the determination of selection have emerged
that put even more compostional power in the lexicon, making
explicit reference to the paradigmatic systems that allow for
grammatical constructions to be partially determined by selec-
tion. Examples of this approach are generative lexicon theory
(Pustejovsky 1995 ; Bouillon and Busa 2001 ), and construc-
tion grammar (Goldberg, 1995 ; Jackendof 1997 , 2002 ). Tese
developments have helped to characterize the approaches to
lexical design in terms of a hierarchy of semantic expressive-
ness. Tere are at least three such classes of lexical description,
defned as follows : sense enumerative lexicons , where lexical
items have a single type and meaning, and ambiguity is treated
by multiple listings of words; polymorphic lexicons , where lexi-
cal items are active objects, contributing to the determination of
meaning in context, under well-defned constraints; and unre-
stricted sense lexicons , where the meanings of lexical items are
determined mostly by context and conventional use . It seems
clear that the most promising direction seems to be a careful
and formal elucidation of the polymorphic lexicons , and this will
form the basis of our subsequent discussion.
Lexical items can be systematically grouped according to
their syntactic and semantic behavior in the language. For this
reason, there have been two major traditions of word cluster-
ing, corresponding to this distinction. Broadly speaking, for
those concerned mainly with grammatical behavior, the most
salient aspect of a lexical item is its argument structure ; for those
focusing on a words entailment properties, the most important
aspect is its semantic class . In this section, we examine these
two approaches and see how their concerns can be integrated
into a common lexical representation.
2.1 Semantic Classes
One of the most common ways to organize lexical knowledge
is by means of type or feature inheritance mechanisms (Evans
and Gazdar 1990 ; Carpenter 1992 ; Copestake and Briscoe
1992 ; Pollard and Sag 1994 ). Furthermore, T. Briscoe, V. de
Paiva, and A. Copestake ( 1993 ) describe a rich system of types
for allowing default mechanisms into lexical type descrip-
tions. Similarly, type structures, such as that shown in Figure
1 , can express the inheritance of syntactic and semantic fea-
tures, as well as the relationship between syntactic classes
Natural Entity
Physical
Mass Individuated
animate
human
inanimate
rock
Abstract
Mental Experiential
Figure 1.
The Cambridge Encyclopedia of the Language Sciences
26
12 CAUSE x, BECOME NOT ALIVE y
Here, the predicate CAUSE is represented as a relation between
an individual causer x and an expression involving a change
of state in the argument y . R. Carter ([ 1976 ] 1988) proposes a
representation quite similar, shown here for the causative verb
darken :
13 x CAUSE y BE.DARK CHANGE
Although there is an intuition that the cause relation
involves a causer and an event, neither Lakof nor Carter makes
this commitment explicitly. In fact, it has taken several decades
for Davidsons ( 1967 ) observations regarding the role of events
in the determination of verb meaning to fnd their way con-
vincingly into the major linguistic frameworks. Recently, a new
synthesis has emerged that attempts to model verb meanings
as complex predicative structures with rich event structures
(cf. Parsons 1990 ; Pustejovsky 1991b; Tenny 1992 ; Krifka 1992 ).
Tis research has developed the idea that the meaning of a verb
can be analyzed into a structured representation of the event
that the verb designates, and has furthermore contributed to
the realization that verbs may have complex, internal event
structures. Recent work has converged on the view that com-
plex events are structured into an inner and an outer event,
where the outer event is associated with causation and agency
and the inner event is associated with telicity (completion) and
change of state (cf. Tenny and Pustejovsky 2000 ; Levin and
Rappaport Hovav 2005).
Jackendof ( 1990 ) develops an extensive system of what he
calls Conceptual Representations , which parallel the syntactic
representations of sentences of natural language. Tese employ
a set of canonical predicates, including CAUSE, GO, TO, and ON,
and canonical elements, including Ting , Path , and Event . Tese
approaches represent verb meaning by decomposing the predi-
cate into more basic predicates. Tis work owes obvious debt to
the innovative work within generative semantics, as illustrated
by McCawleys ( 1968 ) analysis of the verb kill . Recent versions
of lexical representations inspired by generative semantics can
be seen in the Lexical Relational Structures of K. Hale and S.
J. Keyser (1993), where syntactic tree structures are employed
to capture the same elements of causation and change of state
as in the representations of Carter, Levin and T. Rapoport,
Jackendof, and Dowty. Te work of Levin and Rappaport, build-
ing on Jackendofs Lexical Conceptual Structures , has been
infuential in further articulating the internal structure of verb
meanings (see Levin and Rappaport 1995 ) .
J. Pustejovsky (1991b) extends the decompositional approach
presented in Dowty ( 1979 ) by explicitly reifying the events and
subevents in the predicative expressions. Unlike Dowtys treat-
ment of lexical semantics, where the decompositional calcu-
lus builds on propositional or predicative units (as discussed
earlier), a syntax of event structure makes explicit reference
to quantifed events as part of the word meaning. Pustejovsky
further introduces a tree structure to represent the temporal
ordering and dominance constraints on an event and its subev-
ents. For example, a predicate such as build is associated with
a complex event such as that shown in the following (cf. also
Moens and Steedman 1988 ).
meaning associated with the arguments to a predicate. Others
have opted for a more semantically neutral set of labels to
assign to the parameters of a relation, whether it is realized as
a verb, noun, or adjective. For example, the theory of argument
structure as developed by Williams ( 1981 ), Grimshaw ( 1990 ),
and others can be seen as a move toward a more minimal-
ist description of semantic diferentiation in the verbs list of
parameters.
Te interaction of a structured argument list and a rich
system of types, such as that presented previously, provides
a mechanism for semantic selection through inheritance.
Consider, for instance the sentence pairs in (8).
(8) a. Te man / the rock fell.
b. Te man / *the rock died.
Now consider how the selectional distinction for a feature
such as animacy is modeled so as to explain the selectional
constraints of predicates. For the purpose of illustration, the
arguments of a verb will be identifed as being typed from the
system shown previously.
(9) a. x :physical[fall( x )]
b. x :animate[die( x )]
In the sentences in (8), it is clear how rocks cant die and men
can, but it is still not obvious how this judgment is computed,
given what we would assume are the types associated with
the nouns rock and man , respectively. What accomplishes this
computation is a rule of subtyping, , that allows the type asso-
ciated with the noun man (i.e., human) to also be accepted as
the type animate , which is what the predicate die requires of its
argument as stated in (9b) (cf. Gunter 1992 ; Carpenter 1992 ):
(10) [ human animate ]: human animate
Te rule applies since the concept human is subtyped under
animate in the type hierarchy. Parallel considerations rule out
the noun rock as a legitimate argument to die since it is not sub-
typed under animate. Hence, one of the concerns given for the
way that syntactic processes can systematically keep track of
which selectional features are entailed and which are not is par-
tially addressed by such lattice traversal rules as the one pre-
sented here .
2.3 Decomposition
Te second approach to the aforementioned lexical specifca-
tion is to defne constraints internally to the predicate itself.
Traditionally, this has been known as lexical decomposition .
Since the 1960s, lexical semanticists have attempted to for-
mally model the semantic relations between such lexical items
as the adjective dead and the verbs die and kill (cf. Lakof [1965]
1970; McCawley 1968 ) in the sentences that follow.
(11) a. John killed Bill.
b. Bill died.
c. Bill is dead.
Assuming that the underlying form for a verb like kill directly
encodes the stative predicate in (11c) and the relation of causation,
generative semanticists posited representations such as (12).
The Structure of Meaning
27
Although chair and rock are both physical objects , they difer
in their mode of coming into being (i.e., agentive ): Chairs are
man-made; rocks develop in nature. Similarly, a concept such
as food or cookie has a physical manifestation or denotation,
but also a functional grounding pertaining to the relation of
eating. Tese apparently contradictory aspects of a category
are orthogonally represented by the qualia structure for that
concept, which provides a coherent structuring for diferent
dimensions of meaning .
2.5 The Problem of Polysemy
Given the compactness of a lexicon relative to the number of
objects and relations in the world, and the concepts we have
for them, lexical ambiguity is inevitable. Add to this the cul-
tural, historical, and linguistic blending that contributes to the
meanings of our lexical items, and ambiguity can appear arbi-
trary as well. Hence, homonymy where one lexical form has
many meanings is to be expected in a language. Examples of
homonyms are illustrated in the following sentences:
(15) a. Mary walked along the bank of the river.
b. She works for the largest bank in the city.
(16) a. Te judge asked the defendant to approach the bar.
b. Te defendant was in the pub at the bar .
Weinreich (1964) calls such lexical distinctions contrastive
ambiguity , where it is clear that the senses associated with
the lexical item are unrelated. For this reason, it is generally
assumed that homonyms are represented as separate lexical
entries within the organization of the lexicon. Tis accords with
a view of lexical organization that has been termed a sense enu-
meration lexicon (cf. Pustejovsky 1995 ) . Words with multiple
senses are simply listed separately in the lexicon, but this does
not seem to compromise or complicate the compositional pro-
cess of how words combine in the interpretation of a sentence.
Tis model becomes dif cult to maintain, however, when
we consider the phenomenon known as polysemy . Polysemy is
the relationship that exists between diferent senses of a word
that are related in some logical manner, rather than arbitrarily,
as in the previous examples. It is illustrated in the following
sentences (cf. Apresjan 1973; Pustejovsky 1991a, 1998).
(17) a. Mary carried the book home.
b. Mary doesnt agree with the book.
(18) a. Mary has her lunch in her backpack.
b. Lunch was longer today than it was yesterday.
(19) a. Te fight lasted three hours.
b. Te fight landed on time in Los Angeles.
Notice that in each of these pairs, the same nominal form
is assuming diferent semantic interpretations relative to its
selective context. For example, in (17a), the noun book refers
to a physical object, while in (17b), it refers to the informa-
tional content. In (18a), lunch refers to the physical manifes-
tation of the food, while in (18b), it refers to the eating event.
Finally, in (19a), fight refers to the fying event, while in
(19b), it refers to the plane. Tis phenomenon of regular (or
logical) polysemy is one of the most challenging in semantics
(14) [ transition [ e 1:PROCESS ] [ e 2:STATE ] ]
Te process consists of the building activity itself, while the State
represents the result of there being the object built. Grimshaw
( 1990 ) adopts this theory in her work on argument structure,
where complex events such as break are given a similar represen-
tation. In such structures, the process consists of what x does to
cause the breaking, and the state is the resultant state of the bro-
ken item. Te process corresponds to the outer causing event as
discussed earlier, and the state corresponds in part to the inner
change of state event. Both Pustejovsky and Grimshaw difer
from earlier authors in assuming a specifc level of representa-
tion for event structure, distinct from the representation of other
lexical properties. Furthermore, they follow J. Higginbotham
( 1989 ) in adopting an explicit reference to the event place in the
verbal semantics. Recently, Levin and Rappaport (2001, 2005 )
have adopted a large component of the event structure model for
their analysis of verb meaning composition .
2.4 Noun Meaning
Tus far, we have focused on the lexical semantics of verb
entries. All of the major categories, however, are encoded with
syntactic and semantic feature structures that determine their
constructional behavior and subsequent meaning at logical
form. In Generative Lexicon Teory (Pustejovsky, 1995 ), it is
assumed that word meaning is structured on the basis of four
generative factors ( qualia roles ) that capture how humans
understand objects and relations in the world and provide the
minimal explanation for the linguistic behavior of lexical items
(these are inspired in large part by Moravcsiks ( 1975 , 1990)
interpretation of Aristotelian aitia ). Tese are: the formal
role: the basic category that distinguishes the object within
a larger domain; constitutive role: the relation between an
object and its constituent parts; the telic role: its purpose and
function; and the agentive role: factors involved in the objects
origin or coming into being.
Qualia structure is at the core of the generative properties of
the lexicon, since it provides a general strategy for creating new
types. For example, consider the properties of nouns such as
rock and chair . Tese nouns can be distinguished on the basis
of semantic criteria that classify them in terms of general cat-
egories, such as natural kind or artifact object . Although very
useful, this is not suf cient to discriminate semantic types in a
way that also accounts for their grammatical behavior. A cru-
cial distinction between rock and chair concerns the properties
that diferentiate natural kinds from artifacts : Functionality
plays a crucial role in the process of individuation of arti-
facts, but not of natural kinds. Tis is refected in grammatical
behavior, whereby a good chair or enjoy the chair are well-
formed expressions refecting the specifc purpose for which
an artifact is designed, but good rock or enjoy a rock are
semantically ill-formed since for rock the functionality (i.e.,
telic ) is undefned. Exceptions exist when new concepts are
referred to, such as when the object is construed relative to a
specifc activity, for example, as in Te climber enjoyed that
rock; rock itself takes on a new meaning, by virtue of having
telicity associated with it, and this is accomplished by inte-
gration with the semantics of the subject noun phrase (NP).
The Cambridge Encyclopedia of the Language Sciences
28
language of types , we can express the rule of APPLY as a prop-
erty associated with predicates (or functions), and application
as a relationship between expressions of specifc types in the
language.
(24) Function Application:
If is of type a , and is of type a b , then () is of type b .
Viewed as typed expressions, the separate linguistic units
in (23a) combine as function application, as illustrated in
Figure 2 .
As one can see, the -calculus is an expressive mechanism
for modeling the relation between verbs and their arguments
interpreted as function application.
One important extension to the type language used here
provides a compositional analysis of the semantics of proposi-
tional attitude verbs, such as believe and think (Montague 1973 ).
Te sentential complements of such verbs, as is well known,
create opaque contexts for substitutions under identity. For
example, if Lois is unaware of Supermans true identity, then
the belief statement in (25b) is false, even though (25a) is true.
(25) a. Lois believes Superman rescued the people.
b. Lois believes Clark Kent rescued the people .
On this view, verbs such as believe introduce an inten-
sional context for the propositional argument, instead of an
extensional one. In such a context, substitution under identity is
not permitted without possibly afecting the truth value ( truth
conditional semantics ). Tis is an important contribution
to the theory of meaning, in that a property of opacity is associ-
ated with specifc types within a compositional framework .
One potential challenge to a theory of function application
is the problem of ambiguity in language. Syntactic ambiguities
arise because of the ways in which phrases are bracketed in a
sentence, while lexical ambiguity arises when a word has mul-
tiple interpretations in a given context. For example, in the fol-
lowing sentence, the verb treat can mean one of two things:
(26) Te doctor treated the patient well.
Either 1) the patient is undergoing medical care, or 2) the doc-
tor was kind to the patient. More often than not, however, the
context of a sentence will eliminate such ambiguities, as shown
in (27).
(27) a. Te doctor treated the patient with antibiotics. (Sense 1)
b. Te doctor treated the patient with care. (Sense 2)
In this case, the interpretation is constructed from the appro-
priate meaning of the verb and how it combines with its
arguments .
and has stimulated much research recently (Bouillon 1997 ;
Bouillon and Busa 2001 ; Cooper 2006). Te determination
of what such lexical items denote will of course have conse-
quences for ones theory of compositionality, as we will see in
a later section .
3 BUILDING SENTENCE MEANINGS
3.1 Function Application
Te principle of compositionality follows the view that syntax
is an initial guide to the interpretation process. Hence, there
would appear to be a strong relationship between the meaning
of a phrase and where it appears in a sentence, as is apparent
from grammatical function in the following sentences.
(20) a. Te woman loves the child.
b. Te child loves the woman.
However, this is not always a reliable association, as seen
in languages that have freer word order restrictions, such as
German.
(21) a. Die Frau liebt das Kind.
Te woman loves the child.
b. Das Kind liebt die Frau.
Te child loves the woman.
In German, both word orders are ambiguous, since infor-
mation about the grammatical case and gender of the two NPs
is neutralized .
Although there is often a correlation between the gram-
matical relation associated with a phrase and the meaning
assigned to it, this is not always a reliable association. Subjects
are not always doers and objects are not always undergoers
in a sentence. For example, notice how in both (22a) and (22b),
the NP the watch is playing the same role; that is, it is undergo-
ing a change, even though it is the subject in one sentence and
the object in the other.
(22) a. Te boy broke the watch .
b. Te watch broke.
To handle such verbal alternations compositionally requires
either positing separate lexical entries for each syntactic con-
struction associated with a given verb, or expressing a deeper
relation between diferent verb forms .
For most semantic theories, the basic mechanism of com-
positionality is assumed to be function application of some
sort. A rule of application , apply , acts as the glue to assign (or
discharge) the argument role or position to the appropriate
candidate phrase in the syntax. Tus, for a simple transitive
sentence such as (23a), two applications derive the proposi-
tional interpretation of the sentence in (23d).
(23) a. John loves Mary.
b. love(Arg1,Arg2)
c. APPLY love(Arg1,Arg2) to Mary = love(Arg1,Mary)
d. APPLY love(Arg1,Mary) to John = love(John,Mary)
One model used to defne the calculus of compositional
combinations is the -calculus (Barendregt 1984 ). Using the
Figure 2.
The Structure of Meaning
29
(32) a. Every woman sang a song.
b. xy [ woman ( x ) [ song ( y ) & sang ( x, y )]]
c. yx [[ song ( y ) & woman ( x )] sang ( x, y )]]
An alternative treatment for handling such cases is to posit a
rule of quantifer raising, where the scope ambiguity is reduced
to a diference in syntactic structures associated with each
interpretation (May 1985 ) .
3.3 Semantic Modifcation
In constructing the meaning of expressions, a semantic the-
ory must also account for how the attribution of properties
to an entity is computed, what is known as the problem of
modifcation. Te simplest type of modifcation one can imag-
ine is intersective attribution . Notice that in the phrases in (33),
the object denoted correctly has both properties expressed in
the NP:
(33) a. black cofee x [ black ( x ) & cofee ( x )]
b. Italian singer x [ Italian ( x ) & singer ( x )]
c. metal cup x [ metal ( x ) & cup ( x )]
Tere are two general solutions to computing the meaning
of such expressions: a) Let adjectives be functions over com-
mon noun denotations, or b) let adjectives be normal predi-
cates, and have a semantic rule associated with the syntax of
modifcation.
Computing the proper inferences for relative clauses will
involve a similar strategy, since they are a sort of intersective
modifcation. Tat is, for the relative clause in (34), the desired
logical form will include an intersection of the head noun and
the relation predicated in the subordinated clause.
(34) a. writer who John knows
b. x [ writer ( x ) & know ( j, x )]
Unfortunately, however, most instances of adjectival modi-
fcation do not work so straightforwardly, as illustrated in (35).
Adjectives such as good , dangerous , and fast modify polyse-
mously in the following sentences.
(35) a. John is a a good teacher .
b. A good meal is what we need now.
c. Mary took a good umbrella with her into the rain.
In each of these sentences, good is a manner modifer whose
interpretation is dependent on the noun it modifes; in (35a), it
means to teach well; in (35b), it means a tasty meal; and in
(35c), it means something keeping you dry. Similar remarks
hold for the adjective dangerous .
(36) a. Tis is a dangerous road at night.
b. She used a dangerous knife for the turkey.
Tat is, the road is dangerous in (36a) when one drives on it,
and the knife is dangerous in (36b) when one cuts with it.
Finally, the adjective fast in the following sentences acts as
though it is an adverb, modifying an activity implicit in the
noun, that is, programming in (37a) and driving in (37b).
(37) a. Mary is the fastest programmer we have on staf.
b. Te turnpike is a faster road than Main Street.
3.2 Quantifers and Scope
Another type of ambiguity, one that is not associated with
the constituent structure of the sentence or lexical senses in
any obvious way, involves quantifed noun phrases (e.g., every
cookie , some cake , and most pies ). It is interesting that when a
sentence has more than one of these phrases, one often sees
more than one interpretation possible because of the ways the
quantifed NPs relate to each other. Tis is not the case in the
following sentence, however, where there is only one interpre-
tation as to what happened with the cookie.
(28) Some student ate a cookie.
Now consider the sentences in (29), where there is a combi-
nation of a some -NP and an every -NP.
(29) a. Every student saw a movie.
b. Every cookie was eaten by a student.
Te sentence in (29a) can mean one of two things: 1) that
there was one movie, for example, Star Wars , that every stu-
dent saw; or 2) that everyone saw a movie, but it didnt have to
be the same one. Similarly, for (29b), there could be one stu-
dent who ate all the cookies, or each cookie that was eaten
by a diferent student. Tis kind of quantifer scope ambiguity
has to be resolved in order to determine what kind of infer-
ences one can make from a sentence. Syntax and semantics
must interact to resolve this kind of ambiguity, and it is the
theory of sentence meaning that defnes this interaction (cf.
quantification ).
One of the roles of semantic theory is to correctly derive the
entailment relations associated with a sentences logical form,
since this has an obvious impact on the valid reasoning pat-
terns in the language. How these interpretations are computed
has been an area of intense research, and one of the most infu-
ential approaches has been the theory of generalized quan-
tifers (cf. Barwise and Cooper 1981). On this approach, the
denotation of an NP is treated as a set of sets of individuals , and
a sentence structure such as [NP VP] is true if and only if the
denotation of the VP is a member of the family of sets denoted
by the NP. Tat is, the sentence in (30) is true if and only if sing-
ing (the denotation of the VP) is a member of the set of proper-
ties denoted by every woman .
(30) Every woman sang.
On this view, quantifers such as most , every , some , and so
on are actually second-order relations between predicates,
and it is partly this property that allows for the compositional
interpretation of quantifer scope variation seen previously.
Te intended interpretation of (30) is (31b), where the subject
NP every woman is interpreted as a function, taking the VP as
its argument.
(31) a. P x [ woman ( x ) P ( x )]( sang )
b. x [ woman ( x ) sang ( x )]
When combined with another quantifed expression, as in
(32a), the relational interpretation of the generalized quantif-
ers is crucial for being able to determine both scope interpreta-
tions shown in (32).
The Cambridge Encyclopedia of the Language Sciences
30
Each of the italicized phrases is an argument of something,
but is it selected by the matrix predicate? Jackendof has pro-
posed a solution that relies on the notion of construction,
as introduced by A. E. Goldberg ( 1995 ) (cf. construction
grammars ).
Another problem in compositionality emerges from the
interpretation of adjuncts. Te question posed by the exam-
ples in (41) is this: Which NPs are arguments semantically and
which are merely adjuncts?
(41) a. Mary ate the soup.
b. Mary ate the soup with a spoon.
c. Mary ate the soup with a spoon in the kitchen.
d. Mary ate the soup with a spoon in the kitchen at 3:00 p.m .
For Davidson ( 1967 ), there is no semantic distinction between
arguments and adjuncts in the logical form. Under his pro-
posal, a two-place predicate such as eat contains an additional
argument, the event variable, e , which allows each event par-
ticipant a specifc role in the interpretation (cf. Parsons 1990 ;
event structure and grammar ).
(42) yxe [eat( e, x, y )]
Ten, any additional adjunct information (such as locations,
instruments, etc.) is added by conjunction to the meaning
of the main predicate, in a fashion similar to the interpreta-
tion of intersective modifcation over a noun. In this man-
ner, Davidson is able to capture the appropriate entailments
between propositions involving action and event expressions
through conventional mechanisms of logical entailment. For
example, to capture the entailments between (41bd) and (41a)
in the following, each more specifcally described event entails
the one above it by virtue of conjunction elimination (already
encountered) on the expression.
(43) a. e [eat( e, m, the-soup)]
b. e [eat( e, m, the-soup) & with ( e, a spoon)]
c. e [eat( e, m, the-soup) & with ( e, a spoon) & in ( e, the
kitchen)]
d. e [eat( e, m, the-soup) & with ( e, a spoon) & in ( e, the
kitchen) & at ( e, 3:00 p.m. )]
Tis approach has the advantage that no special inference
mechanisms are needed to derive the entailment relations
between the core propositional content in (43a) and forms
modifed through adjunction. Tis solution, however, does not
extend to cases of verbs with argument alternations that result
in diferent meanings. For example, how do we determine what
the core arguments are for a verb like sweep ?
(44) a. John swept .
b. John swept the foor.
c. John swept the dirt.
d. John swept the dirt of the sidewalk.
e. John swept the foor clean.
f. John swept the dirt into a pile.
Te semantics of such a verb should determine what its argu-
ments are, and how the diferent possible syntactic realizations
relate to each other semantically. Tese cases pose an interest-
To account for such cases, it is necessary to enrich the mode of
composition beyond simple property intersection, to accom-
modate the context dependency of the interpretation. Analyses
taking this approach include Borschev and Partee ( 2001 ),
Bouillon ( 1997 ), and Pustejovsky ( 1995 ) .
3.4 Arguments versus Adjuncts
In our discussion thus far of how predicates select arguments to
create compositionally complex expressions, we have assumed
that the matrix predicate (the main verb of the sentence) acts
as the only function over other phrases. In fact, what an argu-
ment of the verb is and what an adjunct is are questions just as
much of meaning as of syntax. In this section, we examine the
semantic issues involved.
In this overview, we have adopted the position that lan-
guage refects the workings of our deeper conceptual sys-
tems in some direct and nonidiosyncratic manner. Lexical
choice as well as specifc grammatical phenomena can be
constrained by underlying conceptual bias. Well-known
examples of this transparency include count/mass noun
distinctions in the lexicon, and case marking and valence
distinctions in the syntax. For example, concepts entail-
ing unindividuated stuf or material will systematically be
semantically typed as mass nouns in the grammar, whereas
naturally individuating (countable) substances will assume
the status of count nouns, with their respective grammatical
consequences, as illustrated in (38). (Some mass terms are
not shared by all languages, such as the concept of paper
or furniture.)
(38) a. { not much/all/lots of } gold/water/dirt/sand
b. { every/two/several } chairs/girls/beaches
Similarly, as presented in previous sections, the classif-
cation of verbs appears to refect their underlying relational
structure in fairly obvious ways.
(39) a. Mary arrived .
b. John greeted Mary.
c. Mary gave a book to John.
Tat is, the argument structure of each verb encodes the
semantics of the underlying concept, which in turn is refected
in the projection to the specifc syntactic constructions, that
is, as intransitive, transitive, and ditransitive constructions,
respectively. For unary, binary, and ternary predicates, there
is a visible or transparent projection to syntax from the under-
lying conceptual structure, as well as a predictable composi-
tional derivation as function application.
So, the question arises as to what we do with nonselected
arguments and adjuncts within the sentence. It is well known,
for example, that arguments not selected by the predicate
appear in certain contexts (cf. Jackendof 1992 ; Levin and
Rappaport Hovav 2005 ).
(40) a. Te man laughed himself sick.
b. Te girl danced her way to fame.
c. Mary nailed the window shut .
The Structure of Meaning
31
be predicted from the meaning of its parts. We have already
encountered modifcation constructions that do not conform to
simple intersective interpretations, for example, good teacher .
Tere are two other constructions that pose a problem for the
principle of compostionality in semantics:
(51) a. Idioms: hear it through the grapevine , kick the bucket ;
b. Coercions: begin the book , enjoy a cofee .
Te meaning of an idiom such as leave well enough alone is
in no transparent way composed of the meanings of its parts.
Although there are many interesting syntactic properties and
constraints on the use of idiomatic expressions in languages,
from a semantic point of view its meaning is clearly associated
with the entire phrase. Hence, the logical form for (52),
(52) Every person kicked the bucket.
will make reference to quantifcation over persons, but not
over buckets (cf. [53]).
(53) x [ person ( x ) & kick.the.bucket ( x )]
We confront another kind of noncompositionality in
semantics when predicates seem to appear with arguments of
the wrong type. For example, in (54a), a countable individ-
ual entity is being coerced into the food associated with that
animal, namely, bits of chicken, while in (54b), the mass terms
water and beer are being packaged into unit measures (Pelletier,
1975 ). In (55), the aspectual verbs normally select for an event,
but here are coercing entities into event denotations. Similarly,
in (56), both object NPs are being coerced into propositional
interpretations. (Cf. Pustejovsky 1995 and Jackendof 2002 for
discussions of coercion phenomena and their treatment.)
(54) a. Teres chicken in the salad.
b. Well have a water and two beers.
(55) a. Roser fnished her thesis.
b. Mary began the novel.
(56) a. Mary believes Johns story.
b. Mary believes John.
Tese examples illustrate that semantics must accommodate
specifc type-shifting and coercing operations in the language
in order to remain compositional. In order to explain just such
cases, Pustejovsky ( 2007 ) presents a general theory of compo-
sition that distinguishes between four distinct modes of argu-
ment selection: a) function application, b) accommodation,
c) coercion by introduction, and d) coercion by exploitation .
4 DISCOURSE STRUCTURE
Tus far we have been concentrating on the meaning of single
sentences. But no sentence is really ever uttered outside of a
context. Language is used as a means of communication and is
as much a way of acting as a means of representing (Austin 1975 ;
Searle 1969 ). In this section, we briefy survey the major areas
of research in discourse semantics. We begin by examin-
ing the semantic models that have emerged to account for
dynamic phenomena in discourse, such as intersentential
ing challenge for the theory of compositionality (cf. Jackendof
2002 ) .
3.5 Presupposition
In computing the meaning of a sentence, we have focused
on that semantic content that is asserted by the proposition.
Tis is in contrast to what is presupposed . A presupposition
is that propositional meaning that must be true for the sen-
tence containing it to have a proper semantic value (Stalnaker
1970 ; Karttunen 1974 ; Potts 2005 ). (Stalnaker makes the dis-
tinction between what a speaker says and what a speaker
presupposes.)
Such knowledge can be associated with a word, a grammat-
ical feature, or a syntactic construction (so-called presuppo-
sition triggers ). For example, in (45) and (46), the complement
proposition to each verb is assumed to be true, regardless of the
polarity assigned to the matrix predicate.
(45) a. Mary realized that she was lost.
b. Mary didnt realize that she was lost.
(46) a. John knows that Mary is sick.
b. John doesnt know that Mary is sick.
Tere are similar presuppositions associated with aspectual
predicates, such as stop and fnish , as seen in (47).
(47) a. Fred stopped smoking.
b. John fnished painting his house.
In these constructions, the complement proposition is assumed
to have been true before the assertion of the sentence.
Such conventional presuppositions are also triggered by
interrogative contexts, such as seen in (48).
(48) a. Why did you go the store?
b. When did you see Mary?
As with all presuppositions, however, they are defeasible, as
the answer to (48b) in (49) illustrates.
(49) But I didnt see Mary.
Conversational presuppositions, on the other hand, are impli-
cated propositions by virtue of a context and discourse situation.
Te response in (50b) conversationally implicates that I am not
hungry (Recanati 2002 ); conversational implicature ).
(50) a. Are you hungry?
b. Ive had a very large breakfast.
Te meaning of such implicatures is not part of the asserted
content of the proposition, but computed within a conversa-
tional context in a discourse. We will return to this topic in a
later section .
3.6 Noncompositionality
While semantic theory seems to conform to the principles of
compositionality in most cases, there are many constructions
that do not ft into the conventional function application par-
adigm. A phrase is noncompositional if its meaning cannot
The Cambridge Encyclopedia of the Language Sciences
32
is added to the listeners interpretation state so that the lis-
tener can use the quantifer to help understand future utter-
ances. In this way, the meaning of a sentence is interpreted
dynamically .
Te dynamics of discourse, of course, involve more than the
binding of anaphors to antecedents across adjacent sentences.
Every utterance is made in the context of a common ground
of shared knowledge (presuppositions), with a communica-
tive intent, and in a particular time and place (cf. discourse
analysis , communicative intention ). Just as sentences
have internal structure, with both syntactic and semantic
dependencies, discourse can also be viewed as a sequence of
structured segments, with named dependencies between them .
For example, the sentences in (62) form a discourse structured
by a relation of narration , implying temporal sequence (Dowty,
1986 ).
(62) a. John entered the room.
b. He sat down.
In (63), on the other hand, the two sentences are related by the
dependency of explanation , where (63b) temporally precedes
and explains (63a).
(63) a. Max fell.
b. John pushed him.
Teories of discourse relations, such as rhetorical struc-
ture theory (Mann and Tompson 1986 ), segmented dis-
course representation theory (SDRT) (Asher and Lascarides 3 ),
and that of Hobbs ( 1985 ) attempt to model the rhetorical func-
tions of the utterances in the discourse (hence, they are more
expressive of discourse structure and speaker intent than dis-
course representation theory [DRT], which does not model
such parameters). For the simple discourses above, SDRT, for
example extends the approach from dynamic semantics with
rhetorical relations and their semantic values, while provid-
ing a more complex process of discourse updates. Rhetorical
relations, as used in SDRT, carry specifc types of illocution-
ary force (cf. Austin 1975 ; Searle 1969 , 9 ), namely, explanation,
elaboration, giving backgrounds, and describing results .
5 CONCLUSION
In this essay, I have attempted to outline the basic components
for a theory of linguistic meaning. Many areas of semantics were
not touched on in this overview, such as issues relating to the
philosophy of language and mind and the psychological conse-
quences of various semantic positions. Many of the accompa-
nying entries herein, however, address these issues directly .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Apresjan , J. D. 1973 . Synonymy and synonyms. In Trends in Soviet
Teoretical Linguistics , ed. F. Kiefer , 17399. Dordrecht, the
Netherlands: Reidel.
Asher , Nicholas , and Alex Lascarides . 2003 . Logics of Conversation .
Cambridge : Cambridge University Press .
Austin , J. L. 1975 . How to Do Tings with Words . Cambridge : Harvard
University Press .
anaphora. We then look at how discourse relations can be used
to model larger units of meaning.
From our previous discussion, we have assumed the sen-
tence as the unit for semantic interpretation, including the
level for the interpretation of quantifer scope and anaphoric
binding, as in (57).
(57) a. Every actress said she was happy.
b. Every actress came in and said hello.
Notice that the anaphoric link between the quantifer and the
pronoun in (57a) is acceptable, while such a binding is not pos-
sible within a larger discourse setting, as in (58) and (59).
(58) a. Every actress came in.
b. *She said she was happy.
(59) a. Every actress came in.
b. *She said hello.
So, in a larger unit of semantic analysis, a bound variable inter-
pretation of the pronoun does not seem permitted.
Now notice that indefnites do in fact allow binding across
the level of the sentence.
(60) a. An actress came in.
b. She said hello.
Te desired interpretation, however, is one that the semantic
model we have sketched out is unable to provide.
(61) a. x [ actress ( x ) & come.in ( x )]
b. [ & say.hello ( x )]
What this example points out is that the view of meaning we
have been working with so far is too static to account for phe-
nomena that are inherently dynamic in nature (Chierchia 1995 ;
Groenendijk and Stokhof 1991 ; Karttunen 1976 ). In this exam-
ple, the indefnite NP an actress is being used as a discourse
referent, and is available for subsequent reference as the story
unfolds in the discourse.
Following Kamp and Reyles ( 1993 ) view, an indefnite NP
introduces a novel discourse referent, while a pronoun or def-
inite description says something about an existing discourse
referent. Using the two notions of novelty and familiarity , we
can explain why she in (60b) is able to bind to the indefnite;
namely, she looks for an accessible discourse referent, the
indefnite. Te reason that (58) and (59) are not good discourses
is due to the universally quantifed NP every actress, which is
inaccessible as an antecedent to the pronoun.
One infuential formalization of this approach is Dynamic
Predicate Logic (Groenendijk and Stokhof, 1991 ), which com-
bines conventional interpretations of indefnites as existentials
with the insight from incremental interpretations, mentioned
previously. On this view, the interpretation of a sentence is a
function of an ordered pair of assignments, rather than a static
single assignment. Te output condition for a sentence with
an indefnite NP, such as (60a), specifes that a subsequent sen-
tence with a pronoun can share that variable assignment: Te
meaning of a sentence lies in the way it changes the representa-
tion of the information of the interpreter (ibid.). Tat is, when
a quantifed expression is used in discourse, something new
The Structure of Meaning
33
Dowty , D. R . 1979 . Word Meaning and Montague Grammar . Dordrecht,
the Netherlands : D. Reidel .
. 1986 . Te efects of aspectual class on the temporal structure
of discourse: Semantics or pragmatics . Linguistics and Philosophy
9.1 : 37 61.
. 1991 . Tematic proto-roles and argument selection . Language
67 : 547 619.
Evans , R. , and G. Gazdar. 1990 . Te DATR papers: February 1990.
Cognitive Science Research Paper CSRP 139, School of Cognitive
and Computing Science, University of Sussex, Brighton, England.
Fillmore , C . 1968 . Te case for case. In Universals in Linguistic
Teory , ed. E. Bach and R. Harms , 188. New York : Holt, Rinehart,
and Winston .
Gazdar , G. , E. Klein , G. Pullum , and I. Sag. 1985 . Generalized Phrase
Structure Grammar . Cambridge : Harvard University Press .
Goldberg , A. E. 1995 . Constructions: A Construction Grammar Approach
to Argument Structure . Chicago : University of Chicago Press .
Grice , H. P. 1969 . Utterers meaning and intentions. Philosophical
Review 78 : 147 77.
.1989 . Studies in the Way of Words . Cambridge : Harvard University
Press .
Grimshaw , J. 1979 . Complement selection and the lexicon Linguistic
Inquiry 10 : 279 326.
. 1990 . Argument Structure . Cambridge, MA : MIT Press .
Groenendijk , Jeroen , and Martin Stokhof . 1991 . Dynamic predicate
logic . Linguistics and Philosophy 14 .1: 39100 .
Gruber , J. S. 1976 . Lexical Structures in Syntax and Semantics .
Amsterdam : North-Holland .
Gunter , C. 1992 . Semantics of Programming Languages . Cambridge,
MA : MIT Press .
Hale , K. and S. J. Keyser. 1993 . On argument structure and the lex-
ical expression of syntactic relations. In Te View from Building
20: Essays in Honor of Sylvain Bromberger , ed. K. Hale and S. J. Keyser,
53109. Cambridge, MA : MIT Press .
Heim , Irene. 1982 . Te semantics of defnite and indefnite noun phrases .
Ph.D. thesis, University of Massachussets , Amherst .
Higginbotham , J. 1989 . Elucidations of meaning . Linguistics and
Philosophy 12 : 465 517.
Hobbs , Jerry R. 1985 . On the coherence and structure of discourse.
Report No. CSLI-8537, Center for the Study of Language and
Information, Stanford University.
Jackendof , R. 1972 . Semantic Interpretation in Generative Grammar .
Cambridge, MA : MIT Press .
. 1983 . Semantics and Cognition . Cambridge, MA : MIT Press .
. 1990 . Semantic Structures . Cambridge, MA: MIT Press .
. 1992 . Babe Ruth homered his way into the hearts of America.
In Syntax and the Lexicon , ed. T. Stowell and E. Wehrli , 15578. San
Diego: Academic Press.
. 1997 . Te Architecture of the Language Faculty . Cambridge, MA:
MIT Press .
. 2002 . Foundations of Language . Oxford : Oxford University
Press .
Kamp , H. 1981 . A theory of truth and semantic representation. In
Formal Methods in the Study of Language , ed. J. A. G. Groenendijk ,
T. M. V. Janssen , and M. B. J. Stokhof . Mathematical Centre Tracts
135, 277322. Amsterdam: Mathematical Centre.
Kamp , H. , and U. Reyle. 1993 . From Discourse to Logic . Dordrecht, the
Netherlands: Kluwer Academic Publishers .
Karttunen , L. 1974 . Presupposition and linguistic context. Teoretical
Linguistics 1 : 181 93.
. 1976 . Discourse referents. In Syntax and Semantics , ed. James
D. McCawley . Vol. 7: Notes from the Linguistic Underground , 36385.
New York : Academic Press .
Barendregt , Henk. 1984 . Te Lambda Calculus, Its Syntax and
Semantics . Amsterdam : North-Holland .
Barwise , Jon , and Robin Cooper . 1981 . Generalized quantifers and
natural language . Linguistics and Philosophy 4 .1: 159 219.
Borschev , Vladimir , and Barbara H. Partee . 2001 . Genitive modifers,
sorts, and metonymy. Nordic Journal of Linguistics 24 .2: 140 60.
Bouillon , P. 1997 . Polymorphie et semantique lexicale: Le cas des adjec-
tifs . Lille : Presses Universitaires du Spetentrion .
Bouillon , P. , and F. Busa , eds. 2001 . Te Language of Word Meaning .
Cambridge : Cambridge University Press .
Bresnan , J. , ed. 1982 . Te Mental Representation of Grammatical
Relations . Cambridge, MA : MIT Press .
Briscoe , T. , V. de Paiva , and A. Copestake , eds. 1993 . Inheritance,
Defaults, and the Lexicon . Cambridge : Cambridge University Press .
Busa , F. 1996 . Compositionality and the semantics of nominals . Ph.D.
diss., Brandeis University.
Carlson , G. 1984 . Tematic roles and their role in semantic interpreta-
tion . Linguistics 22 : 259 79.
Carpenter , B. 1992 . Te Logic of Typed Feature Structures .
Cambridge : Cambridge University Press .
Carter , R. [1976] 1988 . On Linking: Papers by Richard Carter . Ed. B.
Levin and C. Tenny . MIT Lexicon Project Working Papers 25, Center
for Cognitive Science. Cambridge, MA : MIT Press .
Chierchia , G. 1989 . Structured meanings, thematic roles, and con-
trol. In Properties, Types, and Meaning , ed. G. Chierchia , B. Partee ,
and R. Turner , II: 13166. Dordrecht, the Netherlands: Kluwer
Academic Publishers.
. 1995 . Te Dynamics of Meaning . Chicago : University of Chicago
Press .
Chomsky , N. [1955] 1975 . Te Logical Structure of Linguistic Teory .
Chicago : University of Chicago Press .
. 1965 . Aspects of the Teory of Syntax . Cambridge, MA : MIT
Press .
Comrie , Bernard. 1981 . Language Universals and Linguistic
Typology: Syntax and Morphology . Chicago : University of Chicago
Press .
Cooper , Robin . 2006 . A record type theoretic account of copredica-
tion and dynamic generalized quantifcation. In Kvantifkator
for en Dag, Essays Dedicated to Dag Westerstahl on His Sixtieth
Birthday . Avilable online at: http://www.phil.gu.se/posters/
festskrift3/.
Copestake , A. 1992 . Te Representation of Lexical Semantic Information .
Cognitive Research Paper CSRP 280, School of Cognitive and
Computing Science, University of Sussex , Brighton, England .
. 1993 . Defaults in Lexical Representation. In Inheritance,
Defaults, and the Lexicon , ed. Ted Briscoe , Valeria de Paiva , and Ann
Copestake , 22345. Cambridge : Cambridge University Press .
Copestake , A. , and E. Briscoe. 1992. Lexical operations in a unif-
cation-based framework. In Lexical Semantics and Knowledge
Representation , ed. J. Pustejovsky and S. Bergler , 10119. New
York : Springer Verlag .
Copestake , A. , and T. Briscoe . 1995 . Semi-productive polysemy and
sense extension . Journal of Semantics 12 : 15 67.
Davidson , D. 1967 . Te logical form of action sentences. In Te Logic
of Decision and Action , ed. N. Rescher , 8195. Pittsburgh : Pittsburgh
University Press .
Davidson , D. , and G. Harman , eds. 1972 . Semantics of Natural
Language . Dordrecht, the Netherlands : Reidel .
Davis , Steven , and Brendan Gillon. 2004 . Semantics: A Reader .
Oxford : Oxford University Press .
Dolling , J. 1992 . Flexible Interpretationen durch Sortenverschiebung.
In Fgungspotenzen , ed. Ilse Zimmermann and Anatoli Strigen ,
2362. Berlin: Akademie Verlag .
The Cambridge Encyclopedia of the Language Sciences
34
Partee , Barbara , ed. 1976 . Montague Grammar . New York : Academic
Press .
Partee , B. , and M. Rooth. 1983 . Generalized conjunction and type
ambiguity. In Meaning, Use, and Interpretation of Language, ed.
Rainer Buerle , Christoph Schwarze , and Arnim von Stechow ,
36183. Berlin : Walter de Gruyter .
Pelletier , F. J. 1975 . Non-singular reference: Some Preliminaries. In
Mass Terms: Some Philosophical Problems , ed. F. J. Pelletier , 114.
Dordrecht, the Netherlands : Reidel .
Pollard , C. , and I. Sag. 1994 . Head-Driven Phrase Structure Grammar .
Chicago : University of Chicago Press and Stanford CSLI .
Potts , Christopher. 2005 . Te Logic of Conventional Implicatures .
Oxford : Oxford University Press .
Pustejovsky , J. 1991a . Te generative lexicon . Computational
Linguistics 17 : 409 41.
. 1991 b. Te syntax of event structure . Cognition 41 : 47 81.
. 1995 . Te Generative Lexicon . Cambridge, MA: MIT Press .
. 1998 . Te semantics of lexical underspecifcation . Folia
Linguistica 32 : 323 47.
. 2007 . Te Multiplicity of Meaning . Cambridge, MA : MIT Press .
Pustejovsky , J. , and B. Boguraev. 1993 . Lexical knowledge repre-
sentation and natural language processing. Artifcial Intelligence
63 : 193 223.
Recanati , Francois. 2002 . Unarticulated constituents . Linguistics
and Philosophy 25 : 299 345.
Russell , Bertrand , 1905 . On Denoting . Mind 14 : 479 93.
Saussure , Ferdinand de. [1916] 1983 . Course in General Linguistics .
Trans. R. Harris . London : Duckworth .
Searle , John . 1969 . Speech Acts: An Essay in the Philosophy of Language .
Cambridge : Cambridge University Press .
. 1979 . Expression and Meaning . Cambridge : Cambridge
University Press .
Stalnaker , Robert. 1970 . Pragmatics . Synthese 22.1/2 : 272 78.
Strawson , P. F. 1971 . Logico-Linguistic Papers . London : Methuen .
Tenny , C. 1992 . Te Aspectual Interface Hypothesis. In Lexical
Matters , CSLI Lecture Notes , ed. I. Sag and A. Szabolcsi , 127.
Chicago: University of Chicago Press .
Tenny , C. , and J. Pustejovsky. 2000 . Events as Grammatical Objects.
Stanford, CA: CSLI Publications. Chicago: University of Chicago
Press.
Weinreich , U. 1964 . Websters Tird: A Critique of its Semantics .
International Journal of American Linguistics 30 : 405 9.
. 1972 . Explorations in Semantic Teory . Te Hague : Mouton.
Williams , E. 1981 . Argument structure and morphology . Linguistic
Review 1 : 81 114.
Katz , J. , and J. Fodor. 1963 . Te structure of a semantic theory .
Language 39 : 170 210.
Krifka , M. 1992 . Tematic relations as links between nominal reference
and temporal constitution. In Lexical Matters , CSLI Lecture Notes, ed.
I. Sag and A. Szabolcsi , 2953. Chicago: University of Chicago Press .
Kripke , Saul . [1972] 1980 . Naming and necessity. In Semantics
of Natural Language , ed. D. Davidson and G. Harman , 253355.
Dordrecht and Boston : Reidel .
Lakof , G. [1965] 1970 . Irregularity in Syntax . New York : Holt, Rinehart,
and Winston .
Levin , B. , and M. Rappaport Hovav. 1995 . Unaccusativity: At the Syntax-
Semantics Interface . Cambridge, MA : MIT Press .
. ( 2005 . Argument Realization . Cambridge : Cambridge University
Press .
Lewis , David . 1976 . General semantics. In Montague Grammar , ed.
Barbara H. Partee , 150. New York : Academic Press .
Lyons , John . 1968 . Introduction to Teoretical Linguistics .
Cambridge : Cambridge University Press .
Mann , William C. , and Sandra A. Tompson . 1986 . Rhetorical
Structure Teory: Description and Construction of Text Structures .
ISI/RS-86174, 115. Nijmegen, the Netherlands : Information
Sciences Institute .
May , Robert. 1985 . Logical Form: Its Structure and Derivation .
Cambridge, MA : MIT Press .
McCawley , J. D . 1968 . Te role of semantics in a grammar. In
Universals in Linguistic Teory , ed. E. Bach and R. T. Harms , 12469.
New York : Holt, Rinehart, and Winston .
Moens , M. , and M. Steedman . 1988 . Temporal ontology and temporal
reference . Computational Linguistics 14 : 15 28.
Montague , Richard. 1973 . Te proper treatment of quantifcation in ordi-
nary English. In Approaches to Natural Language , ed. Jaakko Hintikka ,
Julius Matthew , Emil Moravcisk , and Patrick Suppes , 22142. Dordrecht,
the Netherlands : D. Reidel . Repr. in Montague 1974, 24770.
. 1974 . Formal Philosophy: Selected Papers of Richard Montague .
New Haven, CT : Yale University Press .
Moravcsik , J. M. 1975 . Aitia as generative factor in Aristotles philoso-
phy. Dialogue 14 : 622 36.
. 1990 . Tought and Language . London : Routledge .
Nunberg , G. 1979 . Te non-uniqueness of semantic solu-
tions: Polysemy. Linguistics and Philosophy 3 : 143 84.
Parsons , T. 1990 . Events in the Semantics of English . Cambridge, MA:
MIT Press .
Partee , Barbara. 1984 . Compositionality. In Varieties of Formal
Semantics , ed. Fred Landman and Frank Veltman , 281311.
Dordrecht, the Netherlands : Foris .
35
diglossia , culture and language , digital media , lit-
eracy ). Tis essay is focused on the issue of the media of com-
munication and the social conditions and consequences of their
evolution. Te reason is that the social practices of speech and
writing both depend on the available technology and lead to tech-
nological and social innovation. As has been argued by Marshall
McLuhan ( 1964 ), Elizabeth L. Eisenstein ( 1979 ), Jan Assmann
( 1991 ), David R. Olson ( 1994 ), and Nicholas Negroponte ( 1995 )
among others, civilizations are characterized by the media they
predominantly use and which shape the way they exchange,
store, and administer information, thus exercising a profound
infuence on social practice.
Te nexus between speech and writing is variable and more
obvious in some cases than in others. For instance, when the
lyrics of a song are read on the monitor and sung in a karaoke
bar, speech and writing are joined together in one activity. On
the other hand, the songs of bards handed down by word of
mouth from one generation to another are part of the culture of
spoken, as opposed to written, language (Ong 1982 , Olson 1991 ;
oral culture , oral composition ) . However, the very idea
of orality is predicated on literacy and would not have become
an object of study without it. Just as there is no silence without
sound, illiteracy exists but in a literate society . On the face of it,
many kinds of verbal behavior, such as speech contests, bidding
at an auction, and election campaign addresses, do not involve
writing. Te institutional frameworks in which they take place in
modern society, school, trade and government, though, rely to a
very large extent on written texts. To analyze social practices of
speech and writing, then, it is necessary to consider technologi-
cal aspects of writing and institutional aspects of literacy.
TECHNOLOGICAL ASPECTS OF WRITING
Many social practices and ligatures of contemporary society
would be impossible without writing. Tis does not imply that the
externalization of language by technological means is the only
force that shaped modern society. Te assumption of an unme-
diated cause-and-efect relationship between writing and social
organization, of a watershed between primitive oral life and com-
plex literate civilization, is a simplifcation that fails to do justice
to the complexity of the interaction. It is surely tempting to argue
that what all great early civilizations had in common was writing
and that it was hence writing that caused complex societies to
come into existence. However, if we look at the uses of writing
in early civilizations, many diferences are apparent. For exam-
ple, economic administration was preeminent in Mesopotamia
(Nissen, Damerow, and Englund 1990 ), whereas cult stood out
in Egypt (Assmann 1991 ). In both cases, it is untenable to argue
that accounting and the cult of the dead, respectively, were an
outfow of the invention of writing. Yet the opposite proposition,
claiming that the demands of bookkeepers and priests led to the
creation of writing, is no less simplistic .
Similarly, the invention of the printing press and mov-
able type has often been seen as a technological breakthrough
with vast social consequences (Febvre and Martin [1958] 1999 ;
Eisenstein 1979 ) . In our day, the digital turn (Fischer 2006 ),
described as the third media revolution after chirographic
culture (Schmandt-Besserat 1992 ) and print culture
4
SOCIAL PRACTICES OF SPEECH AND WRITING
Florian Coulmas
INTRODUCTION
Language is constitutive for human society. As a social fact
it cannot be thought of in the abstract, for the medium of
communication is what allows it to serve social functions. Te
nature of the social relationship that exists by virtue of language
partially depends on the externalization of language, that is, on
how it is transmitted from one actor to another as speech, writ-
ing, sign, or Braille . Te anatomy of speech organs (cf. Liberman
and Blumstein 1988) provides the biological foundation of
human society in the most general sense, which is why oral
speech is considered fundamental for socialization both in the
phylogenetic and ontogenetic sense. But unless we study human
society like that of other primates from the point of view of physi-
cal anthropology, other forms of language externalization must
also be taken into account as communication potential from
the beginning. Tere are two reasons for this. One is that the
invention of writing (sign language, Braille) cannot be undone.
Te other, which follows therefrom, is that writing has brought
about basic changes in the nature of human communication. It
brought in its wake a literate mindset that cannot be reversed.
Research about language in literate societies is carried out by
researchers who, growing up, were socialized into a literate
world organized by and large on the basis of literate principles.
It is not fortuitous, therefore, that social practices of speech and
writing are dealt with here under one heading.
Te scientifc enterprise in general, linguistics in particular, is
a social practice involving speech and writing. Even the investi-
gation of unwritten languages happens against the background
of literate society and by means of the tools developed for what
Goody ( 1977 , 151) felicitously called the technology of the intel-
lect. For the language sciences, it is important to keep in mind
that it is not just the technicians who use a tool to do what they
need to do and want to do, but that the tool restricts what can
be done . Tis holds true for the hardware, that is, the writing
implements, as well as for the software, the code or the writing
systems .
Te social aspects of speech and writing encompass a wide
range of topics many of which are dealt with in other entries of
this encyclopedia ( sociolinguistics , discourse analysis ,
The Cambridge Encyclopedia of the Language Sciences
36
can read the latest stock quotes of a scrolling light-emitting
diode (LED) display. Brand-new buildings are adorned with the
old technique of cutting messages in stone. Stelae with com-
memorative inscriptions, gravestones, and buildings bearing
the names of their owners or occupants are still being put up,
much as in ancient times. Tere are hardly any material objects
to which writing cannot be af xed. Since the earliest times of
literate culture, few have been discarded and many added. Te
choice continues to expand. Hard surfaces made for endurance
are stone, marble, metal, ceramics, wood, and, today, plastics.
Inscriptions are incised, engraved, etched, carved, and chiseled
into them as they were in the past, and malleable surfaces such
as moist clay and molten metal are impressed or molded into
shape.
In addition to monumental inscriptions, writing is found on
various other surfaces, such as whitewashed walls, street signs,
posters, billboards, handbills, notice boards, memorial plaques,
cloth, clothing, commercials carried around by sandwichmen
and mounted on trucks, advertising pillars, buses and other
vehicles covered with commercials, shop windows, and digital
display panels. Tese and some other surfaces, such as palm
leaves, papyrus, parchment, and wax tablets that have gone
out of fashion, are variously suitable for realizing the functional
potential of writing .
Two fundamental functions of writing are memory support
and communication . Tey are not mutually exclusive, but dif-
ferent surfaces lend themselves better to one than to the other.
Hard surfaces answer the requirement of durability. Tey are
inscribed only once, but with a lasting trace that can be recov-
ered after years, decades, even millennia. Baked clay tablets,
the hallmark of cuneiform civilization, and mural inscriptions
on Egyptian monuments embody this type. Memory is in time
turned into history, the recording of the past and the collection
of knowledge, which are the cardinal functional character-
istics of this technology. Inscriptions on hard surfaces are, of
course, also communicative but stationary. Clay tablets can be
transported in limited numbers only, and monumental inscrip-
tions have to be visited to be read. In order to allow written
signs to travel and thus to realize a potential that fundamen-
tally distinguishes writing from speech, freeing the message
from the copresence of sender and receiver, lighter materials
are needed. In antiquity, three main writing surfaces met this
requirement: papyrus, parchment, and paper.
For millennia, Egypt was practically the only producer of
papyrus because the reed of which it is made grows in abun-
dance along the banks of the Nile. Te papyrus scroll hieroglyph
is attested in the most ancient known Egyptian inscriptions, and
the oldest papyrus fragments covered with writing date from the
third millennium b.c.e . Papyrus came to be commonly used for
documentary and literary purposes throughout Greece, Asia
Minor, and the Roman Empire. As of the fourth century c.e. ,
parchment (processed animal hide), a more durable writing
material than the brittle papyrus, began to be more widely used in
Europe, where the scroll was gradually edged out by the book in
codex form (Roberts and Skeat 1983 ) . Te word paper is derived
from papyrus , but paper making is quite diferent from papyrus
making. It was invented by the Chinese some 1,900 years ago
(Twitchett 1983 ). Te earliest Chinese documents on paper date
(Olson 1994 ), is regarded as a driving force of globalization
(Kressel 2007 ). Both of these propositions are defensible, but
not in a unidirectional, monocausal sense. Equally true are the
opposite propositions, that socioeconomic developments led to
the emergence of a larger reading public, thus paving the way for
a simpler and more ef cient reproduction technology than the
copying of manuscripts, and that modern industrial society with
mass participation generated pressure for the development of a
technology of mass dissemination of information. Te invention
of writing facilitated complex social organization, and the print-
ing press was conducive to the spread of education. However,
writing has been a common possession of humanity for more
than 5,000 years and the printing press for half a millennium, if
we disregard the use of cast-metal movable type in Korea in the
early thirteenth century.
Yet we are living in a world with hundreds of millions of adult
illiterates, even, or rather particularly, where writing frst emerged,
that is, in Mesopotamia, in Egypt, in China, and in Mesoamerica.
According to unesco ( 2006 ), there were 781 million adult illiter-
ates worldwide and 100 million school-age children not attend-
ing school in 2005. In spite of the uneven distribution of illiterates
in the world, these fgures suf ce to discredit the notion that a
new technology of communication of and by itself brings about
social change. Economic development, social structure, ethnic
and linguistic composition, fecundity, ideology, and tradition
are intervening variables that determine how a society makes use
of and adjusts to a new technology. It is necessary, therefore, to
reckon with the contemporaneity of diferent levels of develop-
ment, diferent technologies, and diferent literacies . Assuming
a dialectic relationship of mutual infuence between writing and
social change is a more promising approach for understanding
the transition from oral to literate society.
New technologies both respond to practical needs and create
new practices. Any technology is an artifact, but to conclude that
its creators rule over it is a fallacy, for the applications of techno-
logical innovations are often recognized not in advance but after
the fact, when they have been used for some time. Like the genie
let out of the bottle, they may have unplanned and sometimes
unwelcome consequences. Te material and functional proper-
ties of writing technologies determine their potential uses, which,
however, are not necessarily evident at the outset.
Te locus of writing is the city. Even a superfcial look at
present-day urban environments reveals that city dwellers are
surrounded by written messages wherever they go. Of late, this
has given rise to a new branch of scholarship known as linguis-
tic landscape research (Landry and Bourhis 1997 ; Backhaus
2007 ), as it were, a social epigraphy for posterity. Te variety
of writing surfaces on which the literate culture of modern
cityscapes manifests itself is striking. It testifes to the traces of
history in the present and to the contemporaneity of diferent
stages of development, for it includes some of the oldest mate-
rials used for writing side by side with the most recent devices.
Tis contemporaneity is one of the foremost characteristics
of writing. For writing arrests change and enables accumula-
tion of information . Some genuine monuments from antiquity
speak to us today, such as the Egyptian obelisk of Ramses II of
the 19th Dynasty, 13041237 b.c.e. , re-erected on the Place de la
Concorde in the center of Paris. Around the corner, the passerby
Social Practices of Speech and Writing
37
digitalized contents would require some 750 gigabyte (GB) stor-
age space, which easily fts on an external hard disk the size of
a small book. As compared to print, digital information storage
thus reduces the necessary physical space by a factor of 50,000.
Again, this is a coarse measure only. Tere is considerable
variation in the bytes per book page, both in actual fact and
in model calculations, but the correlation between informa-
tion amount and storage space of print media and digital media
transpires from it .
In sum, as clay was followed by paper and paper by digi-
tal storage media, information density per square centimeter
increased exponentially, while the weight and size of written
records decreased. It became, accordingly, ever easier to store
and to transmit written text with many consequences for read-
ing and writing behavior, for reproducing text, and for the role
texts play in everyday life. What the history of writing shows is
that new technologies do not always replace old ones. Rather,
the new supplements the old and often transforms its use. For
instance, parchment was marginalized by paper but for centu-
ries never completely driven out; print never replaced handwrit-
ing and has not become obsolete by texts typed on a cell phone
keypad. Advances in writing technology have greatly expanded
the repertoire of tools that humanity has acquired for handling
information in the form of written language. Te fact that old
technologies continue to be used side by side with new ones
testifes not just to inertia and path dependency but also to the
diferent properties of the materials used. For centuries after
the introduction of paper, it was considered too feeble a mate-
rial to hold contracts and other important documents, which
were preferably executed on parchment. Similarly, although it
is technically possible to keep birth registers as electronic fles
only, birth certifcates on paper continue to be issued. One of the
reasons is that digital information storage is subject to physical
decay and technical obsolence not less but much more than pre-
decessor media .
Writing has made record keeping for posterity and accumu-
lation of knowledge possible. However, with the introduction of
every new writing material, the storage problem that it seemed to
solve became more acute. An archive of records on paper takes
up much less space than one for clay tablets, but it is beset by
dangers that pose no threat to baked clay: dust, humidity, insects,
fre, and water. Teft, too, is a greater threat to libraries than to
clay tablet archives and a greater threat to computers than to
libraries. Keeping books in a usable physical state requires more
work than keeping clay tablets. Te same kind of relationship
holds between libraries and digital data archives. Much more
can be stored, but preservation for future use becomes ever more
dif cult as time intervals of technical innovation shrink. Only a
few specialists are able to handle data stored with software that
ceased to be produced 20 years ago, whereas a book would last
for centuries. Te problem of preserving and organizing the ever-
swelling food of information remains unsolved, and at the same
time many traditional libraries and collections of documents fall
into decay. Technology has hugely expanded human memory,
but it has not yet eliminated the risk that many parts of the heri-
tage committed to writing will disappear forever .
To guard against collective memory loss, the United Nations
Educational, Scientifc and Social Orgnization (UNESCO) has
from the second century c.e . In the wake of the Islamic expan-
sion to Central Asia, the Arabs acquired the paper-making tech-
nology in the eighth century c.e., which they in turn introduced
to Europe in the eleventh century. Relatively cheap, fexible, and
convenient to carry, paper replaced parchment as the principal
writing surface in Europe and in other parts of the world.
Since its invention in China, paper, which Pierre-Marc De
Biasi ( 1999 ) called the greatest invention of all time, gave a
boost to the production of written text wherever it was intro-
duced. In China, it was used for block printing as of the seventh
century. In the tenth century, the entire Buddhist scripture was
printed using 130,000 printing blocks (Taylor and Taylor 1995 ,
156) . Paper was the frst writing material that spread around the
world. In the West, Johannes Gutenbergs invention of print-
ing with movable type would hardly have had the same impact
without it. Of the 180 copies of the Bible he printed, 150 were on
paper and only 30 on parchment, one indication of the impor-
tance of paper for the dissemination of written text. Its position
in this regard is undiminished . Te paperless of ce is far from a
reality even in the most advanced countries; rather, many ancil-
lary devices that presuppose writing on paper form the mainstay
of the thriving of ce machines industry: printers, scanners, copi-
ers, and fax machines. Although nowadays paper holds only a
tiny fraction of all new information released, it is still the unchal-
lenged surface for the formal publication of information. World
paper consumption for information storage and distribution is at
an all-time high. Notwithstanding the shift of many periodicals
and scholarly journals to online publication most continue to
be printed on paper for archival purposes, for paper has a much
longer duration of life than can be guaranteed for any digital
storage medium .
Tis brings to light a more general trade-of of information
processing. Weight and storage capacity are inversely related. A
tablet measuring about 10 10 cm is the ideal size for writing on
wet clay. It holds about 300 characters. Depending on the thick-
ness of the tablet, this yields an information/weight ratio of .2 kg
to 1 kg per 1,000 characters. A text of 300,000 characters would
weigh between 200 kg and 1000 kg. In modern terms, that would
be a short book of fewer than 190 pages, assuming an informa-
tion density of 1,600 characters per page. Give it a solid cover
and it comes to a total of 250 g. With respect to the information/
weight ratio, paper thus outperforms clay by a factor of 4,000.
Such a rough-and-ready calculation may suf ce to illustrate
the point. Papyrus was similarly superior to clay with regard to
storage capacity and transportability; however, many more clay
tablets than papyrus documents have come down to us through
the millennia. How many papyrus rolls were lost in the legendary
blaze of the library of Alexandria is not known, but when in 2004
a fre broke out in the Anna Amalia Library in Weimar, 50,000
books were destroyed, many of them unique or rare. Another
65,000 volumes were severely damaged by fre and water. Baked
clay tablets would have withstood the fames and the water used
to put them out .
Tis line of thought can be extended into the digital age
by another calculation. Computer technology has exponen-
tially increased information storage density. Te 50,000 burnt
books of the Anna Amalia Library took up some 1,660 meters of
shelf space . Assuming an average of 300 pages per book, their
The Cambridge Encyclopedia of the Language Sciences
38
years memorizing the Confucian classics and commentaries.
Tey were then able to recite, understand, and interpret every
clause of the fve canonical works Book of Changes, Book of
Documents, Book of Poetry, Records of Rites, and Spring
and Autumn Annals said to have been redacted by Confucius
himself, as well as a collection of commentaries by subsequent
scholars.
Tat such an education was an adequate preparation for
bureaucrats charged with administering the country was rarely
called into question. It was frmly rooted in the past, and the clas-
sics were thought to hold a solution to any problem that might
arise. Te authority of writing and the conservatism of literature
were never more efective. Te strength of the system lay in the
fact that it encouraged respect for learning and provided the
emperor with a bureaucracy educated in one standard curricu-
lum. Its weakness was its emphasis on commentary that stifed
inquisitiveness and deviation from the trodden path. Te civil
service exam system institutionalized the transmission of texts
down a lineage and was, thus, inherently averse to change. In its
early days, it helped to loosen the hereditary aristocracys grip
on power by rewarding merit rather than birth for recruiting
bureaucrats. In actual fact, however, learning remained largely
a prerogative of aristocratic families out of reach for most com-
moners. Women were not permitted to sit for the exams. In the
end, the civil service examinations served as a system to perpetu-
ate the power of the thin elite of literati bureaucrats .
Controlling literacy has always been the other side of its rela-
tion to authority. Te powerful have lived in fear of the power
of the pen and have had little interest in promoting the art of
writing among the masses. Illiterates are powerless, unable to
challenge the letter of the law or to invoke the laws on the books
to their own advantage. Tat which is written down acquires
authority in its own right as a reference source independent of
the ruler. While helping to project power far beyond earshot, it
gains a measure of objectivity, thereby reducing the arbitrari-
ness of rule. But only the literate can hold the ruler accountable
to his own decrees. Te institutionalization of writing to this end
occurred in ffth-century Greece where government was estab-
lished in the polis through written laws. Tese were not God-
given but man-made laws, aiding the development of a division
between cosmic order and human society (Stratton 1980 ), so
characteristic of the Greek Weltanschauung and so diferent
from the Chinese .
From the objectifcation of language in writing follows
another function with important implications for the exercis-
ing and curbing of power. Writing detaches the author from
his message, which makes it easier and less risky to express the
unspeakable. Two examples suf ce to illustrate. In Athens, ostra-
cism was institutionalized as a procedure to protect democratic
rule. In the event that a charismatic politician (a demagogue)
became too infuential or was otherwise disruptive, the demos
(people) were entitled to exile him from the polis. To this end,
every citizen was given a potsherd ( ostrakon ) on which to write
the name of the man to be sent into exile. Te degree of literacy in
Athens that can be inferred from this practice is a question that
has given rise to much research and speculation (Burns 1981 ;
Harvey 1966 ; Havelock 1982 ; W. Harris 1989 ; Tomas 1992 ). It is
unlikely that it will be possible ever to quote even approximately
launched the Memory of the World Programme to assist in the
preservation of archive holdings and library collections all over
the world . For the time being, this is the endpoint of a develop-
ment begun with the advent of literacy in ancient civilizations: the
institutionalization of writing and the bureaucratization of soci-
ety. Te more serviceable writing became to human society, the
more it penetrated social relations and the more attention it
came to require on the part of society .
INSTITUTIONAL ASPECTS OF LITERACY
From its inception, writing has been an instrument of power. In
ancient civilizations of restricted literacy, its mastery was jeal-
ously guarded by the elite. It was indispensable for the work-
ings of the temple-centered economies of ancient Near Eastern
city states (Nissen 1988 ), symbolized the rule of the pharaohs in
Egypt (Posener 1956 ), became the bedrock of Chinas Confucian
bureaucratic state (Lewis 1999 ), and was a sine qua non of
Athenian democracy (Tomas 1992 ). Certainly, literacy levels
varied widely as did the uses of literacy, but the general ten-
dency of the extension of the human mind by means of writing to
engender and necessitate institutions is unmistakable. Te most
important institutions produced by literate culture have to do
with government, cult, schooling, and economic organization.
Government
Writing was used early on to extend the reach of authority and to
mediate the relationship between ruler and ruled. Monuments
that embody power, such as the Rosetta Stone inscribed with a
decree to commemorate the reception of rulership by Pharaoh
Ptolemy V on March 27, 197 b.c.e. (Parkinson 1999 , 29), as well
as stelae appealing with regulations to the literate public, were
erected throughout the Ancient Near East. Teir inscriptions
were drafted by scribes who created the frst bureaucratic states.
Te Egyptian vizier was responsible for the collection of taxes,
the maintenance of archives, and the appointment of of cials.
Te skills of the scribe aforded privilege in Egyptian society. As
one of them put it, the scribe is released from manual tasks; it
is he who commands (Goody and Watt 1968 , 37) . A thousand
years later, Mencius (372289 b.c.e. ) made the same point in
China: Some labour with their hearts and minds; some labour
with their strength. Tose who labour with their hearts and
minds govern others. Tose who labour with their strength are
governed by others (Book of Mencius, quoted from Lloyd and
Sivin 2002 , 16). Tese observations bring to light the connection
between literacy and social hierarchy that persists to this day .
Wherever societies obtained the art of writing, the literati
were close to the powers that be, but the Chinese institutional-
ized literacy like no other culture. In Confuciuss day, literacy was
already the preeminent mark of the gentleman, and as of the sec-
ond century b.c.e. , the study of the Confucian classics gradually
became institutionalized as the key to attaining public of ce and
infuence . Te civil service examination system was employed
with few adjustments from the Han Dynasty (206 b.c.e. 220 c.e. )
until the closing stages of the Qing Dynasty at the beginning of
the twentiethth century. It was based entirely on the study of
texts. To prepare for the highest degree, students spent up to 20
Social Practices of Speech and Writing
39
sacred and the profane may take a long time to complete. Te
Ten Commandments, written by God and given to Moses, were
a code of conduct regulating the relations between God and the
people, as well as the people among themselves. Te spheres of
spiritual and worldly power were only beginning to be separated
in antiquity, a process to which writing and the ensuing inter-
pretation of texts contributed a great deal. Of Moses legendary
Ten Commandments stone tablets no archaeological evidence
remains, but the Aoka edicts are a tangible example of the
closeness of cult and social order . Engraved in rocks and stone
pillars that have been discovered in dozens of sites throughout
northern India, Nepal, Pakistan, and Afghanistan, these edicts
served to inform the subjects of King Aokas (304232 b.c.e. )
reforms and make them lead a moral life according to the truth
of Buddhism. Moral precepts hewn in stone on public display
once again raise the question of the degree of literacy. It must
have been high enough to disseminate the message through-
out the vast realm. In the rich spiritual world of the Indian sub-
continent, much emphasis was always given to the importance
of oral tradition. Competing with other religious movements,
Buddhism established itself by rigidly regulating monastic life
and assembling a canon of scriptures. At the time of Aoka,
Buddhism was an institutionalized religion that spread from
northern India to other parts of the subcontinent, frst and sub-
sequently to South and Southeast Asia, as well as Central Asia
and China.
Te history of Buddhism and its emergence as a major world
religion is a history of translation. Translation means writ-
ing, turning one text into another and all it involves: exegesis,
doctrine, scholasticism, and schism. As Ivan Illich and Barry
Sanders ( 1988 : 52) have argued, there is no translation in oral-
ity, but only the collaborative endeavor to understand ones
partner in discourse. Writing eternalizes spiritual enlighten-
ment (e.g., the word of God), which must be preserved with
the greatest care and does not allow alteration at will. Of all
religions, Buddhism has produced by far the largest corpus
of sacred texts. Signifcantly, diferent canons resulted from
translation, the Pali canon, the Tibetan canon, and the Chinese
canon. With translation came schism, and with that the delimi-
tation of religious sects and districts. In the evolution of other
book religions, translation had similar consequences. Tese
other religions share with Buddhism the vital importance they
attach to scriptures. Te holy book is what distinguishes world
religions from other cults. Teir legitimacy is derived from the
revelation embodied in the scriptures, which are considered the
true source of enlightenment. Te major world religions vary in
the role they assign sacred texts, in how they make use of them
for purposes of propaganda and the regulation of life, but the
reverence accorded to writing is a feature they share, as is the
question of translation .
Translation is a problem for two reasons that can be labeled
authenticity and correspondence . First, the idea of authentic-
ity of the word of God does not agree with its transposition into
another form. Some book religions are very strict in this regard.
According to believers, God himself chose the Arabic language
for his fnal testament, the Quran, and created every one of the
letters of the Arabic alphabet. Consequently, the only true ver-
sion of the holy book of Islam is the Quran in Classical Arabic.
correct fgures, but what we know of the literate culture of the
Greek city-states is that, unlike in China, there was no scribal
class. Minimal competence such as was necessary to scratch a
name on a potsherd was relatively widespread. Both written law
as the basis of social order and ostracism as a check on power
exemplify the institutionalization of writing as a precondition
for political participation and as a seed of the public sphere. It
enabled the people to express discontent with a leader without
raising their hand individually or speaking out. Anonymity was
a protection from reprisals.
Te other example is from Babylon, as reported in the
Old Testament. Writing means empowerment and, there-
fore, has to be controlled. Te proverbial writing on the wall,
mene-tekel -parsin , an Aramaic phrase that, according to Daniel
5:25, mysteriously appeared on the wall of King Belshazzars pal-
ace, cautioning him that his days were numbered and his empire
was doomed, exemplifes the potential of criticism. A message
by an unknown author exposing abuse of power to the public is
a direct challenge to authority. While it would be problematic to
voice disapproval in the presence of others, writing afords the
originator the protection of anonymity .
Te dialectics of technological innovation come to bear
here. While the mighty use writing to establish, exercise, and
consolidate their power, it also lends itself to embarrassing and
undermining them. Tis was clearly understood in antiquity.
Power holders always tried to curtail the power of the written
word. Confuciuss Analects were burned in 200 b.c.e. on order
of Emperor Pinyin Qin Shi Huang Di . Plato was convinced that
the state had to control the content of what pupils read. Tacitus
refers to banned books (Anales 4,34), and his own books were
later banned in return. Censorship is ubiquitous throughout
history (Jones 2001 ). Many of the fnest literary works were
at one time or another put on a list of forbidden books, such
as the papal Index Auctorum et Librorum Prohibitorum , frst
published in 1559. It took another 500 years for censorship to
be universally censured as an illegitimate means of exercising
power. Article 19 of the Universal Declaration of Human Rights
adopted by the United Nations in 1948 states that everyone has
the right to freedom of expression and information, regardless
of frontiers .
Te struggle against censorship worldwide is far from over.
With every new writing technology it is rekindled, as the debate
about controlling the Internet illustrates. Te power elites try to
control the new media, which they perceive as a threat, although,
ironically, they are often involved in its development. Te battle-
feld shifts with technological advance. What were scriptoria and
publishing houses in the past are servers and the fow of data
through cyberspace today. In the long run, attempts on the part
of governments to defend their lead and keep new information
technology out of reach of their adversaries have failed because
it is the very nature of these technologies that they can be utilized
to uphold and to counter the claim to power .
Cult
In nonliterate societies religious, order and social order are
merged without clear division. Writing tends to introduce a
fracture between spheres, although the diferentiation of the
The Cambridge Encyclopedia of the Language Sciences
40
languages or varieties of the same stock. In multilingual envi-
ronments like medieval Europe, where Latin was the only writ-
ten language, or in present-day India, cultivated languages are
often reserved for writing and formal communication, while ver-
nacular varieties are used in informal settings. A similar division
is found between linguistic varieties of the same stock, where
one is defned by reference to a corpus of classical texts, such as
Classical Arabic, whereas the others fuctuate without artifcial
restriction .
Writing introduces an element of art and artifciality into the
history of language. Every language is the collective product of
its speakers, but a written language is more clearly an artefact
than a vernacular, and the script that it uses more clearly yet.
Historically, the difusion of scripts coincided in large measure
with that of religions, a connection that is still visible today.
Chinese characters arrived in Japan together with Buddhism.
Te spread of Roman traces the expansion of both the Roman
Empire and the Catholic Church, while Orthodox Christianity
uses Cyrillic. Armenian Christians have their own alphabet
designed in the ffth century by St. Mesrob. Estrangela is the script
of the Syrian Church. Te Hebrew square script is the script of
Judaism, the Arabic alphabet that of Islam. Many other examples
could be added; clerks were churchmen (Coulmas 1996 , 435 f.).
Te historical interconnectedness of writing and religion is one
of the reasons that scripts tend to function as symbols of iden-
tity, but ethnic, national, and political identity are also readily
expressed by means of a distinct script or even slightly diferent
orthographic conventions . As Peter Unseth ( 2005 ) has pointed
out, there are clear sociolinguistic parallels between choos-
ing scripts and languages. Because of the visibility and the artif-
cial nature of writing, however, the choice of scripts is generally a
more deliberate departure from tradition in that it involves con-
scious planning .
Schooling
Language is a natural faculty, writing an artifact. Tat is the rea-
son why children acquire language, but not writing, without
guidance. Te dif cult art of writing requires skills that must be
taught, memorized, and laboriously practiced. Te place to do it
is school. For writing to be useful to the community, conventions
have to be established, individual variation curtailed, norms set.
Collective instruction following a curriculum is a more ef cient
way to achieve this than is private tutoring. Already in antiquity,
school became, and still is, the institution that most explicitly
exercises authority over the written language by controlling its
transmission from one generation to the next.
With schooling came the regimentation and the decontex-
tualization of language. Because in writing the continuous fow
of speech has to be broken down into discrete units, analytic
refection about the units of language was fostered. As writing,
language became an object of investigation and normalization.
Both grammar and lexicon are products of writing. Tis is not to
deny the grammaticality of oral speech or that oral people have
a mental lexicon. It just means that the notions of grammar and
lexicon as we know them are entirely dependent upon writing.
At school, units of writing had to be practiced with the stylus, the
brush, or the pen, mechanically through repetition without any
Te Christian tradition until the Reformation movement knew
similar limitations, recognizing only the three sacred tongues of
Hebrew, Greek, and Latin as legitimate languages of the Bible.
In many cases, other languages were considered unft for the
expression of divine revelations, if only because they lacked the
fxity that comes with being reduced to writing. Indeed, trans-
lations of religious scriptures, when they were eventually pro-
duced, were for many languages their frst literary texts, serving
as an anchor in the fuidity of oral discourse and as a starting
point of literacy and textual transmission .
Te other reason for the problematic nature of translation
is that through it, a stable correspondence is to be established
between two codes that though obviously diferent in lexical,
grammatical, and phonetic makeup, must express the same
contents. In order to translate, the meaning of the text has to
be established unequivocally, the assumption being that this
is possible. Te text is thus elevated to the source of meaning.
Authority that formerly accrued to the sage, the reciter, and the
soothsayer was relocated in text. Tis transition from utterance
to text (Olson 1977 ) implies that in order to understand a mes-
sage, at issue no longer is what the speaker means but what the
text contains. Tis is what translation is all about. Language itself
in objectifed form thus becomes a force in its own right with far-
reaching implications in the domains of knowledge production
and social regulation. Preservation of the word of God in text
has provided an objectifed reference plane incorporating the
true meaning waiting to be extracted from it. Olsons notion
of autonomous text that results from the transition from utter-
ance to text as he conceptualizes it has been criticized because
it ignores the readers involvement in constructing meaning
(Nystrand 1986 ) and because it underestimates the oral ele-
ments in literate culture (Wiley 1996 ). Te soundness of these
criticisms in detail, however, does not invalidate the general idea
that writing gives language a measure of stability that it does not
have in speech, and brings with it a shift from the intentional to
the conventional aspects of linguistic meaning, a shift from I
mean to the word means. Te high prestige that the written
word acquired through its association with and instrumentaliza-
tion by organized religion has greatly contributed to the coming
into existence of autonomous text .
Te reverence for holy books had various consequences for
language attitudes and practices, two of which can be men-
tioned here: code choice and script choice (Coulmas 2005 ).
Writing introduces functional domain diferentiation into a
communitys linguistic ecology. Tat the language of cult difers
from that of everyday pursuits has always been the rule rather
than the exception, but with writing the distinction becomes
more pronounced. Te important position of religious texts in
combination with restricted literacy encouraged the perpetu-
ation of the split between spoken and written language. While
the codifcation of the latter was aided by the desire to transmit
the sacred texts inviolately to later generations, the former was
subject to perpetual change. Te result was a situation of coexist-
ing codes, called diglossia in modern scholarship (Ferguson
1959 ; Krishnamurti 1986 ; Schifman 1996 ). Although every case
of diglossia is diferent, the defning characteristic is a domain-
specifc usage of varieties that coincides by and large with the
spoken/written language divide. Tese varieties can be diferent
Social Practices of Speech and Writing
41
issue. In medieval Europe, the ad litteras reform during the reign
of Charlemagne aimed at unifying spoken and written language,
as the widening gap between both was perceived as a problem.
It was eventually reduced, not so much by enforcing a uniform
standard for pronouncing Latin than by dethroning it as the only
written language and transforming lingua illiteratae (Blanche-
Benveniste 1994 ), that is, Romance, Germanic, and Slavonic
vernaculars, into written languages in their own right. Literacy
in these emerging national languages was bolstered by the
Reformation movement that wrested the interpretation monop-
oly of Christian scriptures from the Catholic clergy. Write as you
speak, a maxim that can be traced back to antiquity, became
an increasingly important principle for teaching writing
(Mller 1990 ). Although it unrealistically denies the speech/
writing distinction, generations of teachers have repeated it to
their pupils. It never meant that their writing should be as ellip-
tical, situation-bound, and variable as their speech. Te impli-
cation is that if you cannot write as you speak, something must
be wrong with your speech. Universal education on the basis of
this maxim resulted in a conceptual reduction of the distance
between speech and writing, with some notable consequences.
Mass literacy through schooling led to the disappearance of
diglossia from most European speech communities, although
the dichotomy of speech and writing continued to be expressed
in stylistic diferences.
In other parts of the world, where universal education was
realized later or is still only a desideratum, the split between
spoken and written language remained. In todays world, the
1953 UNESCO declaration recommending mother tongue lit-
eracy notwithstanding, literacy education in the language of
the nation-state often means learning to read and write in a
second language. Te extent to which progress in the promo-
tion of literacy depends on the language of instruction is still
a matter of controversy, as is whether the writing system is a
signifcant variable. To some extent, this can be explained by
the fact that defnitions of literacy are shifting with changing
socioeconomic needs and technical innovations, and because
the range of what are considered varieties of a given language
is variable (as is evident, for example, in the context of decre-
olization and discussions about elevating certain varieties,
such as black English in the United States, to language status).
Tere is, however, wide agreement that the crucial variable
is the efectiveness of the educational system. Mastering the
written language is a dif cult task, which is best executed by
the institution that at the same time administers the written
language: school .
Since the time of the French Revolution, schools have been
charged with establishing the national language and, by way of
spreading the national language ideology, a link between lan-
guage and nationalism . As a result, the demands of multilin-
gual education are often at variance with the state-sponsored
educational system. Because of the nationalization of languages
in the modern nation-state and their privileged position in the
school system, however, the language of literacy training became
a political issue. Minority speech communities in many indus-
trialized countries aspired to the prestige for their language that
comes with a written standard and started to lobby for the inclu-
sion of their language in the school curriculum. Fueled by the
communicative intent. Tese units could be given a phonetic
interpretation; they could be pronounced and thus acquired
as words , an existence as units of language. In time, the image
became the model. Since the correct form of the written sign was
a matter to which the scribe and his pupils had to pay attention,
standards of correctness frst developed with reference to writing
and written language.
Only much later, and as an efect of schooling, did these
notions come to be applied to speech. Te twin questions of
what the units of writing were and how they were to be conjoined
led to the more general and profound question What is a lan-
guage? Right up to the present, answers to this question exhibit
what Roy Harris ( 1980 ) has called a scriptist bias . Only trained
linguists readily recognize unwritten vernaculars as languages,
and even they have to admit that while it is easy to count written
languages, the question of how many speech forms on this planet
qualify as distinct languages is impossible to answer without lay-
ing down analytic criteria that are no less arbitrary than decisions
as to what dialects , varieties, idioms, and speech forms should
be reduced to writing. Languages, as well as the units into which
they are analyzed, are a product of writing, for only in writing can
the fow of speech be arrested and broken down into indepen-
dent stable components with a presumed inherent, rather than
vague and contextually determined meaning.
Among the frst results of school mastering the language
in the Ancient Near East were word lists, the paradigm case of
decontextualized language. Tese lists were the foundation of
lexicography (Civil 1995 ), the science of words. In China, lexi-
cography began with lists of characters, and characters are still
the basic units of dictionaries. Dictionaries provide entries for
lexical items. A lexical item is a word contained in a dictionary.
More refned and less circular defnitions of the orthographic
word , as distinct from the phonological word and the abstract
lexeme have been proposed in great number, but it remains
dif cult if not impossible to defne word without reference to
writing. Te word stands at the beginning of grammatical schol-
arship, which was, as the word grammatical itself indicates,
exclusively concerned with written language. Grammatike ,
combining the Greek words grammata (letters) and techne
(art), was the art of knowing letters . Tese beginnings of the
systematic study of language left a lasting imprint which, as
Per Linell ( 2005 ) has convincingly shown, still informs modern
linguistics. Te word , the sentence , and even the phoneme
are analytic concepts derived from the discrete segments of
writing, not vice versa. Te conceptualization of writing as
a representation of speech is therefore highly problematic
(R. Harris 1980 , 2000 ). To sum up this section, the institution-
alization of writing in school resulted in a changed attitude to
language. It became an object of study and regulation. Both of
these concepts were not in the frst instance developed for, and
applied to, speech.
Under conditions of restricted literacy and diglossia, a wide
divide between spoken and written language was taken for
granted. Speech and writing were two modes of communication
involving varieties or languages that were both grammatically
and stylistically quite separate from each other. It was only when
literacy became accessible to wider sections of the population
that the relationship between speech and writing became an
The Cambridge Encyclopedia of the Language Sciences
42
the relationship between writing and economy is not a simple
one. For one thing, the degree of literacy necessary for economic
growth is a matter of controversy and depends on how economic
growth is measured. Illiteracy is considered a strong indicator of
economic underdevelopment, correlating as it does with infant
mortality, low life expectancy, and poverty; but the question of
whether high literacy drives economic growth or is a function
thereof remains unresolved. Functional illiteracy rates of up to
20 percent in Western countries, notably in the United States,
suggest that at least in the developed world, literacy rates are
more indicative of economic disparity than of overall national
economic development. Similarly, in developing countries,
illiterates are largely rural and without access to public services
(Varghese 2000 ). Te distribution of wealth and the division of
labor in todays national economies are such that they allow for,
or perhaps even sustain, substantial residues of illiteracy. Both in
the developed and the developing world, people who partake of
the practice of writing live side by side with others who continue
to conduct their life in the oral mode .
It is fair to say that the evolution of writing in antiquity both
happened in response to and propelled economic develop-
ment. Yet although writing technology has been available for
economic pursuits for more than fve millennia, fully liter-
ate societies remain an ideal. Tis shows that the relationship
between economy and institutionalized writing is subject to
interference by other variables, notably those discussed in the
previous sections, that is, government, religion, culture, and
education. In antiquity, these spheres and the economy were
not separate. It was a long process that led to their diferentia-
tion in modern times. As the chief instrument of knowledge
management and growth, writing was indispensable for ratio-
nalization and what Max Weber ( 1922 ) called the disenchant-
ment of the world, which, as the religious associations and
emotional values attached to various written languages testify,
is still incomplete.
Te interaction of writing with the economy has been studied
from various points of view. Writing can accordingly be under-
stood as
- a tool,
- social choice,
- a common good,
- human capital,
- transaction cost.
Writing is a tool that enables individuals to expand their com-
munication range and communities to increase social integra-
tion and differentiation. It is useful and valuable. Since writing
systems, as discussed previously, are artifacts, this raises
the question of how this tool evolves so that it is functional.
Students of writing, notably I. J. Gelb ( 1963 ), have invoked
the principle of least effort, which predicts that in time, writ-
ing systems become simpler and easier to use . The relative
simplicity of the Greek alphabet, as compared, for example,
with Egyptian hieroglyphs and early cuneiform, lends support
to this hypothesis. However, intuitive though it is, there are
problems. Obvious counterexamples are found outside the
ancient Near Eastern civilizations mainly studied by Gelb. In
growing awareness of the importance of minority protection,
such movements have met with a measure of success, leading to
a highly complex situation of multiple and multilingual literacies
in modern societies, which of late has attracted scholarly atten-
tion (Martin-Jones and Jones 2000 ; Daswani 2001 ; Cook and
Bassetti 2005 ) .
Te prevailing view sees the establishment of a single national
language with a unifed written standard as facilitating universal
literacy. International comparisons of literacy rates are notori-
ously dif cult (Gurin-Pace and Blum 1999 ), but there is little
doubt that Europe, where national language academies frst
implemented the national language ideology, led the way. Today,
however, the monolingual model of literacy is called into ques-
tion by developments that, on the one hand, favor English as a
supplementary universal written language in many non-English-
speaking countries, and on the other, allow minority languages
to make inroads into the domains of writing. Te question of
whether the diversifcation of literacy will help achieve the goal
of eradicating illiteracy or whether it will compromise the alleged
economic advantage of having one written standard language
continues to be discussed by academics and politicians, while
the complementary developments of globalization of markets
and (re)localization of cultures unfold .
Economic Organization
Te critical importance of writing for economic processes
in early civilizations is best documented for Mesopotamia.
It is widely agreed now that in Sumer, number concepts and
numerical representation stood at the beginning of writing
that evolved into cuneiform (Schmandt-Besserat 1992 ). Te
overwhelming majority of archaeological fnds from Ur, Uruk,
Nineveh, and Babylon are records of economic transactions
kept in clay tablet archives by accountants of the palace admin-
istration (Nissen, Damerow, and Englund 1990 ). Te Sumerians
and their Akkadian successors were the frst to develop book-
keeping into a sophisticated technique of balancing income and
expenditure. Hammurabi , King of Babylon (17281686 b.c.e. ),
created the frst commercial code, 282 articles written on a large
stone monument, which was erected in a public place for all to
observe. Every business transaction was to be in writing and
signed, usually with a seal, by the contracting parties. At the
time, the public-sector economy was far too highly developed
and too complex to function without writing. Tribute quota lists
had to be kept, rations for laborers involved in public works cal-
culated, inventories recorded. Deeds were issued in duplicate
and receipts stored for future reference. Large-scale trading,
often involving credit and futures, had to be regulated and
overseen by the bureaucracy, consisting of a huge scribal class
charged with creating and servicing these documents.
Ancient Mesopotamia is the paradigm case of the close inter-
connectedness of economy and the institutionalization of writ-
ing, but if economic behavior is understood in the wide sense of
human adaptations to the needs and aspirations of society at a
given moment in history, this interconnectedness can be identi-
fed in every literate society. Complex socioeconomic systems of
managed production, distribution and trade, taxation, and credit
did not evolve in cultures that had no writing . Yet the nature of
Social Practices of Speech and Writing
43
pertinent than ever (Levine 1994 ) and can explain processes such
as the accelerated spread of English. Te globalization of mar-
kets and the information technology revolution ofer ever more
people the opportunity to enhance their human capital and, at
the same time, compel them to do so. However, the commodif-
cation of written language (Heller 2003 ) and the new forms and
uses it takes on in the new media have consequences, some of
which become apparent only as technology spreads.
Conceptualizing written language as transaction cost brings
market forces into view. Reducing transaction costs is consid-
ered a quasi-natural precondition of economic growth, which
partly explains the spread of trade languages. Once in place and
controlled by relevant agents, their use is less costly than transla-
tion. In todays world, this principle seems strongly to favor the
further spread of English. However, the efects of technological
innovation are, as always, hard to foresee and even to assess
when it unfolds in front of our eyes. When the Internet ceased
to be a communication domain reserved to the U.S. military,
partly due to the available software at the time, it seemed to be
poised to become an English-only medium. But it turned out
that as the technology caught on, the share of English in cyber-
space communication rapidly declined. Te new information
technology made it much easier for speech communities big and
small around the world to communicate in writing, a possibility
eagerly exploited wherever the hardware became available. For
some communities this meant using their language in writing for
the frst time. In others it led to the suspension of norms, blur-
ring, in many ways yet to be explored, the traditional distinctions
between spoken and written language.
David Crystal ( 2001 ) suggests that Netspeak or computer-
mediated language is a new medium, neither spoken language
nor written language . Te implications of Internet communica-
tion for literacy in the electronic age are only beginning to be
explored (Richards 2000 ). New multilayered literacies are evolv-
ing, responding as they do to the complementary and sometimes
conficting demands of economic rationality, social reproduc-
tion through education, ideology, and the technical properties of
the medium. Tese developments open up a huge new area of
research into how, since the invention of writing, the range of lin-
guistic communication options has been constantly expanding .
CONCLUSION
Writing is a technology that interacts with social practices in
complex ways, exercising a profound infuence on the way we
think, communicate, and conceptualize language. Since it is an
artifact, it can be adjusted deliberately to the evolving needs
of society, but it also follows its own inherent rules that derive
from the properties of the medium. Tis essay has analyzed the
tension between the properties of the medium and the designs
of its users from two points of view, technological advance and
institutionalization. Harnessed by institutions, the technology
of writing is made serviceable to the intellectual advance of
society and modifed in the process, sometimes gradually and
sometimes in revolutionary steps. Tree consequences of writ-
ing remain constant: segmentation, linearization, and accu-
mulation. Te linear and discrete-segmental structure that all
writing systems both derive from and superimpose on language
its long history, the Chinese writing system and its adaptation
in Japan increased rather than decreased in complexity. On
the other hand, it took 500 years for the much simpler Korean
Hangul to be accepted as an alternative to Chinese. If sim-
plicity and adaptability were the forces that drive the spread
of writing systems, Hangul should have supplanted Chinese
characters, not just in Korea but in China and elsewhere long
ago. Looking at the evolution of individual systems, certain
letter forms in the Tibetan script were simplified to the extent
of becoming indistinguishable, rendering the system almost
dysfunctional. One has to conclude that if the principle of
least effort is at work in the evolution of the technology of the
intellect, then it is often compromised by other principles,
such as path dependency, that is, the fact that change in soci-
ety depends on established ways, identity affirmation (see
ethnolinguistic identity ), and cultural exclusivism .
Considering writing and written language from the point
of view of social choice leads to a similar conclusion. A writ-
ten language can be understood as an institution with certain
properties that is shaped by the agents involved in its opera-
tion. However, a single criterion of optimization of reasonable
target functions cannot explain the diversity of systems that
evolved, the multiplicity of uses, or the actual shape of indi-
vidual systems, such as, for example, English spelling or the
Japanese usage of Chinese characters (Coulmas 1994 ). Clearly,
established writing conventions are the result of public choice.
No individual can introduce a better standard, even though its
superior qualities may be indubitable, for conformity with the
existing norm is crucial for the functionality of the institution.
Penalties on norm violations are therefore high, and proposals
for changing established practice invariably meet with strong
resistance. Change is not impossible, but it comes at a cost and
rarely uncompromisingly follows functional optimization.
A written language, if used by a community, has properties of
a common good. Like a public transport system, it is benefcial
for most members of that community and therefore deserving of
the attention of the state. Tis is the rationale underlying the legal
protection enjoyed by national languages , explicitly or implicitly,
in modern nation-states. A written language not used by every-
one is not a common good and treated accordingly. Not serving
anyones interest, dead languages are of interest to the historian
at best. For the same reason, it has always been dif cult to ele-
vate a vernacular to the status of written language; not providing
access to information and not being used by many members of
the community in the beginning, it does not count as a common
good. Its claim to written language status is supported not by its
instrumental value but only by the symbolic value for its commu-
nity. To reconcile the idea of a dominant language as a common
good with the recognition of minority rights is therefore prob-
lematic. With Pierre Bourdieu ( 1982 ), it can be conceptualized
as a struggle over legitimacy, competence, and access ( market ,
linguistic ; habitus , linguistic ). Only if social harmony is
made part of the equation will accommodation be possible .
A written language, the cognitive abilities its mastery implies,
and the information to which it provides access are a resource.
Partaking in it adds to an individuals human capital and hence
to his or her marketability. In the information economy of
the knowledge society, this view on written language is more
The Cambridge Encyclopedia of the Language Sciences
44
Harvey , F. D. 1966 . Literacy in Athenian democracy . Revue des tudes
Grecques 79 : 585 635.
Havelock , Eric A. 1982 . Te Literate Revolution in Greece and Its Cultural
Consequences . Princeton, NJ : Princeton University Press .
Heller , Monica . 2003 . Globalization, the new economy, and the com-
modifcation of language and identity . Journal of Sociolinguistics
7 .4: 473 92.
Illich , Ivan , and Barry Sanders . 1988 . Te Alphabetization of the Popular
Mind . San Francisco : North Point .
Jones , Derek , ed. 2001 . Censorship: A World Encyclopedia . Vols.14.
London : Fitzroy Dearborn .
Kapitzke , Cushla . 1995 . Literacy and Religion: Te Textual Politics and
Practice of Seventh-Day Adventism . Amsterdam and Philadelphia : John
Benjamins .
Kressel , Henry , with Tomas V. Lento . 2007 . Competing for the Future
How Digital Innovations Are Changing the World . Cambridge :
Cambridge University Press .
Krishnamurti , Bh ., ed. 1986 . South Asian Languages: Structure,
Convergence and Diglossia . Delhi : Motilal Banarsidass .
Landry , Rodrigue , and Richard Y. Bourhis . 1997 . Linguistic landscape
and ethnolinguistic vitality . Journal of Language and Social Psychology
16. 1: 23 49.
Levine , Kenneth . 1994 . Functional literacy in a changing world. In
Functional Literacy: Teoretical Issues and Educational Applications ,
ed. Ludo Verhoeven , 11331. Amsterdam and Philadelphia : John
Benjamins .
Lewis , Mark Edward . 1999 . Writing and Authority in Early China. SUNY
Series in Chinese Philosophy and Culture . Albany : State University of
New York Press .
Liberman , Philip , and Sheila Blumstein . 1988 . Speech Physiology, Speech
Perception, and Acoustic Phonetics . New York : Cambridge University
Press .
Linell , Per . 2005 . Te Written Language Bias in Linguistics: Its Nature,
Origin and Transformation . London : Routledge .
Lloyd , Geofrey and Nathan Sivin . 2002 . Te Way and the Word: Science
and Medicine in Early China and Greece . New Haven and London : Yale
University Press .
Martin-Jones , Marilyn , and Kathryn Jones . 2000 . Multilingual
Literacies: Reading and Writing Diferent Worlds . Amsterdam and
Philadelphia : John Benjamins .
McLuhan , Marshall . 1964 . Understanding Media: Te Extension of Man .
New York : McGraw-Hill .
Mller , Karin . 1990 . Schreibe, wie du sprichst!: Eine Maxime im
Spannungsfeld von Mndlichkeit und Schriftlichkeit. Frankfurt am
Main and Bern : Peter Lang .
Negroponte , Nicholas . 1995 . Being Digital . New York : Alfred A. Knopf .
A cyberspace extension is available online at: http://archives.obs-us.
com/obs/english/books/nn/bdintro.htm.
Nissen , Hans J. 1988 . Te Early History of the Ancient Near East . Chicago
and London : Te University of Chicago Press .
Nissen , Hans J. , Peter Damerow , and Robert K. Englund . 1990 . Frhe
Schrift und Techniken der Wirtschaftsverwaltung im alten Vorderen
Orient: Informationsspeicherung und verarbeitung vor 5000 Jahren .
N.p. : Verlag Franzbecker .
Nystrand , Martin . 1986 . Te Structure of Written Communication .
Orlando, FL : Academic Press .
Olson , David R. 1977 . From utterance to text: Te bias of language in
speech and writing . Harvard Educational Review 47.3 : 257 86.
. 1991 . Literacy and Orality . Cambridge : Cambridge University
Press .
. 1994 . Te World on Paper . Cambridge : Cambridge University
Press .
forever informs the perception of language in literate society.
And by making language visible and permanent, it enables and
compels its users to accumulate information far beyond the
capacity of human memory, engendering ever new challenges
for storage and organization .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Assmann , Jan . 1991 . Stein und Zeit: Mensch und Gesellschaft im alten
gypten . Munich : Wilhelm Fink Verlag .
Backhaus , Peter . 2007 . Linguistic Landscapes: A Comparative Study of
Urban Multilingualism in Tokyo . Clevedon, UK : Multilingual Matters .
Blanche-Benveniste , Claire . 1994 . Te construct of oral and written
language. In Functional Literacy: Teoretical Issues and Educational
Applications , ed. Ludo Verhoeven , 6174. Amsterdam and
Philadelphia : John Benjamins .
Bourdieu , Pierre . 1982 . Ce que parler veut dire . Paris : Fayard .
Burns , Alfred . 1981 . Athenian literacy in the ffth century B.C . Journal of
the History of Ideas 42 .3: 371 87.
Civil , Miguel . 1995 . Ancient Mesopotamian lexicography. In
Civilizations of the Ancient Near East , ed. Jack M. Sasson , 230514. New
York : Charles Scribners Sons .
Cook , Vivian , and Benedetta Bassetti , eds. 2005 . Second Language Writing
Systems . Clevedon, UK : Multilingual Matters .
Coulmas , Florian . 1994 . Writing systems and literacy: Te alpha-
betic myth revisited. In Functional Literacy: Teoretical Issues and
Educational Implications , ed. Ludo Verhoeven , 30520. Amsterdam
and Philadelphia : John Benjamins .
. 1996 . Te Blackwell Encyclopaedia of Writing Systems .
Oxford : Blackwell .
. 2003 . Writing Systems: An Introduction to Teir Linguistic Analysis .
Cambridge : Cambridge University Press .
. 2005 . Sociolinguistics: Te Study of Speakers Choices .
Cambridge : Cambridge University Press .
Crystal , David . 2001 . Language and the Internet . Cambridge : Cambridge
University Press .
Daswani , C. J. , ed. 2001 . Language Education in Multilingual India . New
Delhi : UNESCO .
De Biasi , Pierre-Marc . 1999 . Le papier: Une aventure au quotidien .
Paris : Gallimard .
Eisenstein , Elizabeth L . 1979 . Te Printing Press as an Agent of Change .
Cambridge : Cambridge University Press .
Febvre , Lucien , and Henri-Jean Martin . [1958] 1999 . Lapparition du
livre . Paris : Albin Michel (Bibliothque Evolution Humanit) .
Ferguson , Charles A. 1959 . Diglossia . Word 15 : 325 40.
Fischer , Herv . 2006 . Digital Shock: Confronting the New Reality .
Montreal : McGill-Queens University Press .
Gelb , I. J. 1963 . A Study of Writing . Chicago and London : University of
Chicago Press .
Goody , Jack . 1977 . Te Domestication of the Savage Mind . Cambridge :
Cambridge University Press .
Goody , Jack , and Ian Watt . 1968 . Te consequences of literacy. In
Literacy in Traditional Societies , ed. Jack Goody , 2768. Cambridge :
Cambridge University Press .
Gurin-Pace , F. , and A. Blum. 1999 . Lillusion comparative: Les logiques
dlaboration et dutilisation dune enqute internationale sur
lillettrisme . Population 54 : 271 302.
Harris , Roy . 1980 . Te Language Makers . Ithaca, NY : Cornell University
Press .
. 2000 . Rethinking Writing . London : Athlon Press .
Harris , William V. 1989 . Ancient Literacy . Cambridge : Harvard University
Press .
Social Practices of Speech and Writing
45
Taylor , Insup , and M. Martin Taylor . 1995 . Writing and Literacy in
Chinese, Korean and Japanese . Amsterdam and Philadelphia : John
Benjamins .
Tomas , Rosalind . 1992 . Literacy and Ortality in Ancient Greece .
Cambridge : Cambridge University Press .
Twitchett , Denis C. 1983 . Printing and Publishing in Medieval China .
London : Te Wynkyn de Worde Society .
UNESCO Institute for Statistics . 2006 . Available online at: http://portal.
unesco.org/education/en/ev.php-URL_ID=40338&URL_DO=DO_
TOPIC&URL_SECTION=201.html.
Unseth , Peter . 2005 . Sociolinguistic parallels between choosing scripts
and languages . Written Language and Literacy 8 .1: 19 42.
Varghese , N. V. 2000 . Costing of total literacy campaigns in India. In
Adult Education in India , ed. C. J. Daswani and S. Y. Shah , 22750. New
Delhi : UNESCO .
Weber , Max . 1922 . Wirtschaft und Gesellschaft: Grundriss der verstehen-
den Soziologie . Tbingen : J. C. B. Mohr .
Wiley , T. G. 1996 . Literacy and Language Diversity in the United
States . Washington, DC : Center of Applied Linguistics and Delta
Systems .
Ong , Walter J. 1982 . Orality and Literacy: Te Technologizing of the Word .
London : Methuen .
Parkinson , Richard . 1999 . Cracking Codes: Te Rosetta Stone and
Decipherment . London : British Museum Press .
Posener , Georges . 1956 . Littrature et politique dans lgypte de la XIIe
dynastie . Paris : Honor Champion (Bibliothque lcole des hautes
tudes. 3007.)
Richards , Cameron . 2000 . Hypermedia, Internet communication, and
the challenge of redefning literacy in the electronic age . Language
Learning and Technology 4.2 : 59 77.
Roberts , Colin H. , and T. C. Skeat . 1983 . Te Birth of the Codex .
Oxford : Oxford University Press .
Sassoon , Rosemary . 1995 . Te Acquisition of a Second Writing System .
Oxford : Intellect .
Schifman , Harold . 1996 . Linguistic Culture and Language Policy . London
and New York : Routledge .
Schmandt-Besserat , Denise . 1992 . Before Writing. Austin: Te University
of Texas Press.
Stratton , Jon . 1980. Writing and the concept of law in Ancient Greece .
Visible Language 14.2 : 99 121.
46
systems of communication (semaphores, bee language, etc.),
but the study of the biology of language might shed some light
on the case of human language communication. In this essay, I
consider some diverging ideas about the role of communication
in driving language evolution .
Te study of the biology of language (see biolinguistics ) is
traditionally divided into the investigation of three questions: 1)
What is knowledge of language? 2) How does language develop in
the child? 3) And how does language evolve in the species? (See
Chomsky 1980 , 2006 ; Jenkins 2000 , 2004 .) Note that the study of
the neuroscience of language cross-cuts with all three questions.
Tat is, we can ask: 1) What are the neurological underpinnings
of the faculty of language? 2) How does the language faculty
develop in the nervous system of the individual? 3) How did the
language faculty evolve in the species? (See Chomsky and Lasnik
1995 .)
Tese three questions are sometimes referred to as the what
and how questions of biolinguistics. Tere is another question,
the why question, which is more dif cult to answer. Tis is the
question of why the principles of language are what they are
(Chomsky 2004 ). Investigations into these why questions has in
recent years been termed the minimalist program (or minimal-
ism ), but interest in and discussion of these questions go back to
the earliest days of biolinguistics (Chomsky 1995 ; Boeckx 2006 ).
Properties of the attained language derive from three factors
(Chomsky 2005b ): 1) genetic endowment for language, 2) expe-
rience, and 3) principles not specifc to the faculty of language.
Principles in 3) might even be non-domain-specifc or non-
organism-specifc principles. Examples of such principles are
principles of ef cient computation. Note that similar questions
can be posed about any biological system viruses, protein fold-
ing, bacterial cell division, sunfowers, bumblebees, falling cats,
nervous systems, and so on .
Furthermore, similar kinds of questions arise in every linguis-
tics research program (Chomsky 2005b ). To make the discussion
manageable, in what follows I draw examples and discussion
from minimalism. However, the issues and problems carry over
to all other research programs concerned with providing expla-
nations for properties of human language and accounting for
them in terms of neurobiology, genetics , and evolution. For
example, any theory of the language faculty will generate inf-
nitely many expressions that provide instructions to the senso-
rimotor and semantic interfaces. All such generative systems
will have an operation that combines structures (e.g., in mini-
malism, merge ), such as the phrase the boy with the phrase saw
the cat , and so on, formed from the lexical items of the language.
Applying this operation over and over ( unbounded Merge ), we get
a discrete infnity of expressions, part of property 1), our genetic
endowment, in this particular case of the genetic component of
the language faculty .
Many well-known accounts of the evolution of language pro-
pose communication as the primary selective force involved
in the origin of language (see Works Cited). Here, for purposes
of comparison, we present a noncommunicative account of
the origins of language, suggested by Chomsky and based on
work in minimalism. However, we stress again that this kind of
account is also compatible with approaches based on other lin-
guistic research programs. We then discuss another viewpoint
5
EXPLAINING LANGUAGE: NEUROSCIENCE,
GENETICS, AND EVOLUTION
Lyle Jenkins
Before undertaking a discussion of the explanation of language ,
we should point out that we are using this word in a special sense.
As Noam Chomsky has noted (see Jenkins 2000 ), while we can-
not ask serious questions about general notions like vision or
language, we can ask them of specifc systems like insect vision
or human language. In what follows, our focus is on the biologi-
cal foundations of language. As a result, certain areas commonly
referred to as languages are excluded from consideration, for
example, some invented logical systems, computer languages
such as Java, encrypted languages, the language of DNA, and
so on. Tese are all important and interesting areas of investi-
gation. In fact, signifcant insights into human language may
be gained from the study of some of these felds. For example,
it has been argued that particular systems of logical form
may shed light on the semantics of human language. Both bio-
logical factors and nonbiological factors interact in such areas as
pragmatics and sociolinguistics . In addition, the study of
formal languages (e.g., the Chomsky hierarchy) has also led
to some important contributions. However, these areas cannot
be completely accounted for by a consideration of the biology of
human language. It is important to keep in mind that an account
of cognitive systems like human language (as well as systems of
animal communication ) often exhibit a signifcant degree
of modularity . In this view, biological factors interact with
other factors to provide a unifed explanation. Sometimes the
term i-language is used to distinguish the biological object in
the mind-brain that biolinguists study from other uses of word
language .
Te same is true of other areas of study that are even more
closely related to human language, such as the role of language
in poems and novels, or the infuence of economic factors and
conquests and invasions on language change. historical lin-
guistics may involve factors both within and outside of the
scope of human language biology . Again, although analysis of
the biology of human language may shed some light on these
areas (e.g., the study of phonetics and phonology may be
useful in the analysis of poetry), in general it will not provide an
exhaustive account of these areas. Similarly, there has been great
interest in language as a system of communication . For the
reasons discussed here, there is not much to say about arbitrary
Explaining Language
47
directions here (but see the discussion of population dynamics
in a later secion; Nowak and Komarova 2001 ) .
A number of these accounts attempt to interconnect the evo-
lution of manual gestures, sign language, spoken language, and
motor control in various ways. Some of this work is based on the
discovery of a system of mirror neurons ( mirror systems,
imitation, and language ) (Gallese, Fadiga, et al. 1996 ). Tis
and later work demonstrated the existence of neurons in area F5
of the premotor cortex of the monkey, which are activated when
the monkey executes an action, for example, grasping an object,
and also when the monkey observes and recognizes the action
carried out by another monkey, or even the experimenter. In
addition, a subset of audiovisual mirror neurons were discov-
ered that are activated when the sound of an action is perceived,
for example, the tearing of paper (Kohler et al. 2002 ). In addi-
tion to hand mirror neurons, communicative mouth mirror
neurons were discovered that were activated both for ingestive
actions and for mouth actions with communicative content,
such as lip smacking in the monkey (Ferrari et al. 2003 ) .
Since it had been suggested on the basis of cytoarchitectonic
studies that there was a homology between area F5 of the mon-
key and area 44 (in brocas area ) of the human brain (Petrides
and Pandya 1994 ), researchers looked for and found mirror
neuron systems in humans, using fMRI (Iacoboni et al. 1999 )
and event-related magnetoencephalography (MEG) (Nishitani
and Hari 2000 ) in place of single neuron studies. Mirror neu-
rons discharge whether the action is executed, observed, or
heard. Moreover, they even discharge in the human system
when subjects are exposed to syntactic structures that describe
goal-directed actions (Tettamanti et al. 2005 ). In an fMRI study,
17 Italian speakers were asked to listen to sentences describing
actions performed with the mouth (I bite an apple), with the
hand (I grasp a knife), and with the leg (I kick the ball). In
addition, they were presented a control sentence with abstract
content (I appreciate sincerity). In the case of the action-
related words, the left-hemispheric fronto-parietaltemporal
network containing the pars opercularis of the inferior frontal
gyrus (Brocas area) was activated. Other areas were diferen-
tially activated, depending on the body part. Tey conclude that
the experiment showed that the role of Brocas area was in the
access to abstract action representations, rather than in syntactic
processing per se (277) .
On the basis of these outlined fndings, it has been sug-
gested that speech may have evolved from gesture rather than
from vocal communication by utilizing the mirror neuron sys-
tem (Rizzolatti and Arbib 1998 ; Gentilucci and Corballis 2006 ;
Fogassi and Ferrari 2007 ). Leonardo Fogassi and Pier Francesco
Ferrari note that the motor theory of speech perception fts well
with an account in terms of mirror neurons in that the objects
of speech perception are the intended phonetic gestures of the
speaker, represented in the brain as invariant motor commands
(Liberman and Mattingly 1985 , 2) . However, Fogassi and Ferrari
note that, even if the mirror neuron system is involved in speech,
the currently available evidence does not appear to support a
dedicated mirror-neuron system for language in humans.
Additional accounts of the evolution of language may also
be found in Maynard Smith and Szathmary (1998) , Lieberman
(2006) , and Christiansen and Kirby (2003) . See also adaptation ;
based on the idea that language evolved from gestures, an idea
that is represented by a number of approaches to the evolution
of language. After that, we present an example of the compara-
tive approach to evolution in biolinguistics and discuss language
and the neurosciences, focusing on leftright asymmetries of the
language areas to illustrate this research. Finally, we discuss the
genetics of language, using the studies of the FOXP2 gene system
to show how studies of language phenotype, neurobiology, and
molecular biology, as well as comparative and evolutionary stud-
ies with other animal systems, are being carried out.
Work on the principles of ef cient computation governing the
application of the operation of Merge seem to suggest an asym-
metry, namely, that the computational principles are optimized
for the semantic interface, not the sensorimotor interface. Tis is
because conditions of computational ef ciency and ease of com-
munication confict, as is familiar from the theory of language
parsing. Tis in turn has led Chomsky to suggest that external-
ization of language, and hence communication, was a secondary
adaptation of language . Tis would mean that language arose
primarily as an internal language of thought . Supporting
this idea is the existence of sign languages , which develop in
a diferent modality but in other respects are very similar to spo-
ken language (Kegl 2004 ; Pettito 2005 ) .
Tese design features of language have led Chomsky to pro-
pose the following scenario for the evolution of language. Several
decades ago Chomsky suggested, on the basis of results from non-
human primate studies, that the higher apes might well have a
conceptual system with a system of object reference and notions
such as agent, goal, instrument, and so on. What they lack, how-
ever, is the central design feature of human language, namely,
the capacity to deal with discrete infnities through recursive
rules (Chomsky 2004 , 47). He proposed that when you link that
capacity to the conceptual system of the other primates, you get
human language, which provides the capacity for thought, plan-
ning, evaluation and so on, over an unbounded range, and then
you have a totally new organism (Chomsky 2004 , 48) .
Tus, let us assume that some individual in the lineage to mod-
ern humans underwent a genetic change such that some neural
circuit(s) were reorganized to support the capacity for recursion.
Tis in turn provided the capacity for thought and so on over an
unbounded range. Tis in itself provided that individual and any
ofspring with a selective advantage that then spread through the
group. Tus, the earliest stage of language would have been just
that: a language of thought, used internally (Chomsky 2005a , 6).
At some later stage, there was an advantage to externalization, so
that the capacity would be linked as a secondary process to the
sensorimotor system for externalization and interaction, includ-
ing communication (7).
Te evolutionary scenario just outlined is derived from design
principles suggested from work on human languages. Many other
evolutionary scenarios have been proposed that assume that
communication or other social factors played a more primary
role. Tese include accounts involving manual and facial ges-
tures (Corballis 2002 ), protolanguage (Bickerton 1996 ), groom-
ing (Dunbar 1998 ), and action and motor control (Rizzolatti
and Arbib 1998 ). However, the two kinds of accounts are not
incompatible and may represent diferent aspects or stages in
the evolution of language. We cannot review all of these research
The Cambridge Encyclopedia of the Language Sciences
48
a rattle and B to a warble. As for possible reasons why the
starlings succeeded with the context-free task while the tamarins
failed, and for alternative explanations of the learning results, see
additional discussion in Fitch and Hauser (2004) , M. Liberman
(2004) , Gentner et al. (2006) , and Hauser et al. (2007 ) .
LANGUAGE AND THE NEUROSCIENCES
Turning to the neurosciences (see questions 13), with the
advent of more tools both experimental, such as imaging and
the methods of comparative genomics, and theoretical, such
as computational theories of linguistics we can look forward
to more informative models of language and the brain. David
Poeppel and Gregory Hickok ( 2004 ) note three problems with
classical models of language (involving such concepts as Brocas
area and wernickes area ): 1) Tey are inadequate for aphasic
symptomatology, 2) they are based on an impoverished linguis-
tic model, and 3) there are anatomical problems.
As for aphasic symptomatology, the classical models do not
explain certain subtypes of aphasia , like anomic aphasia. Also,
clusters of aphasic symptoms are highly variable and dissocia-
ble, indicating that there is a more complex architecture under-
lying the syndromes.
As for the second problem, there has been a tendency of some
implementations of classical models to incorporate a monolithic
picture of linguistic components, for example, production versus
comprehension , or semantics versus syntax , without regard for
fner computational subdivisions.
And fnally, certain anatomical problems came into view. For
example, Brocas aphasia and Wernickes aphasia did not always
correspond to damage in the areas that bore their names. In addi-
tion, these classical areas were not always anatomically or func-
tionally homogenous. In addition, areas outside these regions
were found to be implicated in language processing, including,
for example, the anterior superior temporal lobe , the middle
temporal gyrus (MTG), the temporo- parietal junction, the
basal ganglia , and many right hemisphere homologs
(Poeppel and Hickok 2004 , 5) . As noted, in the last few decades
many new approaches and tools have been developed in neuro-
linguistics and neurology, genetics and molecular neurobiology
(examples follow), and these have helped to overcome the kinds
of issues pointed out by Poeppel and Hickok .
For a review of some of the attempts to develop a new func-
tional anatomy of language, see the essays published along
with Poeppel and Hickock ( 2004 ) in Cognition 92 (2004). For
new approaches to the study of Brocas area, see Grodzinsky and
Amunts (2006) .
LEFTRIGHT ASYMMETRIES OF THE LANGUAGE AREAS
Ever since Paul Pierre Brocas and Carl Wernickes seminal dis-
coveries of areas involved in language processing, questions
about asymmetries in the brain has been a lively area of research.
As early as 1892, Daniel J. Cunningham reports that he found
the left sylvian fssure longer in the chimpanzee and macaque.
Cunningham, in turn, cites earlier work by Oskar Eberstaller on
the Sylvian fssure in humans, who had concluded that it was
longer in the left hemisphere than in the right (on average). He
grooming, gossip, and language; mirror systems, imi-
tation, and language; evolutionary psychology;
morphology, evolution and; origins of language;
phonology, evolution of; pragmatics, evolution and;
primate vocalizations; semantics, evolution and;
speech anatomy, evolution of; syntax, evolution of;
verbal art, evolution and .
THE COMPARATIVE APPROACH IN BIOLINGUISTICS
Troughout the modern era of biolinguistics, a question that has
been much debated is to what degree the faculty of language
is uniquely human. Marc D. Hauser and colleages ( 2002 ) have
stressed the importance of the comparative approach in the
study of this question. Early on, Chomsky, in making a case for
the genetic determination of language, used arguments from
animal behavior (ethology) to note the similarities in learning
between birdsong and animals, in particular, rapidity of learn-
ing, underdetermination of data, and so on (Chomsky 1959 ).
Hauser, Chomsky, and their colleagues have emphasized a
number of methodological points concerning the comparative
approach, using a distinction between the faculty of language
in the broad sense (FLB) and the faculty of language in the nar-
row sense (FLN). Te basic idea is that before concluding that
some property of language is uniquely human, one should study
a wide variety of species with a wide variety of methods. And
before concluding that some property of language is unique to
language, one should consider the possibility that the property
is present in some other (cognitive) domain, for example, music
or mathematics . Tey tentatively conclude that recursion may
be a property of language that is unique to language and, if so,
belongs to the faculty of language in the narrow sense .
An example of the application of the comparative method is
the investigation of the computational abilities of nonhuman pri-
mates by W. Tecumseh Fitch and Hauser ( 2004 ), who tested the
ability of cotton-top tamarins, a New World primate species, as
well as human controls, to process diferent kinds of grammars.
Using a familiarization/discrimination paradigm, they found
that the tamarins were able to spontaneously process fnite-state
grammars, which generate strings of syllables of the form (AB)
n
,
such as ABAB, ABABAB. However, they were unable to process
context-free grammars, which generate strings of syllables of the
form A
n
B
n
, such as AABB, AAABBB. It is known from formal lan-
guage theory (the Chomsky hierarchy) that context-free gram-
mars are more powerful than fnite-state grammars. Moreover,
the humans tested were able to rapidly learn either grammar.
Te authors conclude that the acquisition of hierarchical pro-
cessing ability, that is, the ability to learn context-free gram-
mars, may have represented a critical juncture in the evolution
of the human language faculty (380) .
In a later study, Timothy Q. Gentner and colleagues ( 2006 )
showed that European starlings, in contrast to tamarins, can
recognize acoustic patterns generated by context-free gram-
mars. Using an operant conditioning paradigm, they found that
9 of 11 starlings were able to learn both fnite-state grammar and
context-free grammar sequences accurately. Te (AB)
n
and A
n
B
n
sequences in this case were made up of acoustic units (motifs)
from the song of the starlings. In this case, A corresponded to
Explaining Language
49
in a region of the planum temporale in human, chimpanzee,
and rhesus monkey brains. It was found that only human brains
exhibited asymmetries in minicolumn morphology, in particu-
lar, wider columns and more neuropil space (Buxhoeveden et al.
2001 ).
It is possible that circuits could be reorganized within a lan-
guage region without a signifcant volumetric change so that a
novel function in language could evolve. Sherwood and col-
leagues conclude: Terefore, it is likely that Brodmanns area
44 homolog in great apes, while similar in basic structure to that
in humans, difers in subtle aspects of connectivity and lacks
homologous function (284).
Allen Braun ( 2003 ) notes that MRI could still turn out to be
useful for the study of microstructure at higher feld strengths,
with the addition of MR contrast agents, and with the use of
difusion-weighted MR methods. He also notes that the pars
orbitalis has often been arbitrarily excluded from the defnition
of Brocas area, and might be important in the search for ante-
cedents of language in nonhuman primates. In particular, some
studies suggest that the pars orbitalis is selectively activated by
semantic processes (as opposed to phonological or syntactic
processes) (Bookheimer 2002 ).
It is known that nonhuman primates have structures homolo-
gous to the perisylvian areas involved in human language, that
is, support both expressive and receptive language (Galaburda
and Pandya 1983 ; Deacon 1989 ). Ricardo Gil-da-Costa and col-
leagues ( 2006 ) presented species-specifc vocalizations in rhesus
macaques and found that the vocalizations produced distinct
patterns of brain activity in areas homologous to the perisyl-
vian language areas in humans using H
2
15
O positron emission
tomography (PET). Two classes of auditory stimuli were pre-
sented to the monkeys. One was species-specifc macaque
vocalizations (coos and screams). As a control, nonbiologi-
cal sounds were presented that matched the species-specifc
vocalizations in frequency, rate, scale, and duration. Tey found,
for example, a greater response to species-specifc calls than to
nonbiological sounds in the perisylvian system with homologs
in humans, for example, to the area Tpt of the temporal planum
and to the anterior perisylvian cortex, roughly corresponding to
the areas studied by Wernicke and Broca in humans. However,
they did not fnd any clear lateralization efects in the macaque
brain comparable to the anatomical and functional asymmetries
documented in humans and anatomical asymmetries in apes
(Gannon et al. 1998 ) .
Gil-da-Costa and colleagues ( 2006 ) note that the perisylvian
regions are not performing linguistic computations in the
macaque, but could be performing a prelinguistic function in
associating the sound and meaning of species-specifc vocal-
izations. Furthermore, this would position the perisylvian sys-
tem to be recruited for use during the evolution of language.
More specifcally, it may have been exapted during the emer-
gence of more complex neural mechanisms that couple sound
and meaning in human language (2006, 1070). Although I have
focused here on the perisylvian system, it should be emphasized
that areas outside this system have also been demonstrated to be
involved in language .
K. A. Shapiro and colleagues ( 2006 ) provide another example
of the application of imaging studies to investigate how linguistic
postulated that this region held the key to what he called the sen-
sible Sprachcentrum (sensible/cognizant language center).
Claudio Cantalupo and William D. Hopkins ( 2001 ) report
fnding an anatomical asymmetry in Brocas area in three great
ape species. Tey obtained magnetic resonance images (MRI)
( n euroimaging ) from 20 chimpanzees ( P. troglodytes ), 5
bonobos ( P. paniscus ), and 2 gorillas ( G. gorilla ). In humans,
Brodmanns area 44 corresponds to part of Brocas area within
the inferior frontal gyrus (IFG). Tis area is larger in the left
hemisphere than in the right. Furthermore, it is known that
Brocas area is vital for speech production (with the qualifca-
tions discussed earlier).
Although the great apes were known to have a homolog of
area 44 on the basis of cytoarchitectonic and electrical stimu-
lation studies, no leftright anatomical asymmetry had been
shown. Cantalupo and Hopkins found a pattern of morpholog-
ical asymmetry similar to that found in the homologous area in
humans. Tis would place the origin of the asymmetry for the
anatomical substrate for speech production to at least fve mil-
lion years ago.
Since the great apes exhibit only primitive vocalizations, these
authors speculate that this area might have subserved a gestural
system (see earlier discussion). Tey note the presence in mon-
keys of mirror neurons in area 44 that subserve the imitation of
hand grasping and manipulation (Rizzolatti and Arbib 1998 ).
Tey also observe that captive great apes have a greater right-
hand bias when gesturing is accompanied by vocalization. Hence
in the great apes, the asymmetry may have subserved the pro-
duction of gestures accompanied by vocalizations, whereas, for
humans, this ability was selected for the development of speech
systems, accompanied by the expansion of Brodmanns area 45
(which, along with Brodmanns area 44, makes up Brocas area)
(Cantalupo and Hopkins 2001 , 505).
However, additional studies of Brodmanns area 44 in
African great apes ( P. troglodytes and G. gorilla ) call into ques-
tion whether the techniques used in Cantalupo and Hopkins
study were suf cient to demonstrate the leftright asymmetry.
Chet C. Sherwood and colleagues ( 2003 ) found considerable
variation in the distribution of the inferior frontal sulci among
great ape brains. Tey also constructed cytoarchitectural maps of
Brodmanns area 44, examining myeloarchitecture and immu-
nohistochemical staining patterns. When they studied the IFG
of great ape brains, they found a poor correspondence between
the borders observed in the cytoarchitectural maps and the bor-
ders in the surface anatomy (e.g., sulcal landmarks). Tere were
similar fndings for human brains in an earlier study (Amunts et
al. 1999 ). Sherwood and colleagues conclude that in the study
by Cantalupo and Hopkins, it is unlikely that the sulci used to
defne the pars opercularis coincided with the borders of cyto-
architectural area 44 (2003, 284). In general then, macrostruc-
ture is a poor predictor of microstructure.
Sherwood and colleagues also point out that even if human-
like asymmetries of the inferior frontal gyrus and of the planum
temporale are confrmed, these gross asymmetries will not suf-
fce to explain the unique neural wiring that supports human
language (284). To that end, comparative studies of microstruc-
ture in humans and great apes are needed. For example, a com-
puterized imaging program was used to examine minicolumns
The Cambridge Encyclopedia of the Language Sciences
50
dyslexia, specific language impairment, and autism
and language ). By understanding how genetic changes can
cause the operation of language to break down, we get an idea of
the genes that are important for language acquisition, process-
ing, and, ultimately, language evolution.
An autosomal-dominant and monogenic disorder of speech
and language was found in the KE family with a 3-generation
pedigree. A monogenic disorder involves a mutation in a single
gene, and here the individuals have one copy of the mutant gene
and one normal gene on two autosomal (non-sex) chromosomes.
Te disorder was mapped to the region 7q31 on chromosome 7,
and it was shown that the gene (called FOXP2 ) had a mutation
in a forkhead-domain of the protein it encoded in the afected
family (Lai et al. 2001 ).
Te individuals were diagnosed as having developmental ver-
bal dyspraxia . Te phenotype was found to be quite complex,
afecting orofacial sequencing, articulation, grammar, and cog-
nition and is still incompletely understood and under investiga-
tion (see also genes and language ; Jenkins 2000 ) .
Te FOXP2 gene was found to code for a transcription factor,
that is, a protein that regulates gene expression by turning a gene
on or of or otherwise modulating its activity. It is natural to ask
what other genes FOXP2 may regulate and how it regulates these
genes (turning them on or of, for example), as well as to deter-
mine whether any of these genes downstream are involved in
speech and language in a more direct way.
To fnd the gene targets of FOXP2 in the brain and to deter-
mine the efects on those genes, methods are being developed to
identify these neural targets both in vitro and in vivo (Geschwind
and Miller 2001 ). Te laboratory of D. H. Geschwind has devel-
oped a genomic screening approach that combines 1) chromatin
immunoprecipitation and 2) microarray analysis (ChIP-Chip).
In chromatin immunoprecipitation, an antibody that recognizes
the protein of interest (e.g., foxp) is used to fsh out a protein-
DNA complex. Te DNA is then hybridized to arrays with DNA
from thousands of human genes. Tis allows the identifcation of
binding sites for transcription factors (in this case, FOXP2). Te
goal is to discover potential gene candidates involved in the devel-
opment of neural circuits supporting speech and language.
Te homologue to the human FOXP2 gene has been discov-
ered in a number of diferent species, including mice ( Foxp2)
and songbirds, such as the zebra fnch (FoxP2). Whatever ones
views on the relationship between human language and other
animal communication systems, it is important to study the evo-
lutionary origin of genes that afect language, such as FOXP2 , for
one can learn about the neural pathways constructed by these
genes, which might not otherwise be possible in experiments
with humans.
It has been found that the zebra fnch and human protein
sequence is 100 percent identical within the DNA-binding
domain, suggesting a possible shared function (White et al. 2006 ;
Haesler et al. 2004 ). In addition, Constance Scharf and Sebastian
Haesler ( 2005 ) report that the FoxP2 pattern of expression in the
brain of birds that learn songs by imitation resembles that found
in rodents and humans. In particular, FoxP2 is expressed in the
same cell types, for example, striatal medium spiny neurons.
Moreover, FoxP2 is expressed both in the embryo and in the
adult. To fnd out whether FoxP2 is required for song behavior in
categories like nouns, verbs, and adjectives are organized in the
brain. An event-related functional MRI imaging study has found
specifc brain sites that are activated by either nouns or verbs,
but not both. In a series of experiments, subjects were asked to
produce nouns and verbs in short phrases as real words ( the
ducks , he plays ), as well as pseudowords (the wugs , he zibs ), both
with regular infections and irregular infections ( geese , wrote ),
including both concrete and abstract words. Specifc brain
areas were selectively activated for either verb production (left
prefrontal cortex and left superior parietal lobule) or for noun
production (left anterior fusiform gyrus) across the entire bat-
tery of tests. Moreover, the areas were nonoverlapping, leading
the authors to conclude that these regions are involved in rep-
resenting core conceptual properties of nouns and verbs (2006,
1644).
In recent years, it has become possible to study asymmetries
on a molecular level as well (Sun and Walsh 2006 ). As discussed
earlier, there are functional, anatomical, and cytoarchitectonic
diferences between the left and right cerebral hemispheres in
humans. To determine what the molecular basis for these asym-
metries might be, Tao Sun and colleagues ( 2005 ) compared left
and right embryonic cerebral hemispheres for leftright difer-
ences in gene expression, using serial analysis of gene expres-
sion (SAGE). Tey discovered 27 genes whose transcriptions
were diferentially expressed on the left and right sides. In par-
ticular, the gene LMO4 , which asymmetrically expressed the Lim
Domain Only 4 transcription factor, is more highly expressed in
the perisylvian regions of the right hemisphere than in the left
at 12 weeks and 14 weeks. Further studies are needed to deter-
mine how LMO4 expression is regulated by factors still earlier in
development .
Mouse cortex was also examined, and it was found that
although Lmo4 expression was moderately asymmetrical in every
individual mouse brain, the expression was not consistently lat-
eralized to either the left or the right. Tis may be related to the
fact that asymmetries like paw preference are seen in individual
mice but are not biased in the population as a whole, as hand
preference is in humans.
Te results of this study are also consistent with the observa-
tion that the genes involved in visceral asymmetries (e.g., of the
heart) are not measurably implicated in cerebral asymmetries. It
had been noted earlier that situs inversus mutations in humans
do not appear to interfere with the left-hemisphere localization of
language and handedness (Kennedy et al. 1999 ). In these earlier
studies, it had been found that the pattern of language laterali-
zation in patients with situs inversus was identical to that found
in 95 percent of right-handed individuals with normal situs. It
was concluded that the pathway afecting language dominance
and handedness was most likely distinct from that afecting the
asymmetry of the visceral organs .
GENETICS AND SPEECH DISORDERS
Biolinguists would like to understand the wiring of networks
underlying language function at the level of genes. We have seen
that one way to study this question is to use such diferential gene
expression methods as SAGE. Another key way of investigating
the genetics of language is by studying language disorders (see
Explaining Language
51
one of the researchers on the study, noted that, in contrast,
Patient 1 had normal spatial ability but could form next to no
complete words. When asked what animal has long ears and eats
carrots, he could only pronounce the r in the word rabbit but was
able to draw the letter on the blackboard and add features such
as whiskers (McMaster 2005 ) .
Te duplicated region on chromosome 7 contains around 27
genes, but it is not yet known which of the duplicate gene copies
are involved in the expressive language delay, although certain
genes have been ruled out.
A gene ( SRPX2 ) responsible for rolandic seizures that are
associated with oral and speech dyspraxia and mental retarda-
tion has been identifed (Roll et al. 2006 ). It is located on Xq22.
Te SPRX2 protein is expressed and secreted from neurons in the
human adult brain, including the rolandic area. Tis study char-
acterizes two diferent mutations.
Te frst mutation was found in a patient with oro-facial dys-
praxia and severe speech delay. Te second mutation was found
in a male patient with rolandic seizures and bilateral perisylvian
polymicrogyria. Te authors also note that Sprx2 is not expressed
during murine embryogenesis, suggesting that SPRX2 might play
a specifc role in human brain development, particularly of the
rolandic and sylvian areas .
RECOVERING ANCIENT DNA
In addition to classical studies of fossils, there is currently
renewed interest in work on ancient DNA. Te Neanderthals
( Homo neanderthalensis ) were an extinct group of hominids that
are most closely related to modern humans ( Homo sapiens ). Up
to now, information about Neanderthals has been limited to
archaeological data and a few hominid remains. Comparing the
genetic sequences of Neanderthals and currently living humans
would allow one to pinpoint genetic changes that have occurred
during the last few hundred thousand years. In particular, one
would be able to examine and compare diferences in such
genes as FOXP2 in living humans, Neanderthals, and nonhuman
primates.
Partial DNA sequences of the Neanderthal have now been
published by two groups led by Svante Pbo and Edward Rubin
(Green et al. 2006 ; Noonan et al. 2006 ). Pbos group identi-
fed a 38,000-year-old Neanderthal fossil from the Vindija cave
(the Neanderthals became extinct around 30,000 years ago). Te
fossil was free enough from contamination to permit DNA to be
extracted and subjected to large-scale parallel 454 sequencing, a
newer and faster system for sequencing DNA.
Pbos group was able to sequence and analyze about one
million base pairs of Neanderthal DNA. Note that the genomes of
modern humans and Neanderthals each have about three billion
base pairs (3 gigabases). Among the conclusions they reached
on the basis of the comparison to human DNA was that modern
human and Neanderthal DNA sequences diverged on average
about 500,000 years ago. Tey also expected to produce a draft of
the Neanderthal genome within two years. (For their preliminary
results, see Pennisi 2009.)
Rubins group obtained around 65,000 base pairs of
Neanderthal sequence. Tey used a combination of the Sanger
method of DNA sequencing and the (faster) pyrosequencing
the zebra fnch, the Scharf laboratory is using RNA to downregu-
late FoxP2 in Area A, a striatal region important for song learning
(Scharf and Nottebohm 1991 ). It is known that young male zebra
fnches express more FoxP2 bilaterally in Area X when learning
to sing (Haesler et al. 2004 ). Tey will then be able to determine
whether the bird is still able to sing normal song as well as copy
the song of an adult male tutor.
Weiguo Shu and colleagues ( 2005 ) found that disrupting the
Foxp2 gene in mice resulted in impairing their ultrasonic vocal-
ization. In addition to their more familiar sonic vocalizations,
mice also make ultrasonic sounds, for example, when they are
separated from their mothers.
In order to study the efect of disrupting the Foxp2 gene on
vocalization, these researchers constructed two versions of
knockout mice. One version had two copies of the defective
Foxp2 gene (the homozygous mice) and the other version had
one copy of the defective Foxp2 gene, as well as one gene that
functioned normally (the heterozygous mice). Te homozygous
mice (double knockout) sufered severe motor impairment,
premature death, and an absence of ultrasonic vocalizations
that are normally produced when they are separated from their
mother. On the other hand, the heterozygous mice, with a sin-
gle working copy of the gene, exhibited modest developmental
delay and produced fewer ultrasonic vocalizations than normal.
In addition, it was found that the Purkinje cells in the cerebel-
lum, responsible for fne motor control, were abnormal. It is con-
cluded that the fndings support a role for Foxp2 in cerebellar
development and in a developmental process that subsumes
social communication functions (Shu et al. 2001 , 9643).
Timothy E. Holy and Zhongsheng Guo ( 2005 ) studied the
ultrasonic vocalizations that male mice emit, when they encoun-
ter female mice or their pheromones. Tey discovered that the
vocalizations, which have frequencies ranging from 30100 kHz,
have some of the characteristics of song, for example, birdsong.
In particular, they were able to classify diferent syllable types
and found a temporal sequencing structure in the vocalizations.
In addition, individual males, though genetically identical, pro-
duced songs with characteristic syllabic and temporal structure.
Tese traits reliably distinguish them from other males. Holy
notes that these discoveries increase the attractiveness of mice
as model systems for study of vocalizations (White et al. 2006 ,
10378).
We have focused on the FOXP2 gene here because there is no
other gene afecting speech and language about which so much
information is available that bears on questions of neurology and
evolution of language. However, the search is underway for other
additional genes .
For example, genetics researchers have also discovered a
DNA duplication in a nine-year-old boy with expressive language
delay (Patient 1) (Fisher 2005 ; Somerville et al. 2005 ). Although
his comprehension of language was at the level of a seven-year-
old child, his expression of language was comparable to that of
only a two-and-a-half-year-old.
Te region of DNA duplication in Patient 1 was found to be
on chromosome 7, and interestingly, was found to be identical to
the region that is deleted in Williams-Beuren syndrome (WBS).
Patients with WBS have relatively good expressive language but
are impaired in the area of spatial construction. Lucy Osborne,
The Cambridge Encyclopedia of the Language Sciences
52
dif cult to discover the signifcance of nucleotide changes in
noncoding regions, since one cannot determine their functional
signifcance by visual inspection. Nonetheless, until this infor-
mation is available, there is no reason to favor the idea that the
two changes in the FOXP2 protein are functional. Carroll notes
that while it may be tempting to reach for the low-hanging fruit
of coding sequence changes, the task of unravelling the regula-
tory puzzle is yet to come (1165).
In fact, some data favor the regulatory sequence hypothesis.
Carroll notes that in evaluating whether FOXP2 is involved in the
evolution of the neural circuits of language, one must ask several
questions. Te frst is, is the gene product used in multiple tis-
sues? (1164).
It is known that FOXP2 appears to act as a repressor in lung
tissue (Shu et al. 2001 ). Moreover, studies in mice have revealed
that in addition to Foxp2, two other transcription factors, Foxp1
and Foxp4 are expressed in pulmonary and gut tissues (Lu et al.
2002 ). Foxp4 is also expressed in neural tissues during embry-
onic development. It is not at all surprising to fnd that a gene
that may be involved in language development is also active in
nonlanguage areas. Te reason for this is that transcription fac-
tors like FOXP2 often act in a combinatorial fashion in conjunc-
tion with other transcription factors in diferent ways in diferent
tissues.
As another example, suppose that a cell has three regula-
tory proteins and that with each cell division a new regulatory
protein becomes expressed in one of the daughter cells, but
not the other. Hence, we only need three regulatory proteins
acting in combination to produce eight diferent cell types.
Combinatorial control is thought to be widespread as a means
of eukaryotic gene regulation. For those familiar with linguistic
models of language acquisition, it may help to think of parame-
ter settings, whereby three binary parameters can specify eight
diferent language types with respect to some structural feature.
Te idea, then, is that FOXP2 can work in conjunction with cer-
tain factors in the lung to repress gene activity in the epithelium,
whereas it might work together with other factors in the brain to
regulate genes there directly (or indirectly) involved in speech
and language.
Te second question to ask is are mutations in the coding
sequence known or likely to be pleiotropic [i.e., causing multiple
efects]? (1164) It is well known that patients with the FOXP2
mutation have multiple defects in speech, orofacial praxis, lan-
guage, and cognition. Te third question to ask is does the locus
contain multiple cis -regulatory elements? (ibid.) ( Cis -elements
are regulatory elements located on the same nucleic acid strand
as the gene they regulate.) Again, since FOXP2 is expressed in
multiple areas in the brain and in other organs, this is a clear
indication that it does. Carroll concludes on this basis that reg-
ulatory sequence evolution is the more likely mode of evolution
than coding sequence evolution (ibid.).
Finally, Carroll notes that some of the data from experi-
ments with birdsong learners and nonlearners also support the
idea of regulatory evolution. When FoxP2 mRNA and protein
levels during development were studied, a signifcant increase
of FoxP2 expression was found in Area X, a center required for
vocal learning. Te increase occurred at a time when vocal learn-
ing in zebra fnches was underway. Moreover, FoxP2 expression
method. Although the Sanger method yields less amounts of
sequence than pyrosequencing, the error rate is lower. From the
sequence data they estimated that Neanderthals and humans
shared a most recent common ancestor around 706,000 years
ago. Tere has been interest in whether Neanderthals might
have contributed to the European gene pool. For example, one
study suggests that humans may have acquired the microceph-
alin gene, which regulates brain size during development, by
interbreeding with another species, possibly the Neanderthals.
However, the data here do not support this possibility, although
more sequence data will be needed to answer the admixture
question defnitively (Evans et al. 2006 ). However, they did
establish the validity of their sequencing approach, which
allows for the rapid recovery of Neanderthal sequences of
interest from multiple independent specimens, without the
need for whole-genome resequencing (Noonan et al. 2006 ,
1118) .
GENETICS AND EVOLUTION
Te work on comparative animal studies and comparative neu-
roanatomy discussed earlier are being increasingly informed by
the rapidly emerging feld of comparative genomics. Work in
the feld of evolutionary development ( evo-devo ) has provided
substantial support for the idea that gene regulation is key to
understanding evolution. Among the methods now available to
us to study the evolution of anatomy (including neural circuits)
and behavior at the genome level are comparative genomics,
gene-expression profling, and population genetics analysis. We
have already seen an example of the profling of gene expression
in the left and right cerebral hemispheres using serial analysis
of gene expression. Such methods have also been applied to
FOXP2 .
As we noted earlier, the FOXP2 gene was discovered to code
for a transcription factor and is therefore involved in regulating
the expression of other genes. In general, transcription factors
are highly conserved. In this case, the FOXP2 protein is diferent
from the chimpanzee and gorilla sequence at two amino acids,
and from the orangutan sequence at three amino acids. Te
human and mouse protein sequence difer at three amino acids.
Te question arises whether these amino acid replacements are
of functional signifcance, that is, whether they played a crucial
role in the evolution of language.
A population genetic study of the FOXP2 locus concluded
that it had been subjected to a selective sweep during the past
200,000 years, correlating closely with estimated age of Homo
sapiens . However, Sean B. Carroll ( 2005 ) notes that one cannot
immediately conclude from the fact that the FOXP2 gene has
been a target of selection during human evolution that it is the
amino acid replacements just discussed that were the function-
ally important targets. Since the FOXP2 gene is 267 kilobases
in size, we should fnd more than 2,000 diferences in DNA
sequence between chimpanzees and humans (assuming an
average base pair divergence of 1.2%). Tis means that there are
many more possibilities for functionally signifcant mutations
in noncoding regulatory areas than in coding regions (the
FOXP2 protein is made from coding regions, while the noncod-
ing regions contain the regulatory information). It is, of course,
Explaining Language
53
(1980, 216). Not only has that proven to be the case, but with the
explosion of knowledge in many areas, including (comparative)
linguistics, comparative neuroanatomy, evolutionary devel-
opment, comparative genomics, to take just a few examples,
biolinguistics promises to be a fascinating feld for decades to
come .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Amunts , K. , A. Schleicher , et al. 1999 . Brocas region revis-
ited: Cytoarchitecture and intersubject variability . Journal of
Comparative Neurology 412 : 319 41.
Bickerton , Derek . 1996 . Language and Human Behavior . Seattle : University
of Washington Press .
Boeckx , Cedric . 2006 . Linguistic Minimalism . Oxford : Oxford University
Press .
Bookheimer , Susan . 2002 . Functional MRI of language: New approaches
to understanding the cortical organization of semantic processing.
Annual Review of Neuroscience 25 (March): 151 88.
Braun , Allen . 2003 . New fndings on cortical anatomy and implications
for investigating the evolution of language . Te Anatomical Record
Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology
271A (March): 273 75.
Buxhoeveden , D. P., A. E. Switala , et al. 2001 . Lateralization of minicol-
umns in human planum temporale is absent in nonhuman primate
cortex . Brain, Behavior and Evolution 57 (June): 349 58.
Cangelosi , Angelo , and Domenico Parisi , eds. 2001 . Simulating the
Evolution of Languge . New York : Springer .
Cantalupo , Claudio , and William D. Hopkins . 2001 . Asymmetric Brocas
area in great apes . Nature 414 (November): 505.
Carroll , Sean B . 2005 . Evolution at two levels: On genes and form . PLoS
Biology 3 (July): 1159 66. Available online at: www.plosbiology.org.
Chomsky , Noam . 1959 . A review of B. F. Skinners Verbal Behavior .
Language 35.1 : 26 58.
. 1980 . On the biological basis of language capacities. In Rules
and Representations , 185216. New York : Columbia University Press .
. 1995 . Te Minimalist Program . Cambridge, MA : MIT Press .
. 2004 . Te Generative Enterprise Revisited: Discussions with Riny
Huybregts, Henk Van Riemsdijk, Naoki Fukui and Mihoko Zushi .
Berlin : Mouton de Gruyter .
. 2005 a. Some simple evo-devo theses: How true might they be for
language? Paper presented at the Morris Symposium, SUNY at Stony
Brook.
. 2005 b. Tree factors in language design . Linguistic Inquiry
36 .1: 1 22.
. 2006 . Language and Mind . Cambridge : Cambridge University
Press .
Chomsky , N. , and H. Lasnik . 1995 . Te theory of principles and param-
eters. In Te Minimalist Program , N. Chomsky , 13127. Cambridge,
MA : MIT Press .
Christiansen , Morten H. , and Simon Kirby , eds. 2003 . Language Evolution .
New York: Oxford University Press .
Corballis , Michael C. 2002 . From Hand to Mouth: Te Origins of Language .
Princeton, NJ : Princeton University Press.
Cunningham , Daniel J. 1892 . Contribution to the Surface Anatomy of the
Cerebral Hemispheres [Cunningham Memoirs]. Dublin : Royal Irish
Academy of Science .
Deacon , T. W. 1989 . Te neural circuitry underlying primate calls and
human language . Human Evolution 4 (October): 367 401.
Dunbar , Robin . 1998 . Grooming, Gossip and the Evolution of Language .
Cambridge : Harvard University Press .
Evans , Patrick D. , Nitzan Mekel-Bobrov , et al. 2006 . Evidence that the
adaptive allele of the brain size gene microcephalin introgressed
levels in adult canaries varied with the season and correlated
with changes in birdsong. Tese facts suggest regulatory control
in development and in the adult brain .
EVOLUTION AND DYNAMICAL SYSTEMS
Another tool used in the study of language evolution is the theory
of dynamical systems (see also self-organizing systems )
(Nowak 2006 ). However, applications of dynamical systems to
language include studies not only of evolution of language but
of language change and language acquisition as well (Niyogi
2004 , 2006 ). Martin A. Nowak and Natalia Komarova ( 2001 ,
292) assume for the case of language acquisition that Universal
Grammar (UG) contains two parts: 1) a rule system that gener-
ates a search space of candidate grammars and 2) a [learning]
mechanism to evaluate input sentences and to choose one of
the candidate grammars that are contained in his [the learners]
search space. One of the main questions to be determined is
what is the maximum size of the search space such that a spe-
cifc learning mechanism will converge (after a number of input
sentences, with a certain probability) to the target grammar.
Te question for language evolution then is what makes a
population of speakers converge to a coherent grammatical sys-
tem. A homogenous population (all individuals have the same
UG) is assumed. Nowak and Komarova derive a set of equations,
which they call the language dynamics equations (293), which
give the population dynamics of grammar evolution.
Te equations represent the average payof (for mutual under-
standing) for all those individuals who use a particular grammar
and contribute to biological ftness (the number of ofspring they
leave), and include a quantity to measure the accuracy of gram-
mar acquisition of the ofspring from their parents. Another
variable denotes the relative abundance of individuals who use
a particular grammar. Still another variable denotes the average
ftness or grammatical coherence of the population, the mea-
sure for successful communication in a population (293).
Nowak and Komarova use the language dynamics equations
to study the conditions under which UG will result in grammati-
cal coherence. A number of factors can be varied in order to run
computer simulations: population size, assumptions about UGs
search space, and assumptions about the learning mechanism
(e.g., memoryless or batch learning, etc.).
A similar kind of dynamical systems analysis has been pro-
posed by Partha Niyogi for language change ( 2004 , 58), for what
he calls the emerging feld of population linguistics . Such con-
cepts as symmetry and stability (stable and unstable equilibria)
are used in the study of the language dynamics equations. Niyogi
uses symmetric and asymmetric nonlinear dynamical models to
study lexical and syntactic change. Nowak and Komarova note
that dynamical systems analysis is compatible with a wide range
of diferent kinds of linguistic analysis and learning theories.
Tere are a number of other approaches to the study of language
evolution with dynamical systems and simulation, some of which
may be found in the suggestions for further reading (Christiansen
and Kirby 2003 ; Cangelosi and Parisi 2001 ; Lyon et al. 2006 ).
More than 25 years ago, Chomsky observed that the study of
the biological basis for human language capacities may prove to
be one of the most exciting frontiers of science in coming years
The Cambridge Encyclopedia of the Language Sciences
54
Kohler , Evelyne , Christian Keysers , et al. 2002 . Hearing sounds, under-
standing actions: Action representation in mirror neurons . Science
297 : 846 8.
Lai , Cecilia S. L. , Simon E. Fisher , et al. 2001 . A forkhead-domain gene
is mutated in a severe speech and language disorder . Nature 413
(October): 519 23.
Liberman , Alvin M. , and Ignatius G. Mattingly . 1985 . Te motor theory
of speech perception revised . Cognition 21 : 1 36.
Liberman , Mark . 2004 . Hi lo hi lo, Its of to formal language theory we
go. Language Log (January 17). Available online at: http://itre.cis.
upenn.edu/~myl/languagelog/archives/000355.html.
Lieberman , Philip . 2006 . Toward an Evolutionary Biology of Language .
Cambridge : Belknap Press of Harvard University Press .
Lu , Min Min , Shanru Li , et al. 2002 . Foxp4 : A novel member of the Foxp
subfamily of winged-helix genes co-expressed with Foxp1 and Foxp2
in pulmonary and gut tissues . Mechanisms of Development 119 ,
Supplement 1 (December): S197 S202.
Lyon , Caroline , Chrystopher L. Nehaniv , et al., eds. 2006 . Emergence of
Communication and Language . New York : Springer .
Maynard Smith , John , and Ers Szathmry . 1998 . Te Major Transitions
in Evolution . New York : Oxford University Press .
McMaster , Geof . 2005 . Researchers discover cause of speech defect.
University of Alberta ExpressNews (October 25) . Available online
at: http://www.expressnews.ualberta.ca.
Nishitani , N. , and R. Hari . 2000 . Temporal dynamics of cortical repre-
sentation for action . Proceedings of the National Academy of Sciences
USA 97 : 913 18.
Niyogi , Partha . 2004 . Phase transitions in language evolution. In
Variation and Universals in Biolinguistics , ed. L. Jenkins , 5774.
Amsterdam : Elsevier .
. 2006 . Te Computational Nature of Language Learning and
Evolution . Cambridge, MA : MIT Press .
Noonan , James P. , Graham Coop , et al. 2006 . Sequencing and analysis of
Neanderthal genomic DNA . Science 314 (November): 1113 18.
Nowak , Martin A . 2006 . Evolutionary Dynamics: Exploring the Equations
of Life . Cambridge : Harvard University Press .
Nowak , Martin A. , and Natalia L. Komarova . 2001 . Towards an evo-
lutionary theory of language . Trends in Cognitive Sciences 5 (July):
288 95.
Pennisi, Elizabeth. 2009. Neanderthal Genomics: Tales of a Prehistoric
Human Genome. Science 323. 5916: 86671.
Pettito , Laura-Ann . 2005 . How the brain begets language. In
Cambridge Companion to Chomsky , ed. J. McGilvray , 84101.
Cambridge : Cambridge University Press .
Petrides , M., and D. N. Pandya . 1994 . Comparative architectonic anal-
ysis of the human and the macaque frontal cortex. In Handbook of
Neuropsychology , ed. F. Boller and J. Grafman , 1758. Amsterdam :
Elsevier .
Poeppel , David , and Gregory Hickok . 2004 . Towards a new functional
anatomy of language . Cognition 92 (May/June): 1 12.
Rizzolatti , Giacomo , and Michael A. Arbib . 1998 . Language within our
grasp . Trends in Neurosciences 21 (May): 188 94.
Roll , Patrice , Gabrielle Rudolf , et al. 2006 . Srpx2 mutations in disor-
ders of language cortex and cognition . Human Molecular Genetics 15
(April): 11951207.
Scharf , Constance , and Sebastian Haesler . 2005 . An evolutionary
perspective on FoxP2: Strictly for the birds? Current Opinion in
Neurobiology 15 (December): 694 703.
Scharf , C. , and F. Nottebohm . 1991 . A comparative study of the behav-
ioral defcits following lesions of various parts of the zebra fnch song
system: Implications for vocal learning . Journal of Neuroscience 11
(September): 2896 913.
into Homo sapiens from an archaic Homo lineage . PNAS 103
(November): 1817883.
Ferrari , Pier Francesco , Vittorio Gallese , et al. 2003 . Mirror neurons
responding to the observation of ingestive and communicative mouth
actions in the monkey ventral premotor cortex . European Journal of
Neuroscience 17 : 1703 14.
Fisher , S. E . 2005 . On genes, speech, and language . New England
Journal of Medicine 353. 16: 16557.
Fitch , W. Tecumseh , and Marc D. Hauser . 2004 . Computational con-
straints on syntactic processing in a nonhuman primate . Science 303
(January): 377 80.
Fogassi , Leonardo , and Pier Francesco Ferrari . 2007 . Mirror neurons
and the evolution of embodied language . Current Directions in
Psychological Science 16 : 136 41.
Galaburda , Albert M. , and Deepak N. Pandya . 1983 . Te intrinsic archi-
tectonic and connectional organization of the superior temporal
region of the rhesus monkey . Journal of Comparative Neurology 221
(December): 169 84.
Gallese , Vittorio , Luciano Fadiga , et al. 1996 . Action recognition in the
premotor cortex . Brain 119 : 593 609.
Gannon , Patrick J. , Ralph L. Holloway , et al. 1998 . Asymmetry of chim-
panzee planum temporale: Humanlike pattern of Wernickes brain
language area homolog . Science 279 (January): 220 2.
Gentilucci , Maurizio , and Michael C. Corballis . 2006 . From manual ges-
ture to speech: A gradual transition . Neuroscience and Biobehavioral
Reviews 30: 949 60.
Gentner , Timothy Q., Kimberly M. Fenn , et al. 2006. Recursive syntactic
pattern learning by songbirds . Nature 440 (April): 1 204 7.
Geschwind , D. H. , and B. L. Miller . 2001 . Molecular approaches to
cerebral laterality: Development and neurodegeneration . American
Journal of Medical Genetics 101 : 370 81.
Gil-da-Costa , Ricardo , Alex Martin , et al. 2006 . Species-specifc calls
activate homologs of Brocas and Wernickes areas in the macaque .
Nature Neuroscience 9 (July): 1064 70.
Green , Richard E. , Johannes Krause , et al. 2006 . Analysis of one million
base pairs of Neanderthal DNA . Nature 444 (November): 330 6.
Grodzinsky , Yosef , and Katrin Amunts , eds. 2006 . Brocas Region .
Oxford : Oxford University Press .
Haesler , Sebastian , Kazuhiro Wada , et al. 2004 . FoxP2 expression in
avian vocal learners and non-learners . Journal of Neuroscience 24
(March): 3164 75.
Hauser , Marc D. , David Barner , et al. 2007 . Evolutionary linguistics: A
new look at an old landscape . Language Learning and Development
3 .2: 101 32.
Hauser , Marc D. , Noam Chomsky , and W. Tecumseh Fitch. 2002 . Te
faculty of language: What is it, who has it, and how did it evolve?
Science 298 (November): 1569 79.
Holy , Timothy E. , and Zhongsheng Guo . 2005 . Ultrasonic songs of male
mice . PLoS Biology 3 (December): 217786. Available online at: www.
plosbiology.org.
Iacoboni , Marco , Roger Woods , et al. 1999 . Cortical mechanisms of
human imitation . Science 286 : 2526 8.
Jenkins , Lyle . 2000 . Biolinguistics: Exploring the Biology of Language .
Cambridge : Cambridge University Press .
Jenkins , Lyle , ed. 2004 . Variation and Universals in Biolinguistics .
Amsterdam : Elsevier .
Kegl , Judy . 2004 . Language emergence in a language-ready
brain: Acquisition. In Variation and Universals in Biolinguistics , ed.
L. Jenkins , 195236. Amsterdam : Elsevier .
Kennedy , D. N., K. M. OCraven , et al. 1999 . Structural and functional
brain asymmetries in human situs inversus totalis . Neurology 53
(October): 1260 5.
Explaining Language
55
Somerville , M. J., C. B. Mervis , et al. 2005 . Severe expressive-language
delay related to duplication of the Williams-Beuren locus . New
England Journal of Medi cine 353 .16: 1694 701.
Sun , Tao , Christina Patoine , et al. 2005 . Early asymmetry of gene tran-
scription in embryonic human left and right cerebral cortex . Science
308 (June ): 1794 8.
Sun , Tao , and Christopher A. Walsh . 2006 . Molecular approaches to
brain asymmetry and handedness . Nature Reviews Neuroscience 7
(August): 65562.
Tettamanti , Marco , Giovanni Buccino , et al. 2005 . Listening to action-
related sentences activates fronto-parietal motor circuits . Journal of
Cognitive Neuroscience 17 : 273 81.
White , Stephanie A. , Simon E. Fisher , et al. 2006 . Singing mice, songbirds,
and more: Models for Foxp2 function and dysfunction in human speech
and language . Journal of Neuroscience 26 (October): 10376 9.
Shapiro , K. A., L. R. Moo , et al. 2006 . Cortical signatures of noun and verb
production . Proceedings of the National Academy of Sciences USA 103
(January): 16444 9.
Sherwood , Chet C. , Douglas C. Broadfield , et al. 2003 . Variability
of Brocas area homologue in African great apes: Implications for
language evolution . Te Anatomical Record Part A: Discoveries
in Molecular, Cellular, and Evolutionary Biology 271A (March):
276 85.
Shu , Weiguo , Julie Y. Cho , et al. 2005 . Altered ultrasonic vocalization in
mice with a disruption in the Foxp2 gene . Proceedings of the National
Academy of Sciences USA 102 (July 5): 9643 8.
Shu , W., H. Yang , et al. 2001 . Characterization of a new subfamily of
winged-helix/forkhead (fox) genes that are expressed in the lung and
act as transcriptional repressors . Journal of Biological Chemistry
276 .29: 27488 97.
56
Yet normally, by about three years of age, a child will have
attained the fundamental knowledge of an infnite combinato-
rial multileveled system as well as essential constraints on this
infnite system, no matter what the language, no matter what
the country or culture the child is born into, no matter what the
limitless contextual, environmental variations across children,
cultures, and countries. Tis mystery, and the paradox of con-
strained infnity, drives the feld of language acquisition. How
is this fundamental knowledge acquired? How is it represented
and efected in the mind and brain?
THE FIELD OF LANGUAGE ACQUISITION
In many ways, the study of language acquisition stands at the
center of the language sciences, subsumes all of its areas, and
thus, perhaps, supersedes all in its complexity. It entails the
study of linguistic theory in order that the systems and com-
plexities of the end state of language knowledge be understood,
so that the status of their acquisition can be evaluated. Tis
involves all areas of linguistics, in addition to phonetics: pho-
nology , morphology , syntax , semantics , and prag-
matics (cf. Chierchia 1999 ). It entails language typology
because acquisition of any specifc language is possible, and
because this acquisition refects any possible typological varia-
tion. Barbara Grimes ( 2000 ) proposes that between 6,000 and
7,000 languages now exist in the end state. Te study of language
acquisition also entails the study of language typology because
language acquisition at any state of development represents a
form of manifestation of language knowledge, itself subject to
typological variation. It entails language change , because
the course of acquisition of language over time reveals changing
manifestations of language; language change must be related to
language acquisition, in ways still not understood (although see
Baron 1977 , deGraf 1999 , and Crain, Gloro, and Tornton 2006
for initial attempts to relate these areas; see also age groups .).
It involves psycholinguistics because any particular mani-
festation of language during language acquisition is revealed
through the cognitive and psychological infrastructure of the
mind and brain during hearing and speaking in real time ,
potentially within a visual modality as in sign languages ,
thus always involving language processing integrated with lan-
guage knowledge . It involves neurolinguistics, not only because
the knowledge and use of language must in some way relate
to its biological and neural instantiation in the brain, but also
because explanations of the gradual course of language acqui-
sition over time must critically evaluate the role of biological
change. (See brain and language and biolinguistics .)
At the same time, the study of language acquisition also stands
at the center of Cognitive Science. It cannot be divorced from other
closely related disciplines within Cognitive Science. Te relation
between language and thought is never dissociated in natural
language, thus crucially calling for an understanding of cognitive
psychology and philosophy ; and the formal computation involved
in grammatical systems is not fruitfully divorced from computer
science. Any use of language involves possibilities for vagueness
and ambiguities in interpretation, given the particular prag-
matic context of any language use. For example, how would one
fully understand the meaning of an ostensibly simple sentence like
6
ACQUISITION OF LANGUAGE
Barbara Lust
INTRODUCTION
How does the child, within the frst few years of life, come to
understand and to produce Hop on Pop or Cat in the Hat
(Dr. Seuss)?
(1) We all play ball
Up on a wall
Fall of the Wall (Dr. Seuss [1963] 199l, 1013)
(2) We looked!
And we saw him!
Te Cat in the Hat!
And he said to us (Dr. Seuss [1957] 1985 , 6)
How does the child come to know the precise sound varia-
tions that distinguish hop and pop or wall and ball, the
multiple meanings of words like play or in, the reference
of pronouns like we or him, the meaning of quantifers (see
quantifcation ) like all, the functional categories determining
noun phrases like the, the idioms like play ball, the system-
atic order and structure necessary for even simple sentences ,
the infnite recursion of sentences allowed by the coordinate
connective and, the infnite possibilities for propositions based
on the manipulation of even a few sound units and word units,
and the infnite possibilities for meaning and truth , which
Dr. Seuss continually demonstrated?
Even more stunning, how does the child both attain the infnite
set of possibilities and, at the same time, know the infnite set of
what is impossible, given this infnitely creative combinatorial sys-
tem? Just as there is no limit to what is possible, there is no limit to
what is impossible in language, for example, (3), or an infnite set
of other impossible word and structure combinations, for example,
(4). Tis provides an essential paradox for language learners: How
do they both acquire the infnite possibilities and, at the same time,
the constraints on the infnite set of impossible expressions?
(3) *Ylay
*lfa
(4) *Play all we ball
*in cat the hat
Acquisition of Language
57
varying positions in the feld on these focal issues, then exem-
plify a range of research results, displaying crucial new discover-
ies in the feld. Ill conclude by formulating leading questions for
the future. For those interested in pursuing these topics further, I
close with selected references for future inquiry in the feld.
Focal Tensions
Classically, approaches to the study of language acquisition have
been categorized as nativist or empiricist (see innateness and
innatism ). Tese approaches, which correspond generally to
rationalist or empiricist approaches to epistemology (see Lust
2006 for a review), have been typically associated with claims
that essential faculties responsible for language acquisition in
the human species involve either innate capabilities for language
or not. However, now the debates have become more restricted,
allowing more precise scientifc inquiry. No current proposals, to
our knowledge, suggest that nothing at all is innate. (See Elman
et al, 1996 for example). No one proposes that every aspect of
language knowledge is innate. For example, in the now-classic
Piaget-Chomsky Debate (see Piatelli-Palmarini 1980 ), Chomsky
did not deny that experience was necessary for a comprehensive
theory of language acquisition. On the other hand, Jean Piaget
proposed an essentially biological model of cognitive develop-
ment and an antiempiricist theory, coherent with Chomskys
position in this way .
Rather, the current issue that is most central to the feld of
language acquisition now is whether what is innate regarding
language acquisition involves components specifc to linguistic
knowledge, for example, specifc linguistic principles and
parameters , or whether more general cognitive knowledge or
learning principles themselves potentially innate can account
for this knowledge . Tis issue corresponds to the question of
whether organization of the mental representation of language
knowledge is modular or not . Proponents of the view that spe-
cifcally linguistic factors are critical to an explanation of human
language acquisition generally work within a linguistic theory of a
modular language faculty, that is, a theory of universal gram-
mar (UG), which is proposed to characterize the initial state
of the human organism, that is, the state of the human organ-
ism before experience. Tis current descendant of Chomskys
LAD provides specifc hypotheses (working within a genera-
tive grammar framework) regarding the identity of linguistic
principles and parameters that may be innately, or biologically,
determined in the human species (e.g., Anderson and Lightfoot
2002 ) . Proponents of the alternative view generally assume a
functionalist theory and a model of cultural learning, that is,
a model where culture in some ways provides the structure of
language, potentially without specifc linguistic printiples, with
only general learning principles (e.g., Van Valin 1991 ; Tomasello
2003 ; Tomasello, Kruger, and Ratner 1993 ) .
Positions in the Field
Debates in the feld of language acquisition continue today
around issues of innateness and modularity (e.g., Pinker, 1994 ;
Tomasello 1995 ; Edelman 2008 ). Research in the feld is often
polarized according to one paradigm or the other. Several
Shes going to leave without understanding the particular con-
text in which this sentence is used, and there is an infnite set of
these contexts possible. Te challenge of developing an infnitely
productive but constrained language system embedded in the
infnite possibilities for pragmatic determination of interpretation
remains a central challenge distinguishing machine language and
the language of the human species. All areas of cognitive science
must be invoked to study this area. Finally, especially with regard to
language acquisition in the child, developmental psychology must
be consulted critically. Te child is a biologically and cognitively
changing organism. Indeed, an understanding of commonalities
and/or diferences between language acquisition in child and adult
requires expertise in the feld of developmental psychology .
In addition, the fact that the human species, either child or
adult, is capable not only of single language acquisition but also
of multilanguage acquisition, either simultaneously or sequen-
tially, exponentially increases the complexity of the area (see
bilingualism and multilingualism ). Te fact that these
languages may vary in their parametric values (see parame-
ters ) and be either oral or visual further complicates the area .
THEORETICAL FOUNDATIONS
Not surprisingly, then, the feld of research involving the study of
language acquisition is characterized by all the complexity and
variation in theoretical positions that characterize the feld of lin-
guistics and the language sciences in general. Te area of language
acquisition research refects varying approaches to the study of
grammar, that is, variations regarding viewpoints on what consti-
tutes the end state of language knowledge that must be acquired,
for example, generative or functionalist approaches. At
the same time, it is characterized by variation in approach to the
study of language acquisition, in particular, ranging from vari-
ous forms of logical to empirical analyses, and including disputes
regarding methodological foundations and varying attempts
at explanatory theories (ranging from rationalist to empiricist
types). (Lust 2006 , Chapter 4, provides a review.)
Te feld is led by a strong theory of what is necessary logi-
cally for a strong explanatory theory of language acquisition in
the form of Noam Chomskys early proposal for a language
acquisition device ( LAD ) . Tis model not only involved
explication of what needed to be explained but also spawned
decades of research either for or against various proposals bear-
ing on components of this model. Yet, today, no comprehensive
theory of language acquisition exists, that is, no theory that would
fully account for all logically necessary aspects of a LAD.
At the same time, decades of research have now produced an
explosion of new discoveries regarding the nature of language
acquisition. Tese empirical discoveries, combined with theo-
retical advances in linguistic theory and with the development
of the interdisciplinary feld of cognitive science, bring us today
to the verge of such a comprehensive theory, one with frm sci-
entifc foundations .
STRUCTURE OF THIS OVERVIEW
In this overview, I frst briefy characterize and distill focal ten-
sions in the feld of language acquisition and briefy survey
The Cambridge Encyclopedia of the Language Sciences
58
of a proposal for grammatical mapping in Lust 1999 , 2006 and
Santelmann et al. 2002 .) Te task of language acquisition goes
beyond the periphery of grammatical knowledge and beyond
mere triggering of parameters; it lies in constructive synthesis of
a specifc language grammatical system, which is constrained by
UG but not fully determined by it .
(iv) Some propose that UG is irrelevant to language acquisi-
tion; the process can be described by an alternative mechanism,
critically involving some form of usage-based learning (e.g.,
Elman et al. 1996; Tomasello 2003 , 2005 ): In general, the only
fully adequate accounts of language acquisition are those that give
a prominent role to childrens comprehension of communicative
function in everything from words to grammatical morphemes to
complex syntactic constructions (Tomasello 2005 , 183) .
Within the proposals of iiii which all work within a generative
UG framework, (i) and (iii) propose a strong continuity hypothesis
of language development, although they difer crucially in what
is claimed to be continuous. In (iii), it is only the essential set of
principles and parameters of UG constituting the initial state
that is proposed to be biologically programmed and to remain
constant over development, while in (i), it is comprehensive
grammatical knowledge, with no distinction made between UG
and SLG. While a maturational approach such as in (ii) would
appear to maintain the premise of biological programming of
language knowledge, and thus be consistent with a theory of
UG, it raises several critical theoretical and empirical issues that
are still unresolved. For example, theoretically, the question
arises: What explains the change from one UG state to another, if
this determination is not programmed in UG itself? Empirically,
in each area where a maturational account has been proposed,
further advanced research has often revealed grammatical com-
petence that was thought to be missing in early development,
for example, early knowledge of functional categories. (See Lust
1999 and Wexler 1999 for debate.)
All proposals must now confront the fundamental develop-
mental issues: what actually changes during language devel-
opment and why. Tose within the generative paradigm must
sharpen their vision of what in fact constitutes UG, and those
outside of it must sharpen their view of how infnitely creative but
infnitely constrained grammatical knowledge can be attained
on the basis of communicative function alone. All proposals
must be accountable to the wide and cumulative array of empir-
ical data now available. (See, for example, Discoveries in the
next section.)
Researchers are also opposed in their view of how the child
uses input data. (See Valian 1999 for a review.) All approaches
must confront this fundamental area, whether the UG frame-
work is involved (e.g., in explaining how the environment trig-
gers parameter setting) or whether an empiricist framework is
involved (where the mechanism of imitation of individual words
and/or utterances may be viewed as an essential mechanism of
data processing and grammar development for the child, e.g.,
Tomasello 1995 , 2002 , 2005 ). Indeed, the very nature of imitation
of language stimuli is under investigation (e.g., Lust, Flynn, and
Foley 1996 ). In each case, the mechanism proposed must reli-
ably account for the childs mapping from these input data to the
specifc knowledge of the adult state, thus solving what has been
called the logical problem of language acquisition and it must
specifc issues focalize these debates. Two of the most pressing
currently concern the nature of development , and the nature of
the childs use of language input .
Critically, all approaches must confront the empirical fact
that childrens language production and comprehension develop
over time. Language acquisition is not instantaneous. Te ques-
tion is whether childrens language knowledge also develops
over time, and if it does, does it change qualitatively, and if it does
change qualitatively, what causes the change. Tis question of
development is parallel to questions regarding cognitive devel-
opment in general, where stage theories are debated against
continuity theories. Positions in the feld today vary in several
ways regarding their understanding of the nature of develop-
ment in language acquisition. Even within researchers working
within a rationalist paradigm that pursues universal grammar as
a specifcally linguistic theory of a human language faculty , there
is disagreement. Te theory of UG does not a priori include an
obvious developmental component. Positions in the feld can be
summarized as in iiv.
(i) Some propose that essential language knowledge is fun-
damentally innate, and do not centrally address the issue of
what change in language knowledge may occur during language
development (e.g., Crain 1991 ). In this view, apparent cases of
language knowledge change in a young childs production or
comprehension are often mainly attributed to methodological
failures, for example, involving the researchers choice of spe-
cifc tasks for testing childrens knowledge. Te view of language
development is characterized by the following: On the account
we envision, childrens linguistic experience drives children
through an innately specifed space of grammars, until they hit
upon one that is suf ciently like those of adult speakers around
them, with the result that further data no longer prompts further
language change (Crain and Pietroski 2002 , 182). (Compare
LAD and a recent proposal by Yang 2006 .) On this view, it would
appear that grammars are predetermined and available for
choice. Presumably, specifc lexicons, as well as other periph-
eral aspects of language knowledge, would be exempt from this
claim of predetermined knowledge. Te range and nature of the
innately specifed space of grammars would have to be expli-
cated, as this framework develops .
(ii) Some propose that UG itself develops over time in a man-
ner that is biologically driven, that is, through maturation. Major
aspects of language development, for example late acquisition of
the passive construction (such as Te ball was thrown by Mary
in English), are attributed to biologically determined changes in
UG (e.g., Wexler 1999 ; Radford 1990 ). Tis maturation is proposed
to be the major explanation of language development. Specifc
language grammars (SLGs) are not, for the most part, treated as
essential independent components of the language acquisition
challenge, but often as simply triggered by UG parameters. Any
learning that would be involved for the child in language devel-
opment would involve merely peripheral aspects of language,
with the nature of the periphery yet to be specifed .
(iii) Some propose that UG is continuous, but UG is inter-
preted as involving just those principles and parameters that con-
strain and guide the language acquisition process. SLG develops
over time through experience with data from a specifc language
and through the childs active construction of it. (See examples
Acquisition of Language
59
languages, the proportion of labials in both late babbling and
frst words will refect the input of labials in the adult language
being acquired (de Boysson-Bardies, Vihman, and Vihman
1991 ).
In fact, children are picking out words as formal elements from
the speech stream even before they understand them. For exam-
ple, 8-month-olds who were exposed to stories read to them for
10 days (30 minutes of prerecorded speech, including three short
stories for children) during a two-week period, two weeks later
distinguished lists of words that had appeared in these stories,
for example, (5a), from those which had not, for example, (5b)
(Jusczyk and Hohne 1997 ).
(5) a. sneeze
elephant
python
peccaries
b. aches
apricot
sloth
burp
Given the age of the children and the nature of the words
tested, it is clear that children are engaged both in word segmen-
tation and in long-term storage of these formal elements, even
without their semantic content. Moreover, like the acquisition of
phonology, childrens sensitivities to the language-specifc struc-
ture of words begins to show refnement after the sixth month.
American 9-month-olds, though not 6-month-olds, listened
longer to English words than to Dutch words, while Dutch
9- month-old infants showed the reverse preference (Jusczyk
et al. 1993 ).
Simultaneously, and in parallel, in the area of syntax, infants
have also been found to be laying the formal foundations for
language knowledge even within the frst 12 months. Tey are
carving out the formal elements that will form the basis for the
syntax of the language they are acquiring. Precise sensitivities
to linear order, as well as to constituent structure , have
been discovered in these frst few months of life. For example,
infants as young as 4 months of age have been found to dis-
tinguish natural well-formed clause structure, like (6a) from
non-natural ones, like (6b), in stories read to them (where /
represents clause breaks through pauses, and the stimuli are
matched in semantic and lexical content) (Hirsh-Pasek et al.
1987 ) .
(6) a. Cinderella lived in a great big house/ but it was sort of dark/
because she had this mean, mean, mean stepmother
b. in a great big house, but it was/ sort of dark because she
had/ this mean
Experimental research has begun to reveal how infants
accomplish this discrimination in the speech stream, suggest-
ing that the mapping of prosodic phrasing to linguistic units may
play an important role.
Again, sensitivities become more language specifc after the
sixth month. Although 6-month-olds did not distinguish natu-
ral (e.g., 7a) and non-natural phrasal (verb phrase) structures
(e.g., 7b), that is, phrasal constituents smaller than the clause,
9-month-olds did:
also be empirically verifed, that is, be a veridical account of how
the child actually does use surrounding data. Tose approach-
ing the logical problem of language acquisition are formally
diagnosing the properties of syntactic and semantic knowledge
that must be acquired, and assessing varied formal learnability
theories that may possibly account for the childs mapping from
actual input data to this knowledge. (e.g., Lightfoot 1989 ) . (See
also projection principle .)
Discoveries
Fortunately, not only have the last decades seen continuous
theoretical developments in areas of linguistics regarding the
nature of language knowledge, and continuous sharpening
of the debate on the nature of language development, but the
feld of language acquisition has also produced a wealth of new
empirical discoveries, all of which promise to inform the crucial
questions and debates in the feld, and to eventuate in a more
comprehensive theory of language acquisition than has yet been
available. Ill exemplify only some of these here, all of which are
foci of intense current research. (See Lust 2006 for a more com-
prehensive review.)
THE FIRST 12 MONTHS. Informed by developed methodologies
for investigating infant perception of language, we now know
that neonates show early sensitivities to language variations
and categorize these variations (e.g., Ramus et al. 2000 ; Mehler
et al. 1996 , 1988 ). For example, given varying sound stimuli from
Japanese and varying sound stimuli from English speech, new-
born infants distinguish variation from Japanese to English sig-
nifcantly more than variation within either English or Japanese,
including speaker variation. In this sense, the infant is seen to
be categorizing variation across languages. (See infantile
responses to language .) More specifcally, we also know
that formal aspects of language (phonetic, phonological, and
syntactic) begin to develop at birth, even before language com-
prehension, and provide the foundations for the appearance of
overt language production and comprehension in the child at
about 12 months of age. Now we also know something of how
that development proceeds even during the frst 12 months of
life.
In the area of acquisition of phonology, we have discovered
the very fne, precise, and extensive phonetic sensitivities of the
newborn, sensitivities that appear to be exactly the right ones
to underlie all potential cross-linguistic phonological contrasts,
for example, contrasts in voicing or place and manner features
that characterize speech sounds. Once again, these sensitivities
are categorical (as in categorical perception; see speech per-
ception in infants ). We know that by about 6 months, these
sensitivities reveal appropriate cross-linguistic selection and
modulation, and by 12 months, this process is nicely attuned to
the childs frst productive words of their specifc language (e.g.,
Werker 1994 ; see Jusczyk 1997 and de Boysson-Bardies 1999 for
reviews). For example, while the 12-month-old infant acquir-
ing Hindi will have maintained sensitivity to Hindi contrasts in
aspiration, the infant acquiring English will show diminished
response to such distinctions, which are not linguistically con-
trastive in English. Although labials appear in frst words across
The Cambridge Encyclopedia of the Language Sciences
60
category sensitivities appear to be available even within the frst
12 months (e.g., Shady 1996 ; Demuth 1994 ).
Contrary to the widespread view that the contentful lexicon
(involving nouns and verbs) is the privileged basis for acquisition
of syntax in early language acquisition, these results are begin-
ning to suggest that, in fact, functional categories are fundamen-
tal (see lexical learning hypothesis ) .
PRINCIPLES AND PARAMETERS. Principles and parameters that
are hypothesized to provide the linguistic content of UG and
of the human language faculty provide leading hypotheses for
language acquisition research. (See principles and param-
eters theory and language acquisition .) Investigators
continue to search not only for theoretical motivation for such
principles and parameters but also for empirical evidence of the
role of UG-based principles and parameters in early language
acquisition. A wide range of empirical research has now accrued
in this paradigm, paralleling theoretical developments (e.g.,
Snyder 2007 ; Crain and Lillo-Martin 1999 ; Guasti 2002 ; Roeper
2007 ; Lust 2006 , among others). Tis research reveals a wide
array of evidence regarding very fundamental linguistic prin-
ciples, including the central and perhaps most fundamental UG
principle of structure dependence :
Te rules operate on expressions that are assigned a certain struc-
ture in terms of a hierarchy of phrases of various types. (Chomsky
1988 , 45)
Evidence for this fundamental linguistic principle has been
adduced in studies of childrens acquisition of several areas,
including various types of question formation (e.g., Crain and
McKee 1985 ; Crain and Nakayama 1987 ; deVilliers, Roeper, and
Vainikka 1990 ), empty category and pronoun interpretation (e.g.,
Cohen Sherman and Lust 1993 ; Nuez del Prado, Foley, and Lust
1993 ; Lust and Cliford 1986 ), and quantifer scope (e.g., Chien
1994 ).
Results from young children at early stages of development
across languages have shown that they very early distinguish
coordinate and subordinate structures and that they diferenti-
ate syntactic processes in these diferent structures accordingly.
For example, in English, they very early distinguish sentences like
(10a) and (10b) in both comprehension and production (Cohen
Sherman and Lust 1993 ):
(10) a. [Te turtle tickles the skunk] and [0 bumps the car]
b. Tom [promises/tells Billy [0 to eat the ice cream cone]]
Chinese children diferentiate coordinate and embedded
structures and diferentiate subjects and topics (see topic
and comment ) in Chinese accordingly (Chien and Lust 1985 ).
Across English, Japanese, and Sinhala, children diferentiate pos-
sibilities for anaphora according to the embedded or adjoined
structure in which proforms appear (Lust and Cliford 1986 ;
Oshima and Lust 1997 ; Gair et al. 1998 ; Eisele and Lust 1996 ) .
In general, very early linguistic knowledge including knowl-
edge of language involving diminished content, where direct
overt phonetic information is not available is attested in stud-
ies of childrens acquisition of sentences with ellipses. For exam-
ple, in sentences such as (11), young children have been found
to reveal not only competence for empty category interpretation
(7) a. Te little boy at the piano/ is having a birthday party
b. Te little boy at the piano is having/ a birthday party
In all these cases, both phonological and syntactic develop-
ment does not reduce either to simple loss of initial sensitivities
or to simple accrual or addition of new ones, but a gradual inte-
gration of a specifc language grammatical system.
More recentl y, research has begun to reveal even more pre-
cisely how these early sensitivities are related to later language
development , thus foreshadowing a truly comprehensive theory
of language acquisition (Newman et al. 2006 ; Kuhl et al. 2005 ) .
BEYOND FIRST WORDS: LINGUISTIC STRUCTURE. Continuous with
advances in our understanding of early infant development, we
are now also seeing a potential revolution in our understand-
ing of the early stages of overt language acquisition, that is,
those periods within the frst three years of life, where the child
is beginning to overtly manifest language knowledge in terms
of language production and comprehension. Child language in
these early periods has traditionally been referred to as holo-
phrastic or, a little later, telegraphic (see two-word stage )
in nature. Even in these early periods, numerous studies are now
revealing childrens very early sensitivity to functional categories
(FCs) in language, that is, to grammatical elements that function
formally to a large degree, often with little or no semantic con-
tent. Tese functional categories play a critical role in defning
constituent structure in language, and in defning the locus of
syntactic operations. Tus, the evidence that infants and tod-
dlers are accessing these FCs in their early language knowledge
begins to provide crucial data on the foundations for linguistic
systems in the child. (Such FC are refected in diferent ways
across languages. In English they are refected in determiners
such as the, auxiliary verbs like do, complementizers like
that, or infection of verbs, for example.)
Early research had revealed that young children perceive and
consult functional categories, such as determiners, even when
they are not overtly producing them regularly in their natural
speech. For example, L. Gerken, B. Landau, and R. Remez ( 1990 )
showed that 2-year-olds recognize the distinction between
grammatical and ungrammatical function words, contrasting
these, for example (8a) and (8b), in their elicited imitation of
these sentences. Gerken and Bonnie McIntosh ( 1993 ) showed
that 2-year-olds used this knowledge in a picture identifcation
task, discriminating between (9a) and (9b), where grammatical
forms facilitated semantic reference.
(8) a. Pete push es the dog
b. Pete push o na dog
(9) a. Find the bird for me
b. Find was bird for me
More recently, a wide range of experimental researchers work-
ing with expanded infant testing methods have replicated these
results, and also revealed similar functional category distinctions
even in younger children. For example Yarden Kedar, Marianella
Casasola, and Barbara Lust ( 2006 ) showed that infants as young
as 18 months also distinguish sentences like (9a) and (9b) in a
preferential looking task, and again, object reference is facili-
tated by the grammatical form. Precursors to these functional
Acquisition of Language
61
learning by hanging on to some aspect of the input that it can
access. Research is beginning to provide evidence on how and
when these forms of bootstrapping may work in the young child
(e.g, work of Gleitman 1990 and others on syntactic bootstrap-
ping and its role in early lexical learning and the the collection of
papers in Morgan and Demuth 1996 ). Precise hypotheses are now
being formed regarding the mechanisms by which certain param-
eters may be set very early by the infant, even before the frst word,
by consulting prosodic and other aspects of the speech stream
(Mazuka 1996 ) .
CROSS-SPECIES COMPARATIVE METHOD. Advances in cross-
species comparisons now provide an additional dimension to the
study of language acquisition, allowing refnement of our speci-
fcation of what is particularly human and of what is particularly
linguistic about human acquisition of language (e.g., Hauser,
Chomsky, and Fitch 2002 ; Call and Tomasello 2007 ; Ramus et al.
2000 ). For example, comparative studies with cotton-top tamarin
monkeys revealed certain capabilities in this species to discrimi-
nate language stimuli (e.g., Dutch and Japanese) that were com-
parable to human infants (Ramus et al. 2000 ). Tis implied that
a general auditory process was involved in the discrimination. In
contrast, other processes discovered in early acquisition of phonol-
ogy have been found not to generalize. (See, for example, the work
of Kuhl et al. 2005 .)
RESILIENCE. Finally, the tremendous resilience of the language
acquisition feat in the face of varying input, including dearth of
input, has been revealed through important work on young deaf
childrens spontaneously created home sign (Goldin-Meadow
2003 ). Te role of community in sign language creation is also
revealed through in-depth studies of the creation of Nicaraguan
Sign Language in young children (Senghas 1995 ; Kegl, Senghas,
and Coppola 1999 ) .
Toward the Future
FIRST AND SECOND LANGUAGE IN CHILD AND ADULT. Current and
future studies that seek to triangulate the language faculty by
precise comparative studies between child frst language acquisi-
tion and adult second language and between child monolingual
frst language acquisition and multilanguage acquisition promise
to be able to dissociate factors related to biological maturation,
universal grammar , and specifc language grammar in ways not
achievable by studies of frst language acquisition alone (e.g.,
Flynn and Martohardjono 1994 ; Yang and Lust 2005 ) .
PRAGMATIC AND COGNITIVE DEVELOPMENT AND GRAMMATICAL
DEVELOPMENT. Studies of the integration of childrens developing
pragmatic knowledge with their grammatical knowledge have only
begun; yet this integration characterizes every aspect of a childs
use of language (cf. Clark 2003 ). Such studies may critically inform
our understanding of childrens early syntax development (e.g.,
Blume 2002 ). Similarly, studies involving interactions between
general cognition and specifc aspects of linguistic knowledge (e.g.,
Gentner and Goldin-Meadow 2003 ) will be critical for resolving
issues of modularity in language acquisition. (See constraints
in language acquisition .)
(in the does too clause, where the second clause does not
state what Bert did), but also competence for construction of the
multiple interpretations allowed in this ambiguous structure (as
in 11ad), and constraint against the ungrammatical possibili-
ties (as in 11 ei). (See Foley et al. 2003 for an example.) In other
words, they evidence early control of and constraints on empty
category interpretation and on other forms of ellipsis. Studies
of Chinese acquisition show similar results (Guo et al. 1996 ).
Without structure dependence and grammatical computation
over abstract structure underlying such sentences (e.g., recon-
structing the verb phrase [VP] in the second clause), children
would not be expected to evidence this competence (see Foley
et al. 2003 ). All interpretations are pragmatically possible.
(11) Oscar bites his banana and Bert does too.
Possible Interpretations:
a. O bites Os banana and B bites Bs banana.
b. O bites Os banana and B bites Os banana.
c. O bites Bs banana and B bites Bs banana.
d. O bites Es banana and B bites Es banana.
Impossible Interpretations
*e. O bites Os banana and B bites Es banana.
*f. O bites Bs banana and B bites Os banana.
*g. O bites Bs banana and B bites Es banana.
*h. O bites Es banana and B bites Os banana.
*i. O bites Es banana and B bites Bs banana.
As is the case in this study of VP ellipsis acquisition, empiri-
cal research results in the Principles and Parameters framework
mutually inform theoretical development; they contribute to and
help to resolve theoretical debates on the representation of this
area of linguistic knowledge.
Evidence for early parameter setting with regard to linear
order in natural language is now provided by a wide range of
cross-linguistic studies showing very early language-specifc
sensitivity to the clausal head direction and directionality of
RECURSION in the language being acquired (e.g., Lust, in prep-
aration; Lust and Chien 1984 ). Very early diferentiation of the
pro drop (i.e., argument omission wherein subjects may not be
overtly expressed) possibilities of a language have been attested
across languages (e.g., Italian and English, Spanish and English)
(Valian 1991 ; Austin et al. 1997 , after Hyams 1986 ). In fact, chil-
dren have been found to critically consult the complementizer
phrase , and the subordinate structure domain in order to make
this diferentiation (Nuez del Prado, Foley, and Lust 1993 ) .
FROM DATA TO GRAMMAR. Our understanding of how the infant,
from birth, consults the surrounding speech stream is now also
expanding quickly. Research has revealed very fne sensitivities
to particular aspects of the speech stream (e.g., Safran, Aslin,
and Newport 1996 ), and research has begun to isolate the pre-
cise role of particular cues in this process, for example, STRESS ,
phonotactic constraints, and statistical distributions (e.g, Johnson
and Jusczyk 2001 ). Various forms of bootstrapping may be avail-
able to the child (phonological or prosodic or syntactic bootstrap-
ping, for example). Here, bootstrapping generally refers to that
process by which the child, in the initial state, might initiate new
The Cambridge Encyclopedia of the Language Sciences
62
line more fully with studies of the language acquisition process.
(See Baker 2005 for an example of an argument for representa-
tion of this integration.) Many current disputes regarding the
fundamentals of language acquisition cannot be resolved until
disputes regarding the nature of the adult state of language
knowledge are further resolved, and until the feld of Linguistics
and Psychology are more fully integrated in the feld of Cognitive
Science .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aitchison , Jean. 2007 . Te Articulate Mammal . 5th ed. London : Routledge .
Almi , C. R. , M. J. Rivkin , R. C. McKinstry , and Brain Development
Cooperative Group . 2007 . Te NIH MRI study of normal brain devel-
opment (Objective-2): Newborns, infants, toddlers, and preschoolers .
NeuroImage 35 : 308 25.
Anderson , Stephen R. , and David W. Lightfoot . 2002 . Te Language
Organ . Cambridge : Cambridge University Press .
Austin , Jennifer , Maria Blume , David Parkinson , Zelmira Nuez del
Prado , Rayna Proman , and Barbara Lust . 1997 . Te status of pro-
drop in the initial state: Results from new analyses of Spanish. In
Contemporary Perspectives on the Acquisition of Spanish , Vol 1:
Developing Grammars , ed. Anna Perez-Leroux and W. Glass , 3551.
Boston : Cascadilla Press .
Baker , Mark C. 2005 . Mapping the terrain of language learning .
Language Learning and Development 1. 1: 93 129.
Baron , Naomi. 1977 . Language Acquisition and Historical Change .
Amsterdam : North Holland .
Binder , Jefrey R. , and Cathy J. Price . 2001 . Functional neuroimaging of
language. In Handbook of Functional Neuroimaging of Cognition , ed.
Robert Cabeza and Alan Kingstone , 187251. Cambridge, MA : MIT
Press .
Blake , Joanna. 2000 . Routes to Child Language: Evolutionary and
Developmental Precursors . Cambridge : Cambridge University Press .
Blume , Maria. 2002 . Discourse-morphosyntax in Spanish non-fnite
verbs: A comparison between adult and child grammars . Unpublished
Ph.D. diss., Cornell University.
Boysson-Bardies , Bndicte de . 1999 . How Language Comes to
Children: From Birth to Two Years . Cambridge, MA : MIT Press .
Boysson-Bardies , Bndicte de , B. Vihman , and M. M. Vihman . 1991 .
Adaptation to language: Evidence from babbling and frst words in
four languages . Language 67. 2: 297 319.
Call , Josep , and Michael Tomasello . 2007 . Te Gestural Communication
of Apes and Monkeys . Mahwah, NJ : Lawrence Erlbaum .
Casey , B. J. , Adriana Galvan , and Todd Hare . 2005 . Changes in cere-
bral functional organization during cognitive development . Current
Opinion in Neurobiology 15 : 239 44.
Chien , Yu-Chin . 1994 . Structural determinants of quantifer scope: An
experimental study of Chinese frst language acquisition. In Syntactic
Teory and First Language Acquisition: Cross-Linguistic Perspectives .
Vol 2: Binding, Dependencies, and Learnability . Ed. B. Lust , G. Hermon ,
and J. Kornflt , 391416. Mahwah, NJ : Lawrence Erlbaum .
Chien , Yu-Chin , and Barbara Lust . 1985 . Te concepts of topic and
subject in frst language acquisition of Mandarin Chinese. Child
Development 56 : 1359 75.
Chierchia , Gennaro. 1999 . Linguistics and language. In Te MIT
Encyclopedia of the Cognitive Sciences , ed. R. Wilson and F. Keil , xci
cix. Cambridge, MA : MIT Press .
Chomsky , Noam. 1988 . Language and Problems of Knowledge .
Cambridge, MA : MIT Press .
Clark , Eve V. 2003 . First Language Acquisition . Cambridge : Cambridge
University Press .
A NEUROSCIENCE PERSPECTIVE AND BEYOND. Current advances
in brain imaging methods (such as fMRI, and neurophysiologi-
cal measures, such as EEG [electroencephalography] and MEG
[magnetoencephalography]) and in genome mapping (Greally
2007 ) promise new discoveries regarding fundamental issues
still open in the feld of language acquisition: How are language
knowledge and language acquisition represented in the brain?
(See neuroimaging and genes and language ). More pre-
cisely, what is the content of the initial state and how is it biologi-
cally represented? How is development in language knowledge
either determined by biological changes or a cause of them?
Tere have been many neuroscientifc results regarding language
in the adult state (e.g., Friederici 2002 ; Binder and Price 2001 ;
see syntax, n eurobiology of ; phonetics and phonology,
neurobiology of ; and semantics, neurobiology of , for
example), including multilingual language (e.g. Paradis 2000 ; see
bilingualism, neurobiology of ). Research on brain devel-
opment has also begun to yield fundamental descriptive results
(e.g., Casey, Galvan and Hare 2005 ). Nevertheless, it appears that
the issue of the language-brain relationship during early develop-
ment is a terra incognita. Te issue of how brain development
and cognitive development in the area of language development
co-occur in early development and over a lifetime will be one of
the key issues in the coming decades of the third millennium
(Friederici 2000 , 66). Advances in the description of brain devel-
opment (e.g., Almi et al. 2007 ) have not yet been related precisely
to advances in language acquisition.
Not only will advances in this area require technology appli-
cable to a young and changing brain but also advances in the
interdisciplinary feld of cognitive science. Tey will depend
on the development of strong linguistic theory, but also on the
development of a theory that maps the elements of linguistic the-
ory to the activation of language in the brain. At present, there is
still no theory of how an NP relates to a neuron (Marshall 1980 )
or how an abstract linguistic principle like structure dependence
or a particular linguistic constraint could be represented at a
neural level. Most imaging results to date provide evidence on
the processing of language, rather than on the knowledge of, or
representations of, language per se .
METHODOLOGY. Advances in this area will also require advance-
ment on issues of methodology: Simply put, we need to under-
stand, precisely, how our behavioral methodologies refect on,
and interact with, the language processes we are attempting to
study and defne (Swinney 2000 , 241). Advancement in this
area is a prerequisite to advancement in our understanding of
language processing, language knowledge, and their interaction
in language acquisition, as well as in the interpretation of brain
imaging results. Many persistent sources of disagreement in
the feld of language acquisition depend on resolution of issues
related to methodology (e.g., Crain and Wexler 1999 ; Lust et al.
1999 ), including those surrounding new infant-based methods
related to preferential looking .
LINGUISTIC THEORY AND LANGUAGE ACQUISITION. Finally, the
study of language acquisition will be advanced when linguistic
studies of the knowledge that must underlie the adult state of
language knowledge, no matter what language, are brought into
Acquisition of Language
63
Gleitman , Lila. 1990 . Te structural sources of verb meanings . Language
Acquisition 1. 1: 3 55.
Goldin-Meadow , Susan . 2003 . Te Resilience of Language: What Gesture
Creation in Deaf Children Can Tell Us About How All Children Learn
Language . New York : Psychology Press .
. 2005 . What language creation in the manual modality tells us
about the foundations of language. Linguistic Review 22 : 199 226.
Greally , John M. 2007 . Encyclopaedia of humble DNA . Nature
447 : 782 3.
Grimes , Barbara , ed. 2000 . Ethonologue: Languages of the World .
Dallas : SIL International .
Guasti , Maria Teresa . 2002 . Language Acquisition: Te Growth of
Grammar . Cambridge, MA : MIT Press .
Guo , F. F. , C. Foley , Y.-C. Chien , B. Lust , and C.-P. Chiang . 1996 .
Operatorvariable binding in the initial state: A cross-linguistic
study of VP ellipsis structures in Chinese and English . In Cahiers de
Linguistique Asie Orientale 25 .1: 3 34.
Hauser , Marc , Noam Chomsky , and Tecumseh Fitch . 2002 . Te fac-
ulty of language: What is it, who has it, and how did it evolve . Science
298 : 1569 79.
Hirsh-Pasek , Kathy , Diane Kemler-Nelson , Peter Jusczyk , K. Wright-
Cassidy , B. Druss , and L. Kennedy . 1987 . Clauses are perceptual units
for young children . Cognition 26 : 269 86.
Hirsh-Pasek , Kathy , and Roberta Michnick Golinkof . 1996 . Te Origins of
Grammar . Cambridge , MA : MIT Press .
Hyams , Nina. 1986 . Language Acquisition and the Teory of Parameters .
Dorcrecht and Boston : Reidel.
Johnson , Elizabeth , and Peter Jusczyk . 2001 . Word segmentation by
8-month-olds: When speech cues count more than statistics . Journal
of Memory and Language 44 : 548 67.
Jusczyk , Peter. 1997 . Te Discovery of Spoken Language . Cambridge,
MA : MIT Press .
Jusczyk , W. , A. D. Friederici , J. M. I. Wessels , V. Y. Svenkerud , and A. M.
Jusczyk . 1993 . Infants sensitivity to the sound patterns of native lan-
guage words . Journal of Memory and Language 32 : 402 20.
Jusczyk , Peter W , and Elizabeth A. Hohne . 1997 . Infants memory for
spoken words . Science 277 : 1984 6.
Kedar , Yarden , Marianella Casasola , and Barbara Lust . 2006 . Getting
there faster: 18- and 24-month-old infants use of function words to
determine reference . Child Development 77. 2: 325 38.
Kegl , Judith , Ann Senghas , and M. Coppola . 1999 . Creation through con-
tact: Sign language emergence and sign language change in Nicaragua.
In Language Creation and Language Change: Creolization, Diachrony
and Development , ed. M. DeGraf , 179237. Cambridge, MA : MIT
Press .
Kuhl , Patricia , Barbara Conboy , Denise Padden , Tobey Nelson , and
Jessica Pruitt . 2005 . Early speech perception and later language devel-
opment: Implications for the critical period . Language Learning and
Development 1. 3/4: 237 64.
Lightfoot , David. 1989 . Te childs trigger experience: Degree 0 learn-
ability . Behavioral and Brain Sciences 12 : 321 34.
Lust , Barbara. 1999 . Universal grammar: Te strong continuity hypoth-
esis in frst language acquisition. In Handbook of First Language
Acquisition , ed. William C. Ritchie and Tej K. Bhatia , 11155. San
Diego, CA : Academic Press .
. 2006 . Child Language: Acquisition and Growth . Cambridge :
Cambridge University Press .
. Universal Grammar and the Initial State: Cross-Linguistic Studies
of Directionality . In preparation.
Lust , Barbara , and Yu-chin Chien . 1984 . Te structure of coordination in
frst language acquisition of Mandarin Chinese . Cognition 17 : 49 83.
Lust , Barbara , and Teresa Cliford . 1986 . Te 3-D study: Efects of
depth, distance and directionality on childrens acquisition of ana-
phora: Comparison of prepositional phrase and subordinate clause
Cohen Sherman , Janet , and Barbara Lust . 1993 . Children are in control.
Cognition 46 : 1 51.
Crain , Stephen. 1991 . Language acquisition in the absence of experi-
ence . Behavioral and Brain Sciences . 14. 4: 597 650.
Crain , Stephen. , Takuya Gloro , and Rosalind Tornton . 2006 . Language
acquisition is language change. Journal of Psycholinguistic Research
35 : 31 49.
Crain , Stephen , and Diane Lillo-Martin . 1999 . An Introduction to Linguistic
Teory and Language Acquisition . Malden, MA : Basil Blackwell .
Crain , Stephen , and Cecile McKee . 1985 . Acquisition of structural restric-
tions on anaphors. In Proceedings of the Sixteenth Annual Meeting
of the North Eastern Linguistics Society , ed. S. Berman , J. Chloe , and
J. McDonough , 94110. Montreal : McGill University .
Crain , Stephen , and Mineharu Nakayama . 1987 . Structure dependence
in grammar formation . Language 63 : 522 43.
Crain , Stephen , and Paul Pietroski . 2002 . Why language acquisition is a
snap . Linguistic Review 19 : 163 84.
Crain , Stephen , and Kenneth Wexler . 1999 . Methodology in the study
of language acquisition: A modular approach. In Handbook of First
Language Acquisition , ed. William C. Ritchie and Tej K. Bhatia , 387
426. San Diego, CA : Academic Press .
deGraf , Michele. 1999 . Creolization, language change and language
acquisition: A prolegomenon. In Language Creation and Language
Change , ed. M. deGraf , 146. Cambidge, MA : MIT Press .
Demuth , Katherine. 1994 . On the underspecifcation of functional cat-
egories in early grammars. In Syntactic Teory and First Language
Acquisition: Cross-Linguistic Perspectives. Vol. 1. Ed. B. Lust , M. Suner ,
and J. Whitman , 11934. Hillsdale, NJ: Lawrence Erlbaum.
deVilliers , Jill , Tom Roeper , and Ann Vainikka . 1990 . Te acquisi-
tion of long-distance rules. In Language Processing and Language
Acquisition , ed. Lynn Frazier and Jill deVilliers , 25798. Dordrecht, the
Netherlands : Kluwer .
Edelman , Shimon. 2008 . Computing the Mind: How the Mind Really
Works . Oxford : Oxford University Press .
Eisele , Julie , and Barbara Lust . 1996 . Knowledge about pronouns: A
developmental study using a truth-value judgment task . Child
Development 67 : 3086 100.
Elman , Jefrey L. , Elizabeth A. Bates , Mark H. Johnson , Annette
Karmilof-Smith , Domenico Parisi , and Kim Plunkett . 1966 . Rethinking
Innateness . Cambridge, MA : MIT Press .
Flynn , Suzanne , and Gita Martohardjono . 1994 . Mapping from the ini-
tial state to the fnal stage: Te separation of universal principles and
language specifc principles. In Syntactic Teory and First Language
Acquisition: Cross Linguistic Perspectives , ed. Barbara Lust , Magui
Suner , and John Whitman , 31936. Hillsdale, NJ : Lawrence Erlbaum .
Foley , Claire , Zelmira Nuez del Prado , Isabella Barbier , and Barbara
Lust . 2003 . Knowledge of variable binding in VP-ellipsis: Language
acquisition research and theory converge . Syntax 6. 1: 52 83.
Friederici , A. D. 2000 . Te developmental cognitive neuroscience of lan-
guage: A new research domain . Brain and Language 71 : 65 8.
. 2002 . Towards a neural basis of auditory sentence processing .
Trends in Cognitive Science 6. 2: 78 84.
Gair , James , Barbara Lust , Lewala Sumangala , and Milan Rodrigo . 1998 .
Acquisition of null subjects and control in some Sinhala adverbial
clauses. In Studies in South Asian Linguistics: Sinhala and Other South
Asian Languages , 27185. Oxford : Oxford University Press .
Gentner , Dedre , and Susan Goldin-Meadow , eds. 2003 . Language in
Mind . Cambridge, MA : MIT Press .
Gerken , L. , B. Landau , and R. Remez . 1990 . Function morphemes in
young childrens speech perception and production . Developmental
Psychology 26 : 204 16.
Gerken , Louann , and Bonnie McIntosh . 1993 . Interplay of function mor-
phemes and prosody in early language . Developmental Psychology
29 .3: 448 57.
The Cambridge Encyclopedia of the Language Sciences
64
Roeper , Tom. 2007 . Te Prism of Grammar: How Child Language
Illuminates Humanism . Cambridge, MA : MIT Press .
Safran , Jenny R. , Richard N. Aslin , and Elissa L. Newport . 1996 . Statistical
learning by 8-month-old infants . Science 274 : 1926 8.
Santelmann , Lynn , Stephanie Berk , Shamita Somashekar , Jennifer
Austin , and Barbara Lust . 2002 . Continuity and development in the
acquisition of inversion in yes/no questions: Dissociating movement
and infection. Journal of Child Language 29 : 813 42.
Senghas , A. 1995 . Childrens Contribution to the Birth of Nicaraguan Sign
Language . Cambridge, MA : MIT Press .
Seuss , Dr. [1957] 1985 . Cat in the Hat . New York : Random House .
. [1963] 1991 . Hop on Pop . New York : Random House .
Shady , M. E. 1996 . Infants sensitivity to function morphemes. Ph.D.
diss., State University of New York at Bufalo.
Snyder , William. 2007 . Child Language . Oxford : Oxford University Press .
Swinney , David. 2000 . Understanding the behavioral-methodology/
language-processing interface . Brain and Language . 71 : 241 4.
Tomasello , Michael. 1995 . Language is not an instinct . Cognitive
Development 10 : 131 56.
. 2002 . Te New Psychology of Language. Cognitive and Functional
Approaches to Language Structure . Mahwah, NJ : Lawrence Erlbaum .
. 2003 . Constructing a Language: A Usage-Based Teory of Language
Acquisition . Cambridge : Harvard University Press .
. 2005 . Beyond formalities: Te case of language acquisition.
Linguistic Review 22 : 183 98.
Tomasello , Michael , Ann Cale Kruger , and Hilary Horn Ratner . 1993 .
Cultural learning . Behavioral and Brain Sciences 16 : 495 552.
Valian , Virginia. 1991 . Syntactic subjects in the early speech of American
and Italian Children . Cognition 40 : 21 49.
. 1999 . Input and language acquisition. In Handbook of Child
Language Acquisition , ed. William Ritchie and Tej Bhatia , 497530. San
Diego, CA : Academic Press .
Van Valin , Robert D. 1991 . Functionalist theory and language acquisi-
tion . First Language 11 : 7 40.
Werker , Janet F. 1994 . Cross-language speech perception:
Developmental change does not involve loss. In Te Development
of Speech Perception: Te Transition from Speech Sounds to Spoken
Words , ed. J. Goodman and H. Nusbaum , 93120. Cambridge, MA :
MIT Press .
Werker , Janet F. , and Richard C. Tees . 1984 . Cross-language speech per-
ception: Evidence for perceptual reorganization during the frst year of
life . Infants Behavior and Development 7 : 49 63.
Wexler , Kenneth. 1999 . Maturation and growth of grammar. In
Handbook of First Language Acquisition , ed. T. Bhatia and Wm. Ritchie ,
55110. San Diego, CA : Academic Press .
Yang , Charles D. 2006 . Te Infnite Gift . New York : Scribner .
Yang , S. , and B. Lust . 2005 . Testing efects of bilingualism on execu-
tive attention: Comparison of cognitive performance on two non-
verbal tests. BUCLD 29 Proceedings Online Supplement. Somerville,
MA: Cascadilla. Available online at http://www.bu.edu/linguistics/
APPLIED/BUCLD/supp29.html.
. 2007 . Cross-linguistic diferences in cognitive efects due to
bilingualism: Experimental study of lexicon and executive attention
in 2 typologically distinct language groups. BUCLD 31 Proceedings .
Somerville, MA: Cascadilla.
embedding. In Studies of the Acquisition of Anaphora: Defning the
Constraints , ed. B. Lust , 20344. Dordrecht, the Netherlands : Reidel
Press .
Lust , Barbara , Suzanne Flynn , and Claire Foley . 1996 . What children
know about what they say: Elicited imitation as a research method.
In Methods for Assessing Childrens Syntax , ed. Dana McDaniel , Cecile
McKee , and Helen Cairns . Cambridge, MA : MIT Press .
Lust , Barbara , Suzanne Flynn, Claire Foley, and Yu-Chin Chien. 1999 . How
do we know what children know? Problems and advances in establish-
ing scientifc methods for the study of language acquisition and linguis-
tic theory. In Handbook of First Language Acquisition , ed. William C.
Ritchie and Tej K. Bhatia , 42756. San Diego, CA : Academic Press .
Marshall , John C. 1980 . On the biology of language acquisition.
In Biological Studies of Mental Processes , ed. D. Caplan , 30120.
Cambridge, MA : MIT Press .
Mazuka , Reiko. 1996 . Can a grammatical parameter be set before the frst
word? Prosodic contributions to early setting of a grammatical param-
eter. In Signal to Syntax: Bootstrapping from Speech to Grammar in
Early Acquisition , ed. James Morgan and Katherine Demuth , 31330.
Hillsdale, NJ : Lawrence Erlbaum .
Mehler , Jacques , Emmanuel Dupoux , Tierry Nazzi , and G. Dehaene-
Lambertz . 1996 . Coping with linguistic diversity: Te infants view-
point. In Signal to Syntax: Bootstrapping from Speech to Grammar in
Early Acquisition , ed. James L. Morgan and Katherine Demuth , 10116.
Mahwah, NJ : Lawrence Erlbaum .
Mehler , Jaques , Peter Jusczyk , Ghislaine Lambertz , Nilofar Halsted ,
Josiane Bertoncini , and Claudine Amiel-Tison . 1988 . A precursor of
language acquisition in young infants . Cognition 29 : 144 78.
Morgan , James L. , and Katherine Demuth , eds. 1996 . Signal to
Syntax: Bootstrapping from Speech to Grammar in Early Acquisition .
Mahwah, NJ : Lawrence Erlbaum .
Newman , Rochelle , Nan Bernstein Ratner , Ann Marie Jusczyk , Peter W.
Jusczyk , and Kathy Ayala Dow . 2006 . Infants early ability to segment
the conversational speech signal predicts later language development: A
retrospective analysis . Developmental Psychology 42. 4: 643 55.
Nuez del Prado , Zelmira , Claire Foley , and Barbara Lust . 1993 . Te sig-
nifcance of CP to the pro-drop parameter: An experimental study of
Spanish-English comparison. In Proceedings of the Twenty-Fifth Child
Language Research Forum , ed. Eve Clark , 14657. Stanford, CA : CSLI .
Oshima , Shin , and Barbara Lust . 1997 . Remarks on anaphora in Japanese
adverbial clauses. In Papers on Language Acquisition: Cornell
University Working Papers in Linguistics , ed. Shamita Somashekar ,
Kyoko Yamakoshi , Maria Blume , and Claire Foley . Ithaca, NY : Cornell
University Press .
Paradis , Michel. 2000 . Te neurolinguistics of bilingualism in the next
decades . Brain and Language 71 : 178 80.
Piatelli-Palmarini , Massimo , ed. 1980 . Language and Learning: Te
Debate Between Jean Piaget and Noam Chomsky . Cambridge : Harvard
University Press .
Pinker , Steven. 1994 . Te Language Instinct . New York : W. W. Morrow.
Radford , Andrew. 1990 . Syntactic Teory and the Acquisition of English
Syntax . Cambridge : Cambridge University Press .
Ramus , Frank , Marc D. Hauser , Cory Miller , Dylan Morris , and Jaques
Mehler . 2000 . Language discrimination by human newborns and by
cotton-top tamarin monkeys . Science 288 .5464: 349 51.
65
theories of language have been developed most prominently
within ordinary language philosophy . Te ideas of Ludwig
Wittgenstein and the principles of speech-act theory have
been taken up by literary theorists, such as Mary Louise Pratt and
Stanley Fish . Some philosophers working in other areas of the
analytic philosophy of language, such as Donald Davidson , have
also had considerable infuence (see Dasenbrock 1993 ).
While intentionalist theories of language have certainly been
infuential in literary study, their use has been limited and, so
to speak, instrumental. Tey tend to be taken up for particular
interpretive aims. Tis is often the case with literary borrowings
from linguistics and the philosophy of language. However, the
uses of representationalist theories have been diferent. In these
cases, the literary and linguistic theories have been much more
thoroughly integrated. Tis is the result of two factors. First, the
research areas of the linguistic and literary theories overlap.
Second, there has been collaboration between linguists and lit-
erary theorists in treating these areas .
More exactly, there are two important representationalist
schools that have had signifcant impact in literary study. One
is cognitive linguistics. Some of the most important work in
cognitive linguistics has treated metaphor. While cognitive lin-
guists were initially interested in ordinary uses of metaphor, they
quickly extended their analyses to poetic metaphor . Tis was
facilitated by the collaboration of a linguist, George Lakof , and a
literary theorist, Mark Turner .
A similar point may be made about Chomskyan or generative
representationalism. One part of Chomskyan theory treats pho-
nology. Patterns in sound and stress are of obvious importance
in verse. Tus, certain aspects of poetic form may be included
within the framework of generative linguistic theories. Work in
this area has been facilitated by collaborations between linguists
and literary critics as well (see Fabb and Halle 2006 ). (Teorists
have used Chomskyan generative principles as a model for other
aspects of literary theory also; see generative poetics .)
Brain-related theorization on language and literature is less
developed, in part because it is highly technical and in part
because neuroscientifc theorization about language itself is
much more recent. Turning again to the divisions in the Preface,
we may distinguish between connectionist approaches and neu-
robiological approaches proper.
Tere has been some work on verbal art and parallel distrib-
uted processing. For example, some writers have used connec-
tionist models to discuss creativity (see Martindale 1995 ) and
there has been some work on connectionism and metaphor
(e.g., Chandler 1991 ). Tough limited, the work done in this feld
is generally well integrated into connectionist theories (i.e., it is
not solely instrumental).
Recently, there has been considerable interest in neuro-
biology and art. Tis work has addressed many nonlinguistic
aspects of brain function. However, some has focused specif-
cally on language. Much of this addresses hemispheric speciali-
zation, exploring distinctive features of verbal art (see Kane 2004 ;
poetic language, neurobiology of ).
Given the general orientation of literary critics, it is unsur-
prising that social aspects of speech and language have fgured
more prominently in literary study. At the level of dialogue, lit-
erary theorists have drawn on, for example, Paul Grices account
7
ELABORATING SPEECH AND WRITING:
VERBAL ART
Patrick Colm Hogan
Te past half century has seen considerable interaction between
the sciences of language and the study of literature. But this
interaction has been largely unidirectional, with infuence
fowing from language science to literature. Tis may be seen
most clearly in the massive impact of Ferdinand de Saussure on
literary study since the 1960s. However, generative gram-
mar, cognitive linguistics, connectionism , and other
approaches have also had efects on poetics and literary theory.
In the following pages, I wish to consider the general issues of
what distinguishes verbal art as an object of study in language
science. However, before I turn to this, it is important to get a
sense of how the analysis of literature and the analysis of lan-
guage have been interconnected since the current phase of lan-
guage study began 50 years ago.
THEORIES OF LANGUAGE AND THEORIES
OF LITERATURE
When the infuence of language science on literary theory is
considered, it is helpful to begin with a division between literary
theorists who have drawn on broad methodological principles
and literary theorists who have taken up particular linguistic
theories. For example, my own work on literary universals
(Hogan 2003 ) does not rely on any particular account of lan-
guage universals. However, I do follow the general methodologi-
cal principles for isolating genetic and areal distinctness (though
see areal distinctness and literature ), distinguishing
diferent varieties of statistically unexpected cross-cultural pat-
terns, and so on.
Tis type of infuence, however, is the exception rather than
the rule. Other writers have drawn on particular theories of
language, using them literally or analogically to explore liter-
ature. Taking up the structure set out in the Preface, we may
distinguish neurobiological, mentalistic, and social theories,
as well as theories that bear on acquisition and evolution.
Mentalistic theories have been the most prominent. As noted
in the Preface, within mentalistic theories, we may distinguish
intentionalist accounts of language (referring to human subjec-
tivity and intention) from representationalist accounts (treat-
ing algorithmic operations on mental symbols). Intentionalist
The Cambridge Encyclopedia of the Language Sciences
66
me please, Im trying to cook. I havent got enough potatoes
(Biber, Conrad, and Reppen 2006 , 69); and 2) the following
excerpt from Shakespeares Sonnet 97 How like a winter hath
my absence been/From thee, the pleasure of the feeting year!
Te frst piece of speech calls to mind a particular, active con-
text. Te second is removed from any such context. Perhaps,
then, works of verbal art are more separable from their imme-
diate context. As a frst approximation, we might say that verbal
art has other verbal art as its primary, distinctive context, and
the current material context has largely inhibitory force. In other
words, our understanding and evaluation of a work of verbal art
are distinguished from other sorts of understanding and evalua-
tion, frst of all, by their relation to other works of verbal art our
sense of storytelling techniques, our awareness of the larger story
cycle in which a particular narrative occurs, our expectations
about characters, and so forth.
Tis does not mean that verbal art is entirely insensitive to
immediate context. Our response to stories may be infected, pri-
marily in an inhibitory way, by current material circumstances.
Consider jokes, a form of verbal art. One standard type of joke
is ethnic. Te standard formats of such jokes (e.g., How many
xs does it take to screw in a light bulb?), the general function of
jokes, and so forth provide the broad, distinctive context for inter-
pretation and response. Te most obvious function of the immedi-
ate, material context comes when a member of the relevant ethnic
community is present and that function is usually inhibitory.
Removal from immediate context cannot be quite the crucial
property, however. Consider, for example, the present essay.
Physically, I am alone as I am writing. Anyone who reads this
essay will be removed from the material context in which I am
writing, and that material context will be irrelevant to the readers
response and understanding. But that does not make this essay
verbal art. Perhaps, then, the most important diference between
the aforementioned speech actions is not context per se but
something closely related to context. Te joke suggests that this
has something to do with the audience. Perhaps the diference is
a matter of the way the speaker addresses his or her audience.
To consider this, we might return to those cases. It is clear that
the person who says, Go away, Im cooking, is talking to his or
her audience. I, too, am addressing my readership in writing this
essay even if my idea of that readership is rather amorphous.
But the sonnet, as a socially circulated poem, is not addressing
its readership. Even if Shakespeare initially drafted the poem as
a private message to a particular person, he made the decision
that its readership would not be its addressee when he made it
a public poem .
More exactly, works of verbal art tend to be marked by indi-
rect address , rather than direct address . When considered from
the perspective of the reader rather than the author, indirect
address is roughly the same as side participation, as discussed
by Richard Gerrig and Deborah Prentice (1996). Gerrig and
Prentice distinguish several possible roles in a conversation.
Obvious roles include speaker, addressee, and overhearer. Te
authors add a fourth role side participant. Suppose I am with
my wife at the grocery store. She sees von Humboldt, a colleague
of hers. Te colleague says, Oh, that meeting with de Saussure,
our new provost, is on the twelfth. When she says the new pro-
vost she is doing so for my beneft. My wife knows perfectly well
of conversational implicature , as well as various ideas of
Mikhail Bakhtin (see dialogism and heteroglossia ). In
terms of larger groups, literary theory has been highly infuenced
by historian Michel Foucaults ideas about discourse and social
power (see discourse analysis [foucaultian] ). Tey
have also drawn on the sociological ideas of Pierre Bourdieu
(see field; market , linguistic; and habitus , linguistic )
and others. Broader social movements, such as Marxism and
feminism (see marxism and language , gender and lan-
guage , and sexuality and language ) and their associated
theories, have also contributed importantly to literary discus-
sion, though not necessarily in a way that bears particularly on
language science .
Te literary use of most social theories has tended to be
instrumental. However, more narrowly sociolinguistic analy-
ses of literature have been integrated into research programs
in sociolinguistics. Tis is largely because, here too, the areas of
research overlap, and language scientists have been involved in
research along with literary interpreters. For example, William
Labovs studies of oral narrative and the researchers of writers in
corpus linguistics have contributed to the advancement of
both sociolinguistics and literary study.
Te same general point holds for the study of acquisition.
Tere has been valuable work done on, for example, the acqui-
sition of metaphor and the development of verbal
humor .
Finally, evolutionary theory has inspired many literary theo-
rists in recent years (see verbal art, evolution and ). Its
advocates propose sweeping evolutionary explanations for a
wide range of literary phenomena. It is not clear that this pro-
gram has gone beyond the stage of conjecture. In any case, it is
general and often not tied specifcally to language.
As I have suggested, much work in the preceding areas is
very signifcant. However, much of it begins, like the classical
European epic, in medias res. It does not set out a clear feld of
study for a language science of verbal art. Rather, the most suc-
cessful work tends to focus on those areas of verbal art that fall
within the purview of nonliterary research programs. Put difer-
ently, it tends to treat verbal art as a case of something else (e.g.,
cognitive metaphor or phonology). In the remainder of this essay,
then, I do not explore these particular approaches and connec-
tions in more detail. Such a review is, in any case, redundant, as
this material is covered in the following entries. Instead, I con-
sider what is distinctive about verbal art and why, as a result, it is
an important area of study for language science .
THE PARTIAL AUTONOMY OF VERBAL ART: INDIRECT
ADDRESS, SIDE PARTICIPATION, AND PLAY
Perhaps the most obvious diferentiating characteristics of verbal
art are that it is normative and rare. While all cultures have verbal
art (see Kiparsky 1987 , 1956), few people in any culture produce
works of verbal art (though, of course, they do produce many con-
stituents of such works novel metaphors, allusions, wit, and so
forth). On the other hand, these diferences are not defnitive in
themselves. Rather, they seem to result from other factors.
Consider two samples of speech actions: 1) the following
excerpt from recorded speech Go away, Im cooking. Excuse
Elaborating Speech and Writing
67
suppose Jane is in third grade and Jill is in second grade. Jane
begins to play a third grade teacher. She starts by referring to
what the class did last week. However, she has to keep in mind
that Jill does not know what the third grade class did last week.
Tus, she may have to say Now, class, you remember that last
week we began by discussing government and binding
theory. As this suggests, play is a purer case of indirection than
ordinary side participation. In the case of side participation,
our adjustments are more likely to be explicit. For example, my
wifes colleague is likely to turn to me when explaining that de
Saussure is the new provost. In play, explicit adjustments occur
when the pretense of play is disrupted. Suppose Jill accidentally
addresses Jane as Jane, then explicitly adjusts that to I mean,
Frau Doktor Wittgenstein. Jane is likely to get annoyed with this
breach in the play, perhaps responding, Jill, you should never
call your teacher by her frst name. Now you have to sit in the cor-
ner and wear the Aversive Stimulus Operant Conditioning Cap.
Following this usage, we may say that verbal art is a form of
indirect speech action in the same sense as one fnds in play.
Indeed, verbal art is, in certain respects, a form of play. Note,
however, that in verbal art, in play, and elsewhere, there is not an
absolute separation between direct and indirect action. Again,
indirect action reduces extrinsic factors. It does not eliminate
them. Te diference is one of degree.
Indeed, the diference between extrinsic and intrinsic is not
fne grained enough to clarify the distinctiveness of verbal art.
Tis becomes clear as soon as we notice that almost all writing
is predominantly intrinsic in that it is not generated or adjusted
primarily by reference to external circumstances. Moreover,
the opportunities for extensive revision in writing allow for virtu-
ally all adjustments to be implicit in the fnal text. How, then, do
we distinguish ordinary writing from writing that is verbal art?
Here, too, the distinction is a matter of degree. Writing
involves diferent gradations of removal from direct address.
Take, for example, a letter. Suppose von Humboldt is not speak-
ing to my wife about a faculty meeting, but is instead writing her
a note. In an obvious way, von Humboldts address to my wife is
indirect and intrinsic. After all, my wife is not present. However,
von Humboldts address to my wife is direct in another way, for it
is oriented precisely toward her. It remains guided by her as von
Humboldt imagines her.
In order to understand how this works, and how it bears on
verbal art, we need to have a clearer understanding of imagina-
tion and action. When I begin to act, I tacitly project or imagine
possible outcomes for my action. For example, when I see a car
coming toward me, I run to the curb. Tis is bound up with a tacit
imagination of where I should be in order to fulfll my purpose
of not being run over. More exactly, we may understand imagi-
nation, like speech, as involving two (continuously interacting)
processes. One generates possible scenarios. Te other makes
adjustments. I project running to the curb, but then notice a man-
hole in my path. Tis leads me to adjust my imagination of the
precise trajectory. Te nature of the generation and the nature of
the adjustment will change, depending on the guiding purposes
of the action. In some cases of speech action, the purpose involves
a real addressee. In some cases, it does not. In a face-to-face dia-
logue, a speaker will continually generate, adjust, and regener-
ate what he or she is saying in part due to the addressees actual
who de Saussure is. Te conversation does not really concern
me. If I had been standing a few feet away, I would have merely
been an overhearer and von Humboldt would have said only,
Oh, that meeting with de Saussure is on the twelfth. But since I
was closer, I became a side participant and von Humboldt had to
take my knowledge and interest into account when speaking.
Te dif culty with the account of Gerrig and Prentice is that
it is, for the most part, analogical. It suggests an area of research
and theorization. However, it does not develop this in algorith-
mic specifcity (i.e., spelling out how it will work, step by step)
in relation to the structures, processes, and contents of human
cognition. To explore the idea further, we might consider a sim-
ple model of speech (oral or written). Tis model begins from
the premise that language is a form of action. Tus, it has the
usual components of action goals, motivations or emotions,
anticipated outcomes, and so forth.
Here, I wish to isolate two stages in the production of speech.
One is the basic generative phase in which the initial utterance
is formulated. Te second is an adjustment phase, which follows
our realization of just what we are saying. Intuitively, we might
expect that awareness of speech would precede generation.
But it does not. Indeed, a moments refection suggests why. In
order to realize that I am about to say Hello, I must in some
sense have already generated the Hello, even if my lips have
not yet moved. More importantly, empirical research indicates
that human action generally involves just this sort of duality. For
example, as Henrik Walter points out, our brains initiate or proj-
ect actions approximately .5 to .7 seconds before the actions are
performed. We are able to modify or inhibit the action .3 to .5
seconds after it is projected (see Walter 2001 , 24850). In keeping
with this temporal sequence, adjustment may precede, interrupt,
or follow the execution of an action.
Suppose someone asks me if I need a ride. I begin to say, No,
thanks. Duns is picking me up. I realize that the person does not
know who Duns is. I may adjust the sentence before speaking
My cousin Duns is picking me up. Or I may succeed in making
the change only after beginning the sentence Duns u h, hes
my cousin hes picking me up. When the adjustment takes
place before speaking, we might refer to it as implicit. When
it occurs in the course of speaking or after speaking, we may
refer to it as explicit. Finally, it is important to note that actions
have two broad sources. One is extrinsic , or externally derived;
the other is intrinsic , or internally derived (see MacNeilage 1998 ,
2256 on the neural substrates for this division).
Now we are in a position to clarify the nature of indirect
address or, more generally, indirect action. Indirect action,
as I am using the phrase, involves a relative decrease in the
extrinsic aspects of action. Tus, the sources of both generation
and adjustment are more intrinsic than is commonly the case.
Moreover, when they occur, extrinsic adjustments tend to be
implicit rather than explicit.
To get a better sense of how indirect action operates, it is use-
ful to look at a paradigmatic case of such action play. Indeed,
play and side participant interaction have a great deal in com-
mon. When Jane and Jill play school, each of them must keep
in mind that the other person has a real identity outside the role
she is playing. Moreover, each of them must continually adjust
her speech and behavior to take that into account. For example,
The Cambridge Encyclopedia of the Language Sciences
68
Tis combines with pragmatic information about the presence of
cookies in the cookie jar to initiate the action of getting a cookie
out of the cookie jar. However, before I reach for the cookie jar,
prudential information stops me as I recall the deleterious efects
that cookies are likely to have on my lithe and shapely form .
For imaginations or actions to result from pragmatic infor-
mation, there must be some motivation present as well. In other
words, I must have some emotion. Emotion is what leads us to
act, whether the action is speech or larger bodily movement. It is
also what leads us to refrain from acting. We may divide emotions
into two sorts, depending on their function in action sequences.
Te frst might be called the initiating emotion. An initiating
emotion is any emotion that one feels in current circumstances
and that impels one to act. Action here includes the initiation of
movement and the initiation of imagination. We imagine various
outcomes of our actions. For example, feeling hungry, I swiftly
and unrefectively imagine eating a cookie. But my imagination
does not stop there. I may imagine my wife seeing me with the
crescent of cookie in my hand and the telltale crumbs sticking to
my lips, then chastising me, reminding me of the doctors warn-
ings, and explaining once again that the cookies are for my nieces
and nephew. Or I may suddenly see an image of myself with wob-
bly love-handles. In each case, I experience what might be called
a hypothetical emotion. A hypothetical emotion is a feeling that
I experience in the course of imagining the possible trajectories
of my action. While initiating emotions give rise to (or generate)
the action sequence initially, hypothetical emotions qualify (or
adjust) that action sequence. Hypothetical emotions may inten-
sify the initiating emotion; they may inhibit it; they may respecify
the precise goals of the action sequence (e.g., from eating cook-
ies to eating carrots), or they may afect the precise means that I
adopt (e.g., checking that my wife is not around).
From the preceding example, it may seem that hypothetical
emotions are all egocentric. However, hypothetical emotions
may also be empathic. For example, I may forego my plan to eat
cookies because I imagine my tiny nephews disappointed face
when he reaches into the empty jar. Empathic hypothetical emo-
tions may have the same qualifying efects on initiating emotions
and actions as do egocentric hypothetical emotions.
Hypothetical emotions are critical for all types of action,
including verbal action. Consider again the case where von
Humboldt explains that de Saussure is the new provost. In doing
this, von Humboldt tacitly imagines the conversation from my
perspective and, sensing that I may not follow and, more impor-
tantly, that I might feel left out she provides the information.
If I wish to alter someones emotions through speech, I will
appeal primarily to initiating emotions. Tus, the alteration of
initiating emotions is usually a central purpose of speech action.
However, certain sorts of hypothetical emotions are important
as well. For example, when requesting a favor, I may foreground
how grateful I will be. One purpose is to encourage my addressee
to imagine my gratitude and experience the related hypothetical
emotion (roughly, feeling appreciated) .
In sum, ordinary speech actions involve an informational
aim and an emotional aim, usually coordinated to produce some
change in the world, currently or at some time in the future. Te
informational aim involves both pragmatic and regulatory com-
ponents, though in ordinary speech, the pragmatic component
response. In writing a letter, a writer will be guided by his or her
tacit imagination of the addressee. Tough that imagination is not
adjusted by reference to actual responses, it nonetheless serves as
a guide for the generation of the speech. Crucially, this imagined
addressee maintains his or her external, independent existence
for the speaker. As such, that addressee is, so to speak, a pseudo-
extrinsic guide to the generation of the speech.
Here, then, we may adjust our account of indirect address
in verbal art. Verbal art minimizes both extrinsic and pseu-
doextrinsic elements in the production and adjustment of
speech and in the imaginations that are connected with speech.
Moreover, it minimizes explicit markers of adjustments for side
participants.
We began with the idea that verbal art is relatively indepen-
dent of direct, material context. Tat context involves authors
and readers. In this connection, we have been considering
the relation between the author and readers as one of indirect
address. Te other crucial element of material context is, of
course, reference. Indeed, verbal art, as commonly understood,
has an even more obvious relation to reference than to address,
for verbal art is paradigmatically conceived of as fction. As such,
it has an unusual degree of referential autonomy. In other words,
it tends to be characterized by indirect reference as well as indi-
rect address. Tis, too, is illustrated most clearly in play. Suppose
Jane and Jill are playing telephone. Jane picks up a banana, puts
it up to the side of her face, then holds it out to Jill and says, Its
for you. In saying it, she is not referring to a banana. She is
referring to a telephone.
In sum, verbal art is not just partially autonomous with
respect to immediate circumstances. It is largely independent of
the extrinsic and pseudoextrinsic aspects of the generation and
adjustment of speech action and associated imagination. Tis is
true with respect to both address and reference .
ART AND ACTION: THE PURPOSES OF VERBAL ART
In discussing action-related imagination, we noted that the pre-
cise nature of imagination varies according to the purposes of the
action. Generally speaking, the goals of physical action concern
the alteration of some situation. To a great extent, speech actions
have the same function. One crucial diference between speech
actions and bodily actions is that speech actions have their efects
only through minds. However much I plead with the packet of
noodles, they wont turn themselves into pad thai. But, if I sweetly
ask someone to make the noodles into pad thai, perhaps that
person will do so. Speech actions, then, tend to aim at altering
the world by altering peoples actions. In order to alter peoples
actions, they aim at altering two things frst, the way those peo-
ple understand the world and, second, the way they feel about it,
including the way they feel about the speaker himself or herself.
More exactly, we may distinguish two psychological pur-
poses of speech actions. Tese are informational and emotional.
Informational purposes may be further divided into pragmatic
and regulative. Pragmatic information concerns factual, direc-
tional, or other information that facilitates our pursuit of goals.
Regulative information concerns broad ethical, prudential, or
related principles, which tend to serve an adjusting function.
For example, feeling hungry, I form the goal of eating something.
Elaborating Speech and Writing
69
and integrated with other information; see encoding ) and
may even draw attentional focus . Tis occurs most obviously
in the phonetic/phonological aspect of language. Poets pattern
stress beyond what occurs spontaneously. Tey organize vowel
sounds and initial consonants to produce assonance and allit-
eration . Te point is less clear in morphology, though corpus
linguistic studies have pointed to diferential tendencies (see
Biber, Conrad, and Reppen 2006 , 5865)
Syntax is a complex and thus particularly interesting case.
Verbal art does undoubtedly tend to pattern syntactic usage in
encodable ways. But the poetic patterns in these cases tend to
be intensifcations of the ways in which ordinary speech actions
pattern syntax. For example, in ordinary speech, we have some
tendency to use parallel syntactic structures. In poetry, we are
more likely to use these. A more obtrusive foregrounding of syn-
tax occurs when we violate expectations of patterning in verbal
art. Such violation is the other primary way in which verbal art
treats linguistic levels more autonomously. In the case of syn-
tax, there are several ways in which this may occur. One obvious
way is through disjoining syntactic units and line units, such that
line breaks do not coincide with syntactic breaks (see Tsur 2006 ,
14652). Another is by rejecting standard principles of word
order.
Verbal art also manifests distinctive tendencies in semantics.
Tese are found most obviously in lexical preferences and meta-
phor. In the case of lexical preferences, verbal art may draw on
rarer or more unexpected words. Perhaps more importantly, it
may pattern the suggestions and associations of terms. In ordi-
nary speech, we tend to concern ourselves with the associative
resonance of a term only in extreme cases (e.g., when its con-
notations may be ofensive to a particular addressee). In verbal
art, however, the writer is much more likely to organize his or her
lexical choices so that the suggestions of the terms are consistent
(e.g., in emotional valence; see dhvani and rasa ) .
As to metaphor, our use of tropes in ordinary speech is surpris-
ingly constrained. Tere are aspects of everyday metaphor that
are creative. However, on the whole, we follow well-worn paths.
Lakof and Turner have argued that a great deal of literary meta-
phor draws on the same broad structures as ordinary speech.
However, literary metaphors extend, elaborate, and combine
these structures in surprising ways (see poetic metaphor ).
Finally, we fnd parallel tendencies in discourse practices.
Consider the principles of conversation articulated by Grice
(see cooperative principle ). Tese principles form a set of
practical conditions for any sort of interaction between speak-
ers. For example, it is a fundamental principle of conversation
that one should not say things that are irrelevant to the con-
versation. Grice points out, however, that one may fout these
principles, violating them in gross and obvious ways. Flouting a
principle of conversation gives rise to interpretation. Jones and
Smith are discussing who should be hired as the new assistant
professor in the Hermeneutics of Suspicion. Jones says, Teres
a new applicant today Heidegger. What do you think of him?
Smith replies, Nice penmanship. Since penmanship is clearly
irrelevant to the topic, Smith may be said to be fouting the prin-
ciple of relevance. Jones is likely to interpret Smiths comment
as indicating a dim view of Heideggers qualifcations. Literary
works often fout conversational principles.
is probably dominant. Te emotional aim involves both initiat-
ing and hypothetical emotions. In ordinary speech, the initiating
emotion is probably dominant .
In being removed from direct address and direct reference,
works of verbal art commonly do not have a goal of directly alter-
ing particular material conditions. Nonetheless, verbal art is ani-
mated by the same two psychological goals that animate other
forms of speech action. Verbal art, too, has an informational
component and an emotional component. Indeed, cross-cultur-
ally, aestheticians and critics tend to see verbal art as successful
insofar as it afects us emotionally and insofar as it develops sig-
nifcant ideas or themes.
Te emphasis in verbal art, however, tends to be diferent
from that of other speech actions in both cases. Specifcally, ver-
bal art commonly does not stress pragmatic aspects of informa-
tion. Certainly, there are, for example, political works that set
out to give the audience pragmatic information. However, this
is not the general tendency of verbal art. Rather, verbal art tends
to develop regulative concerns. Tese regulative concerns make
their appearance in literary themes. When we interpret a work
and seek to understand its point, we are commonly looking
for a theme, which is to say, some sort of regulative informa-
tion. Moreover, in terms of emotion, verbal art tends to inspire
not initiating emotions but hypothetical emotions particularly
empathic hypothetical emotions .
THE MAXIMIZATION OF RELEVANCE
Up to this point, I have been speaking of the content of speech
action generally and verbal art in particular. But the form of ver-
bal art is widely seen as crucial, perhaps its defnitive feature. Te
content/form division is somewhat too crude to form the basis
for a sustained analysis of verbal art. However, it does point to
an important aspect of verbal art, and an important diferentiat-
ing tendency the partial autonomy and patterning of distinct
linguistic levels in verbal art.
All linguistic levels phonology , morphology , syn-
tax , and so forth are, of course, relevant to speech actions of
all types. However, most components are, in most cases, only
instrumentally relevant. Morphology is relevant for communi-
cating whether I want one cookie or two cookies. But it has no
separate function. Put diferently, the specifcation of phonology,
morphology, and syntax is a sort of by-product of my pragmatic
aims. I want to get two cookies. In English, I signal that I want
two cookies rather than one by using the word two, putting
that word before cookie, adding s to cookie, and so forth.
I do not set out to do anything with phonology, morphology, or
syntax. One might even argue that I do not set out to do anything
special with semantics or with discourse principles. I just run
my language processors to achieve the active goal. Features of
phonology and so forth are relevant to my action only insofar
as my speech processor makes them relevant. Tey have, in this
way, minimal relevance.
A number of literary theorists, prominently the Russian
Formalists, have stressed that verbal art foregrounds its lan-
guage (see foregrounding ). In cognitive terms, we might say
that verbal art tends to enhance linguistic patterns to the point
where they are likely to be encoded by readers (i.e., perceived
The Cambridge Encyclopedia of the Language Sciences
70
attention on the friend and simultaneously directs my attention
to ways in which I can reach him. If I see some sort of animal, I
feel fear. Tat fear keeps my attention on the animal and simulta-
neously directs my attention to ways that I can escape .
Te arousal of interest is obviously crucial to literary experi-
ence. It is also important to other speech actions. However, in
nonliterary cases, a great deal of the interest comes from the
direct relation between the people involved (thus, the speaker
and addressee), as well as the practical situation referenced in
the speech action. In other words, a great deal comes from direct
address and direct reference, both of which are greatly inhibited
in verbal art. Nonetheless, verbal art has other means of produc-
ing interest. We have just been discussing two such means the
multiplication of non-normal patterns and violations of normal
or expected patterns at various linguistic levels. Both are precisely
the sorts of deviation from normalcy that produce interest.
Specifc emotional efects are fostered most obviously by
semantics and pragmatics. For example, a great deal of our
emotional response to verbal art seems to be bound up with the
patterning of associative networks that spread out from lexical
items (see Oatley 2002 and suggestion structure ). Te
extension and elaboration of metaphorical structures are clearly
consequential in this regard, particularly as the concreteness
of metaphors often enhances associations with concrete expe-
riential memories, including emotional memories. Te specifc
emotional impact of phonological, morphological, and syntactic
features is less obvious, but no less real, at least in some cases.
For example, the interplay between nonlinguistic organization
(e.g., in line divisions) and linguistic organization (e.g., in sen-
tence divisions) may serve to convey a sense of a speakers voice
and, along with this, an emotional tone.
Tus, once more, we see both continuity and diference
between verbal art and other sorts of speech action. Most speech
actions involve a minimal, incidental patterning of linguistic lev-
els. In contrast, the general tendency of literary art is toward the
maximization of the relevance of diferent linguistic levels. Tis
relevance is a function of the main purposes of the text, thematic
and emotional. Again, this need not be relevance that increases
informational content. Indeed, I suspect that it most often is not.
It need not even contribute to the specifc emotions of the text.
It may be a matter of enhancing interest or of qualifying the spe-
cifc emotions fostered by the text. In each case, though, there is
some partially autonomous variation in the organization of the
linguistic level in question, through the addition of unexpected
patterning, through the violation of expected patterning, or both .
ON INTERPRETATION AND THE USES OF TEXTS
In the preceding section, I referred briefy to the point at which
interpretation ends. When verbal art is considered in relation to
interpretation, the frst thing to remark is that verbal art is notori-
ous for its hermeneutic promiscuity (see philology and her-
meneutics ). It is widely seen as almost infnitely interpretable.
At one level, this is surprising. Novels, for example, are devel-
oped in great detail and with considerable elaboration of the
characters attitudes and actions. It might seem that this would
constrain interpretation relative to the much vaguer and more
elliptical speech actions of ordinary life. But that is not generally
All these points indicate signifcant diferences between ver-
bal art and other sorts of speech action. Again, these diferences
do not create a hard-and-fast division between verbal art and
all other types of speech. Tere is a continuum here, with many
parameters and degrees of variation. Nonetheless, there is a clear
diferential tendency.
Te diferences we have just been considering are a matter
of the various linguistic levels of a literary text bearing autono-
mously or separately on our understanding and experience of
the work. Tis is frst of all and most obviously a matter of creat-
ing patterns. However, given the violations of syntactic rules and
the fouting of conversational principles, it seems clear that ver-
bal art does not simply create extra patterns on top of the usual,
instrumental patterns produced by ordinary language pro-
cesses. It also violates patterns of ordinary language processes.
Most importantly, in both cases, the result renders the linguis-
tic level in some way directly (rather than just instrumentally)
relevant to our experience of the work. Tus, it maximizes the
relevance of language features.
But in what way are these features relevant? As with any sort
of speech action, relevance is, frst of all, relevance to the aims of
the action. Again, the primary aims of verbal art are thematic and
emotional. Tus, the maximization of relevance is the thematic
or emotional use of (ordinarily unexpected) noninstrumental
patterns or violations of (ordinarily expected) instrumental
patterns from diferent linguistic levels.
In touching on these issues, literary critics have tended to
emphasize thematic relevance. Some writers have seen patterns
and violations of patterns in phonology, morphology, and syntax
as consequential for interpretation. It is probably true that such
formal features are thematically interpretable in some cases.
However, it seems doubtful that such features are generally rele-
vant to interpretation. In contrast, extended patterns or violations
in semantics and pragmatics are almost always interpretively
relevant. In these aspects, the main diference between verbal art
and other forms of speech action is where the interpretive pro-
cess ends. We will consider this issue later .
If anything, the maximization of relevance applies more fully
to the communication of emotion than to the communication of
themes. Tere are two types of afective response that bear impor-
tantly on verbal art, and thus on the maximization of relevance. We
might refer to the frst as pre-emotional and the second as specifc
emotional . Pre-emotional efects are efects of interest. Interest is
pre-emotional in two senses. First, it is often an initial stage in an
emotion episode. Second, it is a component of all specifc emo-
tions after those emotions have been activated. Specifc emotions
are simply our ordinary feelings sorrow, joy, disgust, and so on.
More exactly, interest is the activation of our attention system.
Tat activation occurs whenever we experience something new
or unexpected (see Frijda 1986 , 2723, 318, 325, 386). Such acti-
vation prepares us for events that have emotional signifcance.
It directs our attention to aspects of the environment or our own
bodies that may be emotion triggers. Once a specifc emotion is
activated, that emotion system reactivates our attention system,
focusing it on properties relevant to that emotion in particular.
For example, suppose I am out in the woods and hear something
move. Tat arouses my attention. I carefully listen and look. If
I see a friend at a distance, I feel some joy. Tat joy keeps my
Elaborating Speech and Writing
71
someone for the frst time, I may describe my outft so that he
or she can recognize me. It has pragmatic consequences. But
if an author describes a characters outft, that description may
bear on our understanding of the characters emotions, class
status, or religious beliefs. Tose features may in turn bear on
our broader understanding of human relations, class confict,
or religious practice as portrayed in the work. In short, ordi-
narily incidental details may have thematic (thus regulative)
consequences.
Moreover, literary narratives tend to develop subtle and vari-
able af nities and ambivalences (see Hogan 2003 , 12251). In
some cases, the development of these af nities actually runs con-
trary to the authors self-conscious sense of his or her own aims.
Te most famous case may be Miltons portrayal of Satan. Satan
has often been seen as the most engaging fgure in Paradise Lost ,
but this was certainly not Miltons self-conscious intention.
Tis is part of a larger point that intention is not a single, uni-
fed operation. Tere are diferent sorts of intention with diferent
objects, constraints, and processes. Perhaps the most important
form of intention for verbal art is what we might call aestheti-
cal intent (see Hogan 1996 , 16393). Te aesthetical intent of an
author is to produce a work that has the right sort of experiential
efect. Tis right efect is not something that the author is likely
to be able to articulate separately from the work itself. It is simply
what he or she experiences when he or she feels that the work
is now complete. In composing the work, the author generates
and adjusts the text, testing the outcome against his or her own
response. Te authors sense that the work is complete need not
mean that the work conforms to the authors self-conscious atti-
tudes and commitments.
One way of explaining aesthetical intent is that the author
adopts an aesthetical attitude toward the work, or a dhvani atti-
tude, as Anand Amaladass put it. Tis is a matter of broaden-
ing ones attention to the work, expanding ones encoding of the
work, to include its multiple resonances and emotive patterns.
In short, it involves approaching the text as a work in which dif-
ferent linguistic levels are maximally relevant. Correlatively, it
involves approaching the text as a work that is removed from
constraints of direct reference or address, constraints that ordi-
narily orient our judgment of informational relevance and our
construal of emotional bearing. When approaching a work as
verbal art, readers and audience members adopt this attitude as
well to varying degrees .
Te mention of readers approaches to texts brings us to a
fnal complication. We have been considering ways in which
works produced as verbal art tend to be diferent from other
speech actions. But the nature of a given text is not determined
solely by authorial intent. Despite our prototypical concern for
authorial intent, we are free to approach works of verbal art in
pragmatic ways and to approach other works with an aesthetical
attitude. As writers in cultural studies have stressed, practices of
literary analysis may be applied to a wide range of texts that were
not initially intended to be literary. In short, literariness may
be defned by interpretation or reception no less than it may be
defned by production; it may be defned by readers no less than
by authors. In this way, the usual characteristics of verbal art may
be extended to other texts or withdrawn from texts (as when a
novel is studied for its authors psychology).
believed to be the case. Tis suggests that the degree of interpret-
ability of a text is not a function of its elaboration. Rather, inter-
pretability appears to be a function of a texts distance from direct
address and direct reference to practical conditions, particularly
as these are connected with pragmatic information. Put difer-
ently, the limits of interpretation are not so much a matter of
the words themselves. Tey are, frst of all, a matter of action .
Here as elsewhere, action is animated by the agents goals,
emotions, expectations, and so forth. In ordinary life, then, we
usually understand validity in interpretation as a function of
the speakers intention (see intentionality ). More techni-
cally, our prototype of interpretive validity almost certainly
includes the intention of the speaker or author as a critical norm.
(Te point is related to the argument of Steven Knapp and Walter
Benn Michaels that we invariably interpret for speakers inten-
tion, though it is not identical, for there are cases where we do
not adhere to this prototype and thus do not interpret for the
speakers intention.) However, intention is an abstract criterion.
We do not have access to speakers intentions. So even interpret-
ing for intention, we need an operational criterion for validity
as well. Tat is where our own action enters. Action commonly
guides our sense of when we have gotten an interpretation right.
At the dinner table, someone says, Could you pass the that? I
am not sure what he means by that. Unable to read his mind, I
engage in an action either a bodily action (passing the beans) or
a speech action (asking if he meant the beans). When I pass the
beans, he knows that I have understood his intention. When he
accepts the beans, I infer that I understood his intention.
Te interpretation of verbal art is as removed from such prac-
tical action as possible. Tus, our ordinary operational criterion
is rendered inefective. Te only obvious practical behaviors
relating to literary interpretation are professional for example,
the acceptance or rejection of articles in academic journals.
(Te point is codifed in Fishs contention that validity in liter-
ary interpretation is defned by the norms of interpretive
communities .)
A number of further problems for intentional inference arise
in connection with this removal of literary interpretation from
practical action. First, many texts are read or performed, and
thus interpreted, far from their authors and even after their
authors are dead. If we take a strict intentionalist view of valid-
ity, then the author is the only one who has the authority to
determine that a given action does indeed satisfy an operational
criterion. Suppose Jones leaves instructions for his funeral. Te
funeral director follows them as well as she can. But Jones is not
around to confrm that she got things right. Obviously, there
are things that she might do to ascertain Joness intention. For
example, she might talk to Joness friends and relatives or she
might try to learn something about Joness religion and ethnic
background. Tese are the sorts of concerns that lead writers
such as Hans-Georg Gadamer to stress tradition as a crucial
guide to interpretation .
A second problem is more distinctively connected with ver-
bal art per se. Both informational and afective patterns are
more complex in verbal art than in most other speech actions.
For example, a literary work communicates thematically rel-
evant information by maximizing the relevance of virtually
every semantic and discursive detail in the text. If I am meeting
The Cambridge Encyclopedia of the Language Sciences
72
in play, be summer (e.g., a time of professional success for the
speaker). In short, summer becomes a metaphor.
Te mention of metaphors points us toward the maximization
of relevance. However, before going on to this, it is important to
remark on some other contextual features of the poem, contex-
tual features that are themselves bound up with metaphor. As I
indicated earlier, the entire tradition of poetry forms an implicit
context for any given poem (a point stressed by T. S. Eliot). One
obvious way in which Shakespeare suggests a context of verbal
art is in his use of seasonal imagery. Winter is a standard literary
image of romantic separation. Tis is, of course, related to the
conceptual metaphor , discussed by Lakof and others: LIFE
IS A YEAR. As Lakof and Turner explain, there are several pro-
cesses poets use to create novel instances of these schemas .
Shakespeare is clearly creating such a novel instance when he
maps the source metaphor of winter onto the target, summer
(see source and target ). Moreover, in making summer into
(metaphorical) winter, he is intensifying the efect of the meta-
phor by contrast. But he does not say summer only. He expands
the target time to summer and autumn. In a way, this is pecu-
liar. Te contrast would have been more obviously enhanced
by making the target time period spring and summer. Why does
Shakespeare choose autumn? Tere are several reasons. Two
are closely interrelated. First, he wishes to intensify the tacit
emplotment of the speakers isolation. Tat speaker is now
anticipating the very worst, for if summer was like winter, how
devastating will winter be? Second, he wishes to hold out hope
for something beyond that period of terrible loneliness spring.
Tat hope is possible only through an indirect reference to the
broader, literary context in which spring is the season when the
lovers are fnally reunited.
Te indirection of the poem makes it unlikely that it will
have any pragmatic information as its aim. Pragmatic informa-
tion most often involves reference to particular situations, or to
forms of general knowledge. Of course, there is not a strict divi-
sion between general pragmatic knowledge and, say, prudential
regulatory information. But to the degree that the poem com-
municates thematic points, those points do incline toward pru-
dential regulatory information. For example, suppose summer
is metaphorical (i.e., winter is metaphorical for summer,
which is itself metaphorical for, say, professional success). Ten
the poem suggests that the enjoyment of external conditions
(e.g., professional success) is lost when it cannot be shared a
point with clear regulatory consequences.
Te indirection of the poem also indicates that it is unlikely
to be aimed at the elicitation of initiating emotions. Tis can be
seen if we contrast it with a similar letter, sent by the poet to his
beloved. Such a letter could be aimed at convincing the beloved
to return home. Te poem, in contrast, does not have any such
initiating emotional aim. Its emotional aim is, rather, confned to
provoking empathic hypothetical emotions.
Te maximization of relevance contributes to the poems
achievement of its primary aims. As usual, this maximization is
most obvious at the phonological level. Consider only the frst
two lines. Te poem is in iambic pentameter. However, when
spoken naturally, these lines do not follow the meter. Indeed,
there is considerable tension between spontaneous stress
On the other hand, the expansion of hermeneutic liberality
results from the existence of verbal art. Interpretive autonomy
arises in the frst place through the removal of speech action
from direct address and direct reference, along with the attenu-
ation of pragmatic information and initiating emotion. In this
way, interpretive practices that bridge literary and nonliterary
speech actions are themselves a distinctive product of literary
speech action.
SHAKESPEAREAN INDIRECTION
As the preceding discussions have been rather abstract, it is
valuable to end with a more developed example. Consider
Shakespeares Sonnet 97:
How like a winter hath my absence been
From thee, the pleasure of the feeting year!
What freezings have I felt, what dark days seen!
What old Decembers bareness every where!
And yet this time removed was summers time,
Te teeming autumn big with rich increase,
Bearing the wanton burthen of the prime,
Like widowed wombs after their lords decease:
Yet this abundant issue seemd to me
But hope of orphans, and unfathered fruit,
For summer and his pleasures wait on thee,
And thou away, the very birds are mute;
Or if they sing, tis with so dull a cheer
Tat leaves look pale, dreading the winters near.
Te poem presents a straightforward instance of indirect
address. Tere is, of course, a narratee in the poem, an explicit
thee. But the poem only plays at addressing this thee. Te
point is clear, for example, when the reader is brought in as a side
participant with the otherwise superfuous information, this
time removed was summers time. If there were a defning, mate-
rial context and if there were direct address in this speech action,
the speaker would not need to explain that the separation had
occurred over the summer. Te beloved would surely remember
this. Te point is reinforced more subtly by the shifts in spatial ori-
entation in the poem (what some writers see as particular types of
deixis [see Stockwell 2002 , 4157]). Tese may be understood as a
matter of indirect reference. If this speech action were grounded
in a particular situation, then there would almost certainly be
some fxed reference point, a specifc home that would defne
which of the lovers was stationary and which had departed. But
the poem is contradictory on this score. In the opening lines,
the speaker refers to my absence /From thee and his time
removed. But toward the end of the poem, he reverses the spatial
orientation and related direction of movement. Now, the beloved
is away and summer will wait for her return.
Te introduction of summer in this context suggests some-
thing else. Te poet is not only like little Jane, covertly explaining
what the third graders did in their last class. He may also be like
Jane in referring to a telephone by way of a banana. Put difer-
ently, the indirectness of both address and reference make the
reference of summer itself uncertain. Without a context, read-
ers are free to associate summer tacitly with anything that can,
Elaborating Speech and Writing
73
Verbal art both fosters the proliferation of interpretations
and sharpens our sense of the human embodiment that lim-
its those interpretations. We fnd the same tension between
sameness and diference in all the characteristics we have con-
sidered. Although I have been stressing diference, the same-
ness is no less consequential. For example, in the context of
an encyclopedia of language sciences, it is important that in
studying verbal art, we are likely to isolate properties and rela-
tions in language that we might otherwise have passed over
properties and relations of address, reference, informational
structure and orientation, and type and force of emotional
consequence or function. In short, verbal art is a critical ele-
ment of human life. As such, it is a critical object of study in its
own right. It is also a crucial part of human speech action. As
such it is a crucial, if sometimes neglected, part of the language
sciences as well .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Amaladass , Anand . 1984 . Philosophical Implications of Dhvani:
Experience of Symbol Language in Indian Aesthetics . Vienna : De Nobili
Research Library .
Biber , Douglas , Susan Conrad , and Randi Reppen . 2006 .
Corpus Linguistics: Investigating Language Structure and Use .
Cambridge : Cambridge University Press .
Chandler , Steven . 1991 . Metaphor comprehension: A connection-
ist approach to implications for the mental lexicon . Metaphor and
Symbolic Activity 6.4 : 227 58.
Dasenbrock , Reed Way , ed. 1993 . Literary Teory After Davidson .
University Park : Pennsylvania State University Press .
Eliot , T. S. 2001 . Tradition and the individual talent. In Te Norton
Anthology of Teory and Criticism , ed. Vincent B. Leitch , 10928. New
York : W. W. Norton .
Fabb , Nigel , and Morris Halle . 2006 . Metrical complexity in Christina
Rossettis verse . College Literature 33.2 : 91 114.
Fish , Stanley . 1980 . Is Tere a Text in Tis Class? Te Authority of
Interpretive Communities . Cambridge: Harvard University Press.
Frijda , Nico . 1986 . Te Emotions . Cambridge: Cambridge University
Press.
Gadamer , Hans-Georg . 1989 . Truth and Method . 2d ed. Trans. Joel
Weinsheimer and Donald Marshall . New York : Cr ossroad .
Gerrig , Richard , and Deborah Prentice . 1996 . Notes on audience
response. In Post-Teory: Reconstructing Film Studies , ed. David
Bordwell and Nol Carroll , 388403. Madison : University of Wisconsin
Press .
Grice , Paul . 1989 . Studies in the Way of Words . Cambridge : Harvard
University Press .
Hogan , Patrick Colm . 1996 . On Interpretation: Meaning and Inference
in Law, Psychoanalysis, and Literature . Athens : University of Georgia
Press .
. 2003 . Te Mind and Its Stories: Narrative Universals and Human
Emotion . Cambridge : Cambridge University Press .
Jakobson , Roman . 1987 . Language in Literature , ed. Krystyna Pomorska
and Stephen Rudy . Cambridge, MA : Belknap Press .
Kane , Julie . 2004 . Poetry as right-hemispheric language . Journal of
Consciousness Studies 11 .5/6: 21 59.
Kiparsky , Paul . 1987 . On theory and interpretation. In Te Linguistics
of Writing: Arguments Between Language and Literature , ed. Nigel
Fabb , Derek Attridge, Alan Durant, and Colin MacCabe , 18598.
New York : Methuen .
patterns and the meter. One natural way of speaking the lines
would be as follows:
Hw lke wntr hth m bsnce ben
Frm the, th plasre f th fetng yar!
Tere are several things one might remark on here. First, in
natural speech, the lines are in a somewhat irregular tetram-
eter. However, this irregularity is not unpatterned. Tere is a
striking rhythmic motif that occurs in the middle of both lines.
Te sequences lke wntr hth m bsnce and the, th
plasre f th fetng have the same stress pattern. Tis is not
monotonous because the lines also manifest three variations.
First, a caesura appears in only one of the sequences. Second,
the frst word of the line changes from stressed to unstressed.
Finally, the last word changes from unstressed to stressed. Tus
we fnd novelty, therefore the triggering of at least mild interest,
at two levels.
In addition, there are interpretive and specifc emotional con-
sequences. Te disjunction of syntactic and verse breaks may help
to give the poem a sense of voice. Specifcally, it suggests to me a
speaker who pauses before saying what is most painful about his
absence. Being away from home means that one is absent from
many things. But, here, there is only one crucial attachment. Te
line break imitates the emotion that so often interrupts ones
speech when such occasions are real and directly addressed.
Tere may also be efects of tempo in these lines. Hw lke
wntr is heavy with accents and thus slower than th fetng
yar. Tis provides an instance of sound echoing sense. It also
suggests an interpretive point time fies in the sense that life
passes quickly, leaving us little time together; but, being apart, we
experience each moment of that feeting time as a slow drag.
We would have many further points to interpret if we were
to consider lexical choice, metaphorical patterning, the fouting
of conversational principles, and so forth. Each of these directs
us toward the endless interpretability of verbal art. As in other
works, there is no clear operational criterion that would tell us
that we have reached the end of our interpretation or that our
interpretation is correct. Of course, as Fish indicates, there are
professional constraints on our judgments in these areas. A
psychoanalytic critic might discuss how the poet seems to shift
between the position of a separated lover and that of an orphan,
suggesting an oedipal relation to the beloved. A writer in queer
theory might stress that the poet puts his beloved in the position
of the (deceased) father, not the mother, thus suggesting a male
beloved.
Tis brings us back to interpretation and reception. Again, we
are always free to take indirect address and put it into a more
directly referential context. For example, we may seek to read
through the sonnets to Shakespeares own life and sexual feel-
ings. Indeed, as Walter Ong pointed out many years ago, we have
a strong tendency to do just that, placing the decontextualized
voice of the poem back in a human body at a particular place in a
particular time. Tat sense of concrete human embodiment is
itself no doubt a crucial part not only of literary response but also
of all human communication.
Te preceding point suggests once again the distinctive-
ness of verbal art and its continuity with other speech actions.
The Cambridge Encyclopedia of the Language Sciences
74
Ong , Walter J. , S.J. Te jinnee in the well wrought urn. In Te
Barbarian Within and Other Fugitive Essays and Studies , 1525. New
Yor k: Macmillan .
Pratt , Mary Louise. 1977 . Toward a Speech Act Teory of Literary Discourse .
Bloomington : Indiana University Press .
Shakespeare , William . 2006 . Te Sonnets , ed. G. Blakemore Evans .
Cambridge : Cambridge University Press .
Stockwell , Peter . 2002 . Cognitive Poetics: An Introduction .
London : Routledge .
Tsur , Reuven . 2006 . Kubla Khan: Poetic Structure, Hypnotic Quality and
Cognitive Style A Study in Mental, Vocal and Critical Performance .
Amsterdam : John Benjamins .
Walter , Henrik . 2001 . Neurophilosophy of Free Will: From Libertarian
Illusions to a Concept of Natural Autonomy . Trans. Cynthia Klohr .
Cambridge, MA : MIT Press .
Knapp , Steven , and Walter Benn Michaels . 1985 . Against theory. In
Against Teory: Literary Studies and the New Pragmatism , ed. W. J. T.
Mitchell , 1130. Chicago : University of Chicago Press .
Labov , William . 1972 . Language in the Inner City: Studies in the Black
English Vernacular . Philadelphia : University of Pennsylvania Press .
Lakof , George , and Mark Turner . 1989 . More Tan Cool Reason: A Field
Guide to Poetic Metaphor . Chicago : University of Chicago Press .
MacNeilage , Peter . 1998 . Evolution of the mechanism of language
output: Comparative neurobiology of vocal and manual commu-
nication. In Approaches to the Evolution of Language: Social and
Cognitive Bases , ed. James Hurford , Michael Studdert-Kennedy, and
Chris Knight , 22241. Cambridge : Cambridge University Press .
Martindale , Colin . 1995 . Creativity and connectionism. In Te Creative
Cognition Approach , ed. Steven Smith , Tomas Ward, and Ronald
Finke , 24968. Cambridge, MA : MIT Press .
Oatley , Keith . 2002 . Emotions and the story worlds of fction. In Narrative
Impact: Social and Cognitive Foundations , ed. Melanie Green , Jefrey
Strange , and Timothy Brock , 3969. Mahwah, NJ : Erlbaum .
THE CAMBRIDGE ENCYCLOPEDIA OF
THE LANGUAGE SCIENCES
77
and system of language that following generations infer often
difer from the system earlier generations are using. Tis often
results in semantic change, syntactic change , and sound
change.
Albert Atkin
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Henning . 1973 . Abductive and deductive change. Language
49 : 765 93.
Burks , Arthur . 1946 . Peirces theory of abduction. Philosophy of Science
13 : 301 6.
McMahon , April . 1994 . Understanding Language Change .
Cambridge : Cambridge University Press .
Peirce , C. S. 1935 . Te Collected Papers of Charles S. Peirce . Vol. 5.
Cambridge : Harvard University Press .
ABSOLUTE AND STATISTICAL UNIVERSALS
Language universals are statements that are true of all lan-
guages; for example, all languages have stop consonants. But
beneath this simple defnition lurks deep ambiguity, and this
triggers misunderstanding in both interdisciplinary discourse
and within linguistics itself. A core dimension of the ambiguity
is captured by the opposition absolute versus statistical uni-
versal, although the literature uses these terms in varied ways.
Many textbooks draw the boundary between absolute and statis-
tical according to whether a sample of languages contains excep-
tions to a universal. But the notion of an exception-free sample
is not very revealing, even if the sample contained all known
languages: Tere is always a chance that an as yet undescribed
language, or an unknown language from the past or future, will
provide an exception.
It is impossible, in principle, to survey all languages of our spe-
cies. If we nevertheless want to make claims about all languages,
only two routes are open: a priori deduction of necessarily true
statements or statistical extrapolation from empirical samples to
the entire set. Absolute universals can then be defned as those
that are necessarily true, statistical universals as those that are
extrapolated from samples.
Absolute Universals
For statements to be necessarily true, they must follow from a
priori assumptions. Te assumptions that linguists make are
diverse and heavily debated. An example is the assumption that
words consist of morphemes , that is, minimal form-mean-
ing pairs. If one accepts this, then it is necessarily true that all
languages have morphemes, and there cannot be exceptions.
Why? Suppose someone claims to have discovered a language
without morphemes. One can of course simply analyze the lan-
guage without mentioning morphemes, but obviously that can-
not challenge the universal just because one can always defend
it by reanalyzing the language with morphemes. Te only true
challenge would be to show that analyzing some data in terms
of morphemes leads to structures that are in confict with other
assumptions, for example, that form-meaning pairs combine
exclusively by linear concatenation. Te confict can be illustrated
by languages with morphologies like the English plural geese ,
where the meanings plural and goose do not correspond to linear
A
ABDUCTION
Abduction is a form of reasoning frst explicated by the nine-
teenth-century philosopher C. S. Peirce. Te central concept
he wishes to introduce is that of generating new hypotheses to
explain observed phenomena partly by guesswork or specu-
lation. In his early work, Peirce tried to explain abductive rea-
soning, as distinct from deductive and inductive reasoning, by
reference to syllogistic form. For instance, the following schema
is an example of deductive reasoning:
All the beans in the bag are white
Tese beans came from this bag
Terefore, these beans are white
Tis is distinct from inductive reasoning which, Peirce argues,
follows this pattern:
Tese beans came from this bag
Tese beans are white
Terefore, all the beans in this bag are white
And both these forms are distinct from abductive reasoning
which, Peirce argues, follows this pattern:
Tese beans are white
All the beans in this bag are white
Terefore, the beans came from this bag
In later work, however, Peirce felt that trying to ft abduc-
tive reasoning into such a strict syllogistic form was restric-
tive, and instead he opted for the following schema to explain
abduction:
Te surprising fact C is observed
But if A were true, C would be a matter of course
Hence, there is a reason to suspect that A is true.
(Peirce 1935 , 189)
For example, suppose I observe that my car will not start.
One good explanation for this would be that it is out of fuel.
Consequently, it seems that we have a good reason to think that
my cars refusal to start is due to its being out of fuel. Of course,
we may very quickly discover that my car has plenty of fuel,
and a diferent hypothesis must be adopted, but Peirce always
intended that abductive reasoning was fallible and conjectural,
awaiting confrmation from other testing.
Peirces account of abduction has been widely adopted in
the philosophy of science, but it has also been of some interest
to linguists. One particularly prominent use of abduction has
been in historical linguistics for explaining language
change (see, for instance, Anderson 1973 ). Te systematic
features of a language that govern the use of one generation
are opaque to the following generation as they acquire that lan-
guage the only access is through language output. It appears,
then, that following generations must use abductive inferences
to access the rules of language before applying those rules to
new cases . And, of course, since abduction is fallible, the rules
Abduction Absolute and Statistical Universals
78
Tis view of absolute universals is highly controversial: Many
linguists limit absolute universals to what is descriptively necessary
in every language; many psychologists propose that children apply
diferent and much more general principles in acquiring a lan-
guage than those found in linguists metalanguages; and to date,
no absolute universal has been confrmed by genetic research.
Statistical Universals
What is not an absolute universal is a variable (or character, or
parameter ): some languages have a certain structure or they
dont have it, or to diferent degrees. It is interesting to note that
most variables show some skewing in their distribution; some val-
ues of a variable are favored only in certain geographical areas (rel-
ative pronouns in Europe) or only in certain families (stem-internal
infection in Afroasiatic). But some values are globally favored (e.g.,
nasals) or, what is more typical, globally favored under certain
structural conditions (e.g., postnominal relative clauses among
languages with objects following the verb). Tese global prefer-
ences are called unconditional (unrestricted) and conditional
(restricted) statistical universals, respectively. (An alternative term
for conditional statistical universals is implicational univer-
sals , but this invites confusion because their probablistic nature
diferentiates them from logical implications; cf. Cysouw 2005 )
Statistical universals are mostly motivated by theories of how
languages develop, how they are used, how they are learned,
and how they are processed. One such theory, for example, pro-
poses that processing preferences in the brain lead to a universal
increase in the odds for postnominal structures among verb-ob-
ject languages (Hawkins 2004 ).
Statistical universals take the same forms as statistical hypoth-
eses in any other science for example, they can be formulated
in terms of regression models. Tey can be tested with the same
range of statistical methods as in any other science, and, again
as in other sciences, the appropriate choice of models, popula-
tion assumptions, and testing methods is an issue of ongoing
research (e.g. Cysouw 2005 ; Janssen, Bickel, and Ziga 2006 ;
Maslova 2008).
A central concern when testing statistical universals is to
ascertain true globality, that is, independence of area and fam-
ily. Areas can be controlled for by standard factorial analysis, but
it is an unsettled question just what the relevant areal relations
are; for example, should one control for the infuence of Europe
or the entire Eurasia or both? A quick solution is to assume a
standard set of fve or six macroareas in the world and accept as
universal if a distribution is independent of these areas (Dryer
1989 ). But the rationale for such a set is problematic, and this
has led to a steep surge of interest in research on areas and their
historical background (e.g., Nichols 1992 ; Haspelmath et al.
2005 ).
Controlling for family relations poses another problem.
Under standard statistical procedures, one would draw ran-
dom samples of equal size within each family and then model
families as levels of a factor. However, over a third of all known
families are isolates, containing only one member each. And
picking one member at random in larger families is impossible
if at the same time one wants to control for areas (e.g., admitting
an Indo-European language from both Europe and South Asia).
strings of morphemes. Confronted with such data, there are
three options:
(i) give up the notion of morpheme;
(ii) give up the assumption of linear concatenation;
(iii) add additional assumptions that reconcile the confict.
On any of these options, the universal remains exception-
less: On solution (i), no language has morphemes; on solu-
tions (ii) and (iii), all languages have morphemes. As a result,
absolute universals can never be falsifed by individual data.
Teir validity can only be evaluated by exploring whether they
are consistent with other absolute universals that are claimed
simultaneously.
Absolute universals can also be thought of as those aspects
of ones descriptive metalanguage often called a theoreti-
cal framework that are necessarily referred to in the analysis
of every language, that is, that constitute the descriptive a priori.
Depending on ones a priori, this includes, apart from the mor-
pheme, such notions as distinctive feature, constituent (see con-
stituent structure ), argument, predicate (see predicate
and argument ), reference, agent, speaker, and so on. In some
metalanguages, the a priori also includes more specifc assump-
tions, for example, that constituents can only be described by
uniform branching (all to the left, or all to the right), or only by
binary branching, and so on .
Te status of absolute universals is controversial. For many
linguists, especially in typology and historical linguis-
tics , absolute universals are simply the descriptive a priori,
with no additional claim on biological or psychological real-
ity. Te choice between equally consistent universals/meta-
languages for example, among options (i), (ii), and (iii) in
the previous example is guided by their success in describing
structures and in defning variables that capture distributional
patterns, an evaluation procedure comparable to the way in
which technical instruments for analyzing objects are evaluated
in the natural sciences. In the morphology problem, typologists
would most likely chose option (ii) because it allows for defning
a variable of stem-internal versus af xal plural realization that
has an interesting distribution (suggesting, for example, that
within-stem realization is favored by a few families in Africa and
the Near East).
In generative grammar , by contrast, absolute universals
are thought of not only as descriptively a priori but also as bio-
logically given in what is called universal grammar : they are
claimed to be innate (see innateness and innatism ) and to
be identical to the generalizations that a child makes when learn-
ing language . Tus, if the morpheme is accepted as a universal,
that is, a priori term of our metalanguage, it will also be claimed to
be part of what makes languages learnable (see learnability )
and to be part of our genetic endowment. An immediate conse-
quence of such an approach is that something can be claimed as
universal even if it is not in fact necessary in the analysis of every
language. For example, even if some language (e.g., the Rotokas
language of Bougainville) lacks evidence for nasal sounds, one
could still include a distinctive feature [ nasal] in Universal
Grammar. Rotokas speakers are then said to have the feature as
part of their genetic endowment even if they dont use it.
Absolute and Statistical Universals
79
Maslova , E. . 2000 . A dynamic approach to the verifcation of distribu-
tional universals . Linguistic Typology 4 : 307 33.
. 2008. Meta-typological distributions. Sprachtypologie und
Universalien forschung 61: 199207.
Newmeyer , F. J . 2005 . Possible and Probable Languages: A Generative
Perspective on Linguistic Typology . New York : Oxford University Press .
Nichols , J. 1992 . Language Diversity in Space and Time . Chicago : Te
University of Chicago Press .
Pinker , S. , and R. Jackendof . 2005 . Te faculty of language: Whats spe-
cial about it? Cognition 95 : 201 36.
Plank , F. , and E. Filimonova . 2000 . Te Universals Archive: A
brief introduction to prospective users . Sprachtypologie und
Universalienforschung 53 : 109 23.
Plank , F. , ed. 2007 . Linguistic Typology 11 .1 (Special issue treating the
state of typology.)
ACCESSIBILITY HIERARCHY
Edward L. Keenan and Bernard Comrie (1972, 1977) introduce
the accessibility hierarchy (AH) as a basis for several cross-
linguistic generalizations regarding the formation of relative
clauses (RCs).
AH: SUBJ > DO > IO > OBL > GEN > OCOMP
Te terms of the AH are main clause subject , direct object , indi-
rect object , object of pre- or postposition , genitive (possessor), and
object of comparison . Keenan and Comrie cross-classifed RCs
along two parameters: 1) the head noun precedes or follows the
restrictive clause (RestCl), and 2) the case of the position relativ-
ized, NP
rel
, is pronominally marked or not. In (1), from German,
the RestCl, underlined, follows the head in (1a) and precedes it
in (1b). In (2a,b), from Hebrew and Russian, NP
rel
is pronomi-
nally marked but not in English.
(1) a. der Mann, der in seinem Bro arbeitet
the man, who in his study is+working
the man who is working in his study
b. der in seinem Bro arbeitende Mann
the in his study working man
the man who is working in his study
(2) a. ha-isha she Dan natan la et ha-sefer
the-woman that Dan gave to+her acc the-book
the woman that Dan gave the book to
b. devuka, kotoruju Petr ljubit
girl, who(acc) Peter loves
the girl who Peter loves
A given choice of values for the two parameters defnes an
RC-forming strategy . A strategy that applies to SUBJ is called pri-
mary . German has two primary strategies, a postnominal, +case
one, (1a), and a prenominal, case one, (1b). Keenan and Comrie
support three hierarchy generalizations:
(3) a. All languages have a primary strategy
b. A given RC-forming strategy must apply to a continuous
segment of the AH
c. A primary strategy may cease to apply at any position on
the AH
For example, many West Austronesian languages, such as
Malagasy (Madagascar), only have primary strategies. So we can
only relativize the agent in (4a).
In response to this problem, typologists seek to ensure represen-
tativity of a sample not by random selection within families but
by exhaustive sampling of known families, stratifed by area. In
order to then control for unequal family sizes, one usually admits
only as many data points per family as there are diferent values
on the variables of interest (Dryer 1989 ; Bickel 2008).
Samples that are not based on random sampling do not
support parametric inference by statistical tests. An alternative
to this is randomization methods (Janssen, Bickel, and Ziga
2006 ): Te null hypothesis in these methods is that an observed
preference can be predicted from the totals of the sample (e.g.,
that an observed 90% postnominal relatives in VO [verb-object]
languages could be predicted if 90% of the entire sample had
postnominal relatives) not that the sample stems from a popu-
lation without the observed preference. Extrapolation to the total
population (the entire of set of human languages) can then only
be based on plausibility arguments: If a preference signifcantly
deviates from what is expected from the totals of the observed
sample, it is likely that the preference holds in all languages. A
key issue in such argumentation is whether the tested variables
are suf ciently unstable over time so that a present sample can
be assumed to not refect accidental population skewings from
early times in prehistory (Maslova 2000 ). In response to this,
typologists now also seek to test universals by sampling language
changes instead of language states a move that is sometimes
called the dynamization of typology (Greenberg 1995 ; Croft
2003 ).
While the number of proposed statistical universals is
impressive the Universals Archive at Konstanz has collected
more than 2,000 (Plank and Filimonova 2000 ) very few of them
have been rigorously tested for independence of area, family,
and time.
Balthasar Bickel
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bickel , B. 2008. A refned sampling procedure for genealogical control.
Sprachtypologie und Universalienforschung 61: 2233.
Croft , W. 2003 . Typology and universals . 2d ed. Cambridge : Cambridge
University Press .
Cysouw , M. 2005 . Quantitative methods in typology. In Quantitative
Linguistics: An International Handbook , ed. G. Altmann , R. Khler, and
R. Piotrowski , 55478. Berlin : Mouton de Gruyter .
Cysouw , M, ed. 2008. Special issue on analyzing Te World Atlas of
Language Structures. Sprachtypologie und Universalienforschung 61.
Dryer , M. S. 1989 . Large linguistic areas and language sampling . Studies
in Language 13 : 257 92.
Greenberg , J. H. 1995 . Te diachronic typological approach to lan-
guage. In Approaches to Language Typology , ed. M. Shibatani and T.
Bynon , 14366. Oxford : Clarendon .
Haspelmath , M. , M. S. Dryer , D. Gil , and B. Comrie , eds. 2005 . Te World
Atlas of Language Structures . Oxford : Oxford University Press .
Hauser , M. D. , N. Chomsky , and W. T. Fitch . 2002 . Te faculty of lan-
guage: What it is, who has it, and how did it evolve? Science 298 : 1569
79. Tis paper and the response by S. Pinker and R. Jackendof 2005
launched an ongoing debate on the nature and extent of absolute uni-
versals in generative grammar.
Hawkins , J. A. 2004 . Ef ciency and Complexity in Grammars .
Oxford : Oxford University Press .
Janssen , D. , B. Bickel , and F. Ziga . 2006 . Randomization tests in lan-
guage typology. Linguistic Typology 10 : 419 40.
Absolute and Statistical Universals Accessibility Hierarchy
80
causativizing a transitive verb, its agent argument may surface as
an IO ( Jai fait manger les pinards aux enfants I made-eat the
spinach to the children). Lastly, S. Hawkins and Keenan ( 1987 )
show psycholinguistically that recall of RCs formed on high posi-
tions on the AH was better than recall of ones formed on low
positions.
One interesting modifcation to the hierarchy generaliza-
tions concerns syntactic ergativity. Keenan and Comrie noted
that Dyirbal (Dixon 1972 ) relativizes absolutives intransitive
subjects and transitive objects, but not transitive subjects. A
verbal af x (antipassive) derives intransitive verbs from transi-
tive ones with the agent as subject, hence relativizable. Mayan
languages such as Jacaltec (Craig 1977 , 196) are similar. Tis is
an elegant solution to the requirement that agents be relativ-
izable, analogous to Bantu applicatives or Austronesian voices
af xes.
(7) a. x s watxe naj hun-ti
asp 3abs 3erg make cl:man one-this
He made this
b. naj x watxe n hun-ti
cl:man asp 3abs make ap one-this
the man (who) made this
Edward L. Keenan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Comrie , Bernard. 1976 . Te syntax of causative constructions: Cross-
language similarities and divergences. In Te Grammar of Causative
Constructions , Syntax and Semantics 6, ed. Masayoshi Shibatani , 261
312. Amsterdam : Academic Press .
Craig , Colette. 1977 . Jacaltec . Austin : University of Texas Press.
Dixon , Robert M. W. 1972 . Te Dyirbal Language of North Queensland .
Oxford : Cambridge University Press.
Hawkins , Sarah , and Edward L. Keenan . 1987 . Te psychological validity
of the accessibility hierarchy. In Universal Grammar : 15 Essays , ed. E.
L. Keenan , 6089. London : Croom Helm .
Keenan , Edward L. 1975 . Variation in Universal Grammar. In Analyzing
Variation in English , ed. R. Fasold and R. Shuy , 13648. Washington,
DC : Georgetown University Press .
Keenan , Edward L. , and Bernard Comrie. 1972 . Noun phrase accessibil-
ity and Universal Grammar. Paper presented at the Annual Meetings
of the Linguistic Society of America, Atlanta.
. 1977. Noun phrase accessibility and Universal Grammar .
Linguistic Inquiry 8.1: 63 99.
ACOUSTIC PHONETICS
Like the rest of linguistics, acoustic phonetics combines descrip-
tion with theory. Descriptions are images of acoustic properties
and quantitative measures taken from these images, and theory
accounts for the way in which a sounds articulation determines
its acoustics.
Description
Te three most commonly used images of speech are the wave-
form , spectrum , and spectrogram ( Figure 1 ). Te waveform dis-
plays diferences in sound pressure level (in pascals) over time
( Figure 1a , d ) , the spectrum diferences in sound pressure level
(4) a. Manolotra (m+aN+tolotra) vary ho anny vahiny aminny
lovia vaovao ny tovovavy
ofers (pres+act+ofer) rice forthe guests onthe dishes new the
young+woman
Te young woman ofers rice to the guests on the new dishes
b. ny tovovavy (izay) manolotra vary ho anny vahiny aminny
lovia vaovao
the woman (that) ofers rice tothe guests onthe dishes new
the young woman who ofers rice to the guests on the new dishes
c. *ny vary (izay) manolotra ho anny vahiny aminny lovia
vaovao ny tovovavy
the rice (that) ofers tothe guests onthe dishes new the
young+woman
the rice that the young woman ofers to the guests on the new
dishes
Te frst four words in (4c) claim that the rice is doing the
ofering a nonsense. Malagasy does not, however, have an
expressivity gap here since it has a rich voice system allow-
ing any major NP in a clause as subject. Te form of ofer that
takes theme subjects is atolotra , recipient subjects tolorana ,
and oblique subjects anolorana . (5a,b) illustrate Teme and
Instrument RCs.
(5) a. ny vary (izay) atolo-dRasoa ho anny vahiny aminny lovia
vaovao
the rice (that) ofered-by+Rasoa forthe guests onthe new
dishes
the rice that the young+woman ofers to the guests on the
new dishes
b. ny lovia vaovao (izay) anoloran-dRasoa vary ho anny
vahiny
the dishes new (that) ofered-by+Rasoa rice for the guests
the new dishes on which the young woman ofered rice to the
guests
Bantu languages, such as Luganda, (6), illustrate the DO cutof.
Only subjects and objects are directly relativizable. Obliques can
be promoted to object using applicative af xed verbs. So the
instrumental in (6a) is only relativizable from (6c).
(6) a. John yatta enkoko n (= na) ekiso
John killed chicken with knife
b. *ekiso John kye-yatta enkoko (na) knife
John rel-killed chicken (with)
c. John yattisa (yatt+is+a) ekiso enkoko
John kill+with knife chicken
John killed+with a knife the chicken
d. ekiso John kye-yattisa enkoko knife
John rel-kill+with chicken
the knife John killed the chicken with
Independent support for the AH: Keenan ( 1975 ) shows that
stylistically simple texts used RCs formed high on the AH pro-
portionately more than texts independently judged stylistically
hard. Second, Comries work (1976) supports the conclusion that
the positioning of demoted subjects in morphological causatives
tends to assume the highest function on the AH not already flled.
Tus in the French Jai fait rire les enfants I made-laugh the chil-
dren, the children surfaces as a DO as laugh lacks a DO. But in
Accessibility Hierarchy Acoustic Phonetics
81
Note that the numerator in (2) is not 1 but instead c , the speed
of sound.
Te spectrum and spectrogram of [i] ( Figures 1b , c ) show peaks
and horizontal bands, respectively, known as formants , at roughly
300, 2,200, and 2,800 Hz. Te corresponding images of [s] ( Figures
1e , f ) show a broad energy band spanning 4,0007,000 Hz.
Whether a sound is periodic and where in its spectrum energy
is concentrated are nearly suf cient to distinguish all speech
sounds from one another acoustically, and these two proper-
ties also refect the two components of the theoretical model for
transforming articulations into acoustics. All voiced sounds are
periodic, as are trills. Sonorants (vowels, glides, liquids, nasals)
are usually periodic, while obstruents (fricatives, stops, afri-
cates) are aperiodic. Voiced obstruents are both periodic and
aperiodic. Diferences in vowel quality and consonantal place of
articulation are both realized acoustically as diferences in where
energy is concentrated in their spectra. Te remaining property
is duration, which besides conveying short:long contrasts also
contributes to the conveying of tense:lax contrasts between vow-
els and the voicing and manner contrasts between consonants .
Theory
Speech sounds are the product of the application of a flter that
determines the frequencies in which energy is concentrated
(in deciBels) over frequency ( Figure 1b , e ) , and the spectrogram
diferences in frequency over time ( Figure 1c , f ,); darkness indi-
cates the sound pressure level at particular frequencies and
moments in the spectrogram .
Te images in Figures 1a c difer from those in Figure 1d f
in every conceivable respect: Sound pressure level varies more
or less regularly and repetitively, every 0.0073 second, in the
vowel [i] (as in heed ), while in the fricative [s] (as in see ), it
instead varies nearly randomly. Te vowel is thus nearly peri-
odic , while the fricative is decidedly aperiodic . Tis diference
gives the vowel a clear pitch, while making the fricative instead
sound noisy.
A single cycle s duration in a periodic sound is its period ( T );
the distance it travels in space is its wavelength ( ). As measures
of a single cycles extent, both period and wavelength are recip-
rocally related to frequency ( F ), the number of cycles per second,
or Hz:
(1)
1
T (sec/cycle)
F (cycles/sec) =
(2)
c (cm/sec)
F (cycles/sec)
(cm/cycle)
=
Figure 1. Waveforms, spectra, and
spectrograms of 30 ms intervals of
the vowel [i] (ac), and the fricative
[s] (df).
Acoustic Phonetics
82
released or continuously through a fricatives narrow con-
striction. In strident sounds, the jet breaks up against a baffle
just downstream, increasing turbulence and noise intensity
considerably .
RESONANCE. Both periodic and aperiodic sound sources intro-
duce acoustic energy into the oral cavity across a broad enough
range of frequencies to excite any resonance of the air inside
the oral cavity. If the articulators are in their rest positions
and vocal folds are in the voicing position, this cavitys shape
approximates a uniform tube, closed at the glottis and open at
the lips ( Figure 2 ). A resonance is produced by the propaga-
tion of acoustic energy away from the source and its refection
back and forth of the two ends of the tube, which establishes a
standing wave . In a standing wave resonance, the locations of
zero and maximum pressure variation are fxed. To understand
how air resonates, it is easier to consider the change in pressure
level in the standing wave, rather than pressure level itself, that
is, the extent to which the air molecules are being displaced
longitudinally, or equivalently the velocity of the pressure
change, rather than the extent of their instantaneous compres-
sion or rarefaction. Air is most freely displaced longitudinally
at the open end, the lips, and least freely at the closed end, the
glottis. As a result, the standing waves that ft best inside the
oral cavity are those whose wavelengths, and thus frequencies,
are such that they have a velocity maximum (antinode) at the
lips, and a velocity minimum (node) at the glottis. Because the
air resonates more robustly at these frequencies than at oth-
ers, the oral cavitys resonant response flters the sound source,
passing energy in the source at some frequencies and stopping
it at others .
Figures 2a c show the three lowest-frequency standing
waves that fit these boundary conditions. How are their fre-
quencies determined? Figures 2a c show that one-quarter
of the first resonances wavelength spans the distance from
the glottis to the lips (the oral cavitys length, L
oc
), three-
quarters of the seconds, and five-quarters of the thirds. More
generally:
(3)
n oc
n
L
4
1 2
=
where n is the resonance number. Solving for wavelength and
substituting into (2) yields:
(4)
oc
n
L
n
c
F
1 2
4
=
Substituting 35,000 cm/sec for c and 17.5 cm for L
oc
(the aver-
age adult males oral cavity length) yields 500, 1,500, and 2,500
Hz as the frst three resonances frequencies, values close to
schwas.
Because the variable L
oc
is in the denominator, resonance
frequencies are lower in adults and mens longer oral cav-
ities than in childrens or womens shorter ones, and like-
wise when the lips are protruded in a rounded vowel, such
in the sounds spectrum to a periodic and/or aperiodic sound
source .
SOUND SOURCES. Sound sources are produced by using valves
that control air fow through the vocal tract to transform the
energy in that fow into sound. In periodic sound sources, the
fow of air causes a valve to open and close rapidly and regularly,
which in turn causes air pressure to rise and fall just downstream.
Te repeated opening and closing of the glottis, known as vocal
fold vibration , is the most common periodic sound source; oth-
ers are uvular, alveolar, and bilabial trills .
Aperiodic sound sources are produced by keeping a valve
completely or nearly closed, in stops and fricatives, respec-
tively. Either way, oral air flow is obstructed enough that
oral air pressure rises behind the obstruction. This pressure
rise speeds up flow enough to turn it into a turbulent and
thus noisy jet, in either a brief burst when a stop closure is
First resonance
0 2.5 5 7.5 10 12.5 15 17.5
Distance from glottis (cm)
Glottis = closed Lips = open
Minimum Maximum
(a)
Second resonance
0 2.5 5 7.5 10 12.5 15 17.5
Distance from glottis (cm)
(b)
(c)
Third resonance
0 2.5 5 7.5 10 12.5 15 17.5
Distance from glottis (cm)
Figure 2. The oral cavity as a tube closed at one end and open at the
other: (ac) standing waves corresponding to the frst three resonances,
each with a velocity minimum at the closed end and a velocity maximum
at the open end.
Acoustic Phonetics
83
is higher in front unrounded [i] than schwa but lower in low
back unrounded [a] and especially high back rounded [u]. F1
varies inversely with tongue height, and F2 varies directly with
tongue backness and lip rounding. The formant frequencies
of all other vowel qualities lie between the extremes observed
in these vowels, just as all other vowels lingual and labial
articulations lie between those of these vowels. F1 starts low
and rises following [b, d, g] ( Figure 4 ), both F2 and F3 start low
following [b] ( Figure 4a ), both formants start higher following
[d] (Figure 4b), and they diverge from very similar frequencies
following [g] (Figure 4c). Although consonants are articulated
at other places of articulation, these three places are distin-
guished in nearly all languages, and many distinguish only
these three.
Te frst heuristic treats the constriction as dividing the oral
cavity into separate resonating cavities ( Figure 5 ), and it applies
the length rule independently to each of them. Te frst three for-
mants are the three lowest of the six resonances produced by the
two cavities. Tis heuristic may be called the cavity association
heuristic because each formant can be associated with the cavity
from which it came. Tere are two complications. First, the cav-
ity behind the constriction is efectively closed at both ends, and
as the [u] in whod , rather than spread in an unrounded one,
such as the [i] in heed . These observations yield the length
rule : Resonance frequencies vary inversely with resonating
cavity length.
Te frst three formants of [i] ( Figures 1b , c ) difer decid-
edly in frequency from those of schwa, because raising the
tongue body toward the front of the palate decreases the oral
cavitys cross-sectional area there and increases it in the phar-
ynx, while spreading the lips, which shortens the oral cavity.
Although the length rule predicts how shortening changes for-
mant frequencies, an additional rule is needed to predict how
decreasing and increasing cross-sectional area afects formant
frequencies.
The predictions of two heuristics that are widely used for
this purpose are tested here against the observed formant
frequency differences between the three vowels [i, u, a], and
between the three places of articulation of the stops in [b,
d, g]. F1 is lower in the high vowels [i, u] ( Figures 3a , b )
than in schwa, but higher in the low vowel [a] ( Figure 3c ). F2
Figure 3. Spectra of the vowels (a) [i] as in heed , (b) [u] as in whod ,
and (c) [a] as in hod . The individual peaks are the harmonics of the fun-
damental frequency (F0) of voice sound source, and the formants are the
ranges of amplifed harmonics. Peaks corresponding to F1F3 are labeled
at the top of each panel.
Figure 4. Spectrograms of the frst 150 ms of the words (a) bad , (b)
dad , and (c) gad . The onsets of F1F3 are labeled.
Acoustic Phonetics
84
efects of lip rounding, which closes the front cavity at both ends
and introduces another Helmholtz resonance. None of the reso-
nances produced by this front cavity are lower than the back cav-
ity resonances, but the additional Helmholtz resonance is low
enough to constitute the F2 observed in [u] (657 Hz if the labial
constriction has a cross-sectional area of 1 cm
2
and a length of 2
cm, and the front cavity is 4.5 cm long).
In the second, perturbation , heuristic, a constrictions prox-
imity to a resonances velocity minimum or maximum deter-
mines how it perturbs that resonances frequency away from its
schwa-like value: A constriction near a velocity minimum raises
the formants frequency, while one near a maximum lowers it
instead (expansions have the opposite efects). Figures 4a c show
that minima occur at even quarters of a resonances wavelength
and maxima at odd quarters, and that their locations are at fxed
proportions of the length of the oral cavity. Because constric-
tion locations are also a certain proportion of the distance from
the glottis to the lips, whether they coincide with a minimum or
maximum can be calculated by multiplying both sides of (4) by
the proportion of the oral cavitys length that corresponds to the
constrictions location and rounding the result on the right-hand
side to the nearest quarter ( Table I ).
Te perturbation heuristic successfully predicts the efects of
the bilabial, palatal, velar, and pharyngeal constrictions on all
three formants of [b, i, g, a], and likewise the efects of the alveo-
lar and velar constrictions on F1 and F3 in [d, u], but it fails to
predict F2 raising after [d], and F2 lowering in [u]. Te latter can
so its resonances must have velocity minima at both ends. Teir
frequencies are predicted by:
(5)
rc
n
L
n
c
F
2
=
where L
rc
is the length of the resonating cavity. Te second com-
plication is that the acoustic interaction of the constriction with
the cavity behind it produces a Helmholtz resonance . Its fre-
quency ( F
h
) is:
(6)
c b b
c
h
L L A
A c
F
2
=
A
c
is the constrictions cross-sectional area, L
c
is its length, and
A
b
and L
b
are the cross-sectional area and length of the cavity
behind the constriction. If a 3 cmlong constriction with a cross-
sectional area of 1 cm
2
is moved incrementally from 3 cm above
the glottis to 0.5 cm back of the lips along a 17.5 cm oral cavity,
the back and front cavities produce the resonance frequencies
displayed in Figure 6 , along with the Helmholtz resonance. Te
arrows projected down from the intersections between back and
front cavity resonances show where F2 and F3 change associa-
tion from the front to the back cavity.
Te constriction centers in [a, u, i] are roughly one-quarter
(4.5 cm), two-thirds (11.5 cm), and three-quarters (13 cm) of the
distance from the glottis to the lips. Te constriction of [g] is close
to [u]s, while [d]s is about seven-eighths the distance from the
glottis (15.5 cm), and [b]s is of course at the lips (17.5 cm).
Te Helmholtz resonance is lowest for all constriction loca-
tions and thus constitutes F1. It also lowers progressively as the
constriction is moved forward because the cavity behind the
constriction lengthens. Te cavity-association heuristics suc-
cessful predictions include the following : 1) Te low or pharyn-
geal vowel [a] has a higher F1 than the high or velar and palatal
vowels [u, i]; 2) F1 is low following [g, d, b]; 3) [a]s F2 (the front
cavitys frst resonance) is low, 4) F2 and F3 (the front and back
cavities frst resonances) start at very similar frequencies follow-
ing [g], because a velar constriction is close to where the front
and back cavities frst resonances cross at 11 cm from the glottis;
5) F2 and F3 start low following [b] (the back cavitys frst and
second resonances). It incorrectly predicts: 6) Te F2 of [i] (the
back cavitys frst resonance) is low, indeed lower than [u]s; and
7) F2 and F3 (the back cavitys frst and second resonances) are
low following [d]. For [u], the calculations leave out the acoustic
back front
constriction
1 cm
2
3 cm
8.5 cm 6 cm
8 cm
2
Figure 5. The confguration of the oral cavity with a constriction partway along its length.
Figure 6. The frst three resonance frequencies of the back cavity (flled
symbols) and front cavity (empty symbols) and the Helmholtz resonance
(crosses) produced by incremental movement of the constriction in
Figure 5.
Acoustic Phonetics
85
If a characteristic lacks any of these features, it is not an adap-
tation. An adaptation is not, therefore, simply anything in an
individual with a good or functional outcome, or that has
useful efects by intuitive standards. Rice cultivation, useful as it
is, is not a biological adaptation because it lacks a specifc genetic
basis. Similarly, the English language is not an adaptation,
however useful it might be. In contrast, if a mutation occurred
that modifed a neural structure so that the vocal chords could
more reliably produce distinct phonemes, and this gene spread
throughout the species because its bearers prospered due to the
advantages resulting from a lifetime of more ef cient communi-
cation, then the modifed neural structure would qualify as an
adaptation.
Researchers judge whether something is an adaptation by
assessing how likely or unlikely it is that its functional organiza-
tion was produced by random mutation and spread by genetic
drift. For example, the eye has hundreds of elements that are
arranged with great precision to produce useful visual inputs. It is
astronomically unlikely that they would have arrived at such high
levels of mutual coordination and organization for that function
unless the process of natural selection had diferentially retained
them and spread them throughout the species. Consequently,
the eye and the visual system are widely considered to be obvi-
ous examples of adaptations. For the same reason, evolutionary
scientists consider it overwhelmingly likely that many neurocog-
nitive mechanisms underlying language are adaptations for
communication (a proposition that Noam Chomsky has dis-
puted; see Lyle Jenkinss essay, Explaining Language, in this
volume). Language competence reliably develops, is believed
to have a species-typical genetic basis, and exhibits immensely
complex internal coordination that is functionally organized to
produce ef cient communication, which vastly enhances the
achievement of instrumental goals, plausibly including those
linked to ftness .
again be predicted once the acoustic efects of lip rounding are
added, as the simultaneous labial constriction, together with the
protrusion of the lips, lowers F2 along with all other formants.
Summary
Speech sounds articulations produce sound sources by trans-
forming aerodynamic energy into acoustic form, and those
sound sources in turn cause air inside the oral cavity to resonate,
at frequencies determined by the length of the resonating cavi-
ties and where they are constricted.
John Kingston
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Fant , C. , and M. Gunnar . 1960 . Acoustic Teory of Speech Production . Te
Hague : Mouton .
Jakobson , Roman , C. Fant , M. Gunnar , and Morris Halle . 1952 .
Preliminaries to Speech Analysis , Cambridge, MA : MIT Press .
Ladefoged , Peter , and Ian Maddieson. 1996 . Sounds of the Worlds
Languages . Oxford: Blackwell Publishers .
Stevens , Kenneth N. 1998 . Acoustic Phonetics . Cambridge, MA : MIT Press.
ADAPTATION
An adaptation is a characteristic in an organism that evolved
because it helped the organism or its relatives to survive and
reproduce. Examples include the vertebrate eye, claws, mam-
mary glands, the immune system, and the brain structures that
underlie the human capacity for language. More completely,
an adaptation is 1) a reliably developing set of characteristics
2) whose genetic basis became established and organized in the
species (or population) over evolutionary time because 3) the
adaptation interacted with recurring features of the body or envi-
ronment 4) in a way that, across generations, typically caused
this genetic basis to increase its gene frequency.
Table 1. Calculating a constrictions proximity to a resonances velocity minimum or maximum from the constrictions proportional distance from
the glottis to the lips.
Place of
constriction
Segment Proportion of oral
cavity length
Proximity to minimum or maximum
Calculation Odd/even Lower/higher
Labial b 1 1* 1 /4=1/4
1*3 2 /4=3/4
1*5 3 /4=1/4
Odd
Odd
Odd
F1 lower
F2 lower
F3 lower
Alveolar d 7/8 7/8* 1 /4=7/321/4
7/8*3 2 /4=21/323/4
7/8*5 3 /4=35/324/4
Odd
Odd
Even
F1 lower
F2 lower
F3 higher
Palatal i 3/4 3/4* 1 /4=3/161/4
3/4*3 2 /4=9/162/4
3/4*5 3 /4=15/164/4
Odd
Even
Even
F1 lower
F2 higher
F3 higher
Velar g, u 2/3 2/3* 1 /4=2/121/4
2/3*3 2 /4=6/122/4
2/3*5 3 /4=10/123/4
Odd
Even
Odd
F1 lower
F2 higher
F3 lower
Pharyngeal a 1/4 1/4* 1 /4=1/160/4
1/4*3 2 /4=3/161/4
1/4*5 3 /4=5/161/4
Even
Odd
Odd
F1 higher
F2 lower
F3 lower
Adaptation
86
relevant goal by organizing the current situation in a way that
supports efective goal pursuit.
Ad hoc categories contrast with thousands of well-established
categories associated with familiar words (e.g., cat, eat, happy ).
Extensive knowledge about these latter categories resides in
memory and may often become active even when irrelevant to
current goals. When ad hoc categories are used frequently, how-
ever, they, too, become highly familiar and well established in
memory. Te frst time that someone packs a suitcase, the cate-
gory things to pack in a suitcase is ad hoc. Following many trips,
however, it becomes entrenched in memory.
Ad hoc categories constitute a subset of role categories,
where roles provide arguments for verbs, relations, and sche-
mata . Some role categories are so familiar that they become
lexicalized (e.g., seller, buyer, merchandise, and payment name
the agent, recipient, theme, and instrument roles of buy ). When
the conceptualization of a role is novel, however, an ad hoc cate-
gory results (e.g., potential sellers of gypsy jazz guitars ). Pursuing
goals requires the constant specifcation and instantiation of
roles necessary for achieving them. When a well-established cat-
egory for a role doesnt exist, an ad hoc category is constructed
to represent it.
Both conceptual and linguistic mechanisms appear central to
forming ad hoc categories. Conceptually, people combine exist-
ing concepts for objects, events, settings, mental states, proper-
ties, and so on to form novel conceptual structures. Linguistically,
people combine words in novel ways to index these concepts.
Sometimes, novel concepts result from perceiving something
novel and then describing it (e.g., seeing a traditional opera set in
a modern context and describing this newly encountered genre
as modernized operas). On other occasions, people combine
words for conceptual elements before ever encountering an
actual category instance (e.g., describing mezzo sopranos who
have power, tone, and fexibility before experiencing one). Te
conceptual and linguistic mechanisms that formulate ad hoc
categories are highly productive, given that components of these
categories can be replaced systematically with alternative values
from semantic felds (e.g., tourist activities to perform in X, where
X could be Rome, Florence, Venice, etc.). Syntactic structures
are also central for integrating the conceptual/linguistic com-
ponents in these categories (e.g., the syntax and accompanying
closed class words in tourist activities to perform in Rome ).
Lawrence Barsalou ( 1983 ) introduced the construct of ad hoc
categories in experiments showing that these categories are not
well established in memory and do not become apparent without
context. Once constructed, however, they function as coherent
categories, exhibiting internal structures as indexed by typicality
gradients. Barsalou ( 1985 ) showed that these gradients are orga-
nized around ideal values that support goal achievement and
also around frequency of instantiation. He also showed (1987)
that these internal structures are generally as stable and robust
as those in familiar taxonomic categories.
Barsalou ( 1991 ) ofered a theoretical framework for ad hoc
categories (see also Barsalou 2003 ). Within this framework, ad
hoc categories provide an interface between roles in knowl-
edge structures (e.g., schemata) and the environment. When a
role must be instantiated in order to pursue a goal but knowl-
edge of possible instantiations does not exist, people construct
Within the evolutionary sciences, the concept of adaptation
plays an indispensable role not only in explaining and under-
standing how the properties of organisms came to be what they
are, but also in predicting and discovering previously unknown
characteristics in the brains and bodies of species .Evolutionary
psychologists , for example, analyze the adaptive problems our
ancestors were subjected to, predict the properties of previ-
ously unknown cognitive mechanisms that are expected to have
evolved to solve these adaptive problems, and then conduct
experimental studies to test for the existence of psychological
adaptations with the predicted design (see evolutionary
psychology ). An understanding that organisms embody sets
of adaptations rather than just being accidental agglomerations
of random properties allows organisms to be properly studied as
functional systems. If language is accepted as being the product
of adaptations, then there is a scientifc justifcation for studying
the underlying components as part of a functional system.
Te concept of adaptation became more contentious when
human behavior and the human psychological architecture
began to be studied from an adaptationist perspective. Critics
have argued that not every characteristic is an adaptation an
error adaptationists also criticize. More substantively, critics
have argued that it is impossible to know what the past was like
well enough to recognize whether something is an adaptation.
Adaptationists counter that we know many thousands of things
about the past with precision and certainty, such as the three-
dimensional nature of space and the properties of chemicals, the
existence of predators, genetic relatives, eyes, infants, food and
fertile matings, and the acoustical properties of the atmosphere,
and that these can be used to gain an engineers insight into why
organisms (including humans) are designed as they are.
Julian Lim , John Tooby, and Leda Cosmides
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Gould , S. J. , and R. C. Lewontin . 1979 . Te spandrels of San Marco and
the Panglossian paradigm: A critique of the adaptationist programme.
Proceedings of Te Royal Society of London, Series B 205 . 1161 : 581 98.
Pinker , Steven. 1994 . Te Language Instinct . New York : Morrow .
. 2003 . Language as an adaptation to the cognitive niche. In
Language Evolution , ed. M. Christiansen and S. Kirby , 1637. New
York : Oxford University Press.
Tooby , John , and I. DeVore . 1987 . Te reconstruction of hominid
behavioral evolution through strategic modeling. In Te Evolution
of Primate Behavior: Primate Models . ed. W. G. Kinsey , 183237. New
York : SUNY Press .
Williams , George C. 1966 . Adaptation and Natural Selection: A Critique
of Some Current Evolutionary Tought . Princeton, NJ : Princeton
University Press .
AD HOC CATEGORIES
An ad hoc category is a novel category constructed spontane-
ously to achieve a goal relevant in the current situation (e.g., con-
structing tourist activities to perform in Beijing while planning a
vacation). Tese categories are novel because they typically have
not been entertained previously. Tey are constructed sponta-
neously because they do not reside as knowledge structures in
long-term memory waiting to be retrieved. Tey help achieve a
Ad hoc Categories
87
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Barsalou , L. W. 1983 . Ad hoc categories . Memory & Cognition
11 : 211 27.
. 1985 . Ideals, central tendency, and frequency of instan-
tiation as determinants of graded structure in categories . Journal
of Experimental Psychology: Learning, Memory, and Cognition
11 : 629 54.
. 1987 . Te instability of graded structure: Implications for the nature
of concepts. In Concepts and Conceptual Development: Ecological
and Intellectual Factors in Categorization , ed. U. Neiser , 10140.
Cambridge : Cambridge University Press .
. 1991 . Deriving categories to achieve goals. In Te Psychology of
Learning and Motivation: Advances in Research and Teory . Vol. 27.
Ed. G. Bower , 164. San Diego, CA : Academic Press .
. 1999 . Perceptual symbol systems . Behavioral and Brain Sciences
22 : 577 660.
. 2003 . S ituated simulation in the human conceptual system .
Language and Cognitive Processes 18 : 513 62.
Cech , C. J. , E. J. Shoben , and M. Love . 1990 . Multiple congru-
ity efects in judgments of magnitude . Journal of Experimental
Psychology: Learning, Memory, and Cognition 16 : 1142 52.
Chrysikou , E. G. 2006 . When shoes become hammers: Goal-derived cat-
egorization training enhances problem-solving performance. Journal
of Experimental Psychology: Human Learning and Performance
32 : 935 42.
Glucksberg , S. , and B. Keysar . 1990 . Understanding metaphorical com-
parisons: Beyond similarity. Psychological Review 97 : 3 18.
Lucariello , J. , and K. Nelson . 1985 . S lot-fller categories as mem-
ory organizers for young children. Developmental Psychology
21 : 272 82.
Medin , D. L. , N. Ross , S. Atran , D. Cox , J. Coley , J. Prof tt , and S. Blok .
2006 . Folkbiology of freshwater fsh. Cognition 99 : 237 73.
Ross , B. H. , and G. L. Murphy . 1999 . Food for thought: Cross-classifcation
and category organization in a complex real-world domain. Cognitive
Psychology 38 : 495 553.
Valle-Tourangeau , F. , S. H. Anthony , and N. G. Austin . 1998 . S trategies
for generating multiple instances of common and ad hoc categories.
Memory 6 : 555 92.
ADJACENCY PAIR
conversation analysis , an inductive approach to the micro-
analysis of conversational data pioneered by Harvey Sacks
( 1992 ), attempts to describe the sequential organization of
pieces of talk by examining the mechanics of the turn-taking sys-
tem. Adjacency pairs refect one of the basic rules for turn-taking
(Sacks, Scheglof, and Jeferson 1974 ), in which a speaker allo-
cates the conversational foor to another participant by uttering
the frst part of a paired sequence, prompting the latter to pro-
vide the second part. Examples are question-answer, greeting-
greeting as in (1), and complaint-excuse:
(1) A: Hi there
B: Oh hi
Te constitutive turns in adjacency pairs have the following
structural characteristics:
(i) Tey are produced by two diferent speakers.
(ii) Tey are, as the term suggests, adjacent. Tis is not a strict
requirement, as the two parts can be separated by a so-called
insertion sequence , as in (2):
an ad hoc category of possible instantiations (e.g., when going
camping for the frst time, constructing and instantiating
activities to perform on a camping trip ). Te particular instan-
tiations selected refect their ft with a) ideals that optimize
goal achievement and b) constraints from the instantiations
of other roles in the knowledge structure (e.g., activities to
perform on a camping trip should, ideally, be enjoyable and
safe and should depend on constraints such as the vacation
location and time of year). Once established, the instantiations
of an ad hoc category are encoded into memory and become
increasingly well established through frequent use (e.g., estab-
lishing touring back roads and socializing around the camp-
ground as instances of activities to perform on a camping trip ).
Barsalou ( 1999 ) describes how this framework can be realized
within a perceptual symbol system. Specifcally, categories
(including ad hoc categories) are sets of simulated instances
that can instantiate the same space-time region of a larger
mental simulation (where a simulation is the reenactment of
modality-specifc states, as in mental imagery).
Ad hoc categories have been studied in a variety of empiri-
cal contexts. S. Glucksberg and B. Keysar ( 1990 ) proposed that
ad hoc categories underlie metaphor (e.g., the metaphor jobs are
jails conceptualizes the category of confning jobs ). C. J. Cech, E.
J. Shoben, and M. Love ( 1990 ) found that ad hoc categories are
constructed spontaneously during the magnitude comparison
task (e.g., forming the ad hoc category of small furniture , such
that its largest instances anchor the upper end of the size dimen-
sion). F. Vallee-Torangeau, S. H. Anthony, and N. G. Austin
( 1998 ) found that people situate taxonomic categories in back-
ground settings to form ad hoc categories (e.g., situating fruit
to produce fruit in the produce section of a grocery store ). E. G.
Chrysikou ( 2006 ) found that people rapidly organize objects into
ad hoc categories that support problem solving (e.g., objects that
serve as platforms ).
Research has also addressed ad hoc categories that become
well established in memory, what Barsalou ( 1985 , 1991 ) termed
goal-derived categories (also called script categories , slot fller
categories , and thematic categories ). J. Luciarello and K. Nelson
( 1985 ) found that children acquire goal-derived categories associ-
ated with scripts before they acquire taxonomic categories (e.g.,
places to eat ). B. H. Ross and G. L. Murphy ( 1999 ) examined how
taxonomic and goal-derived concepts simultaneously organize
foods (e.g., apples as belonging simultaneously to fruit and snack
foods ). D. L. Medin and colleagues (2006) found that goal-derived
categories play central roles in cultural expertise (e.g., tree experts
form categories relevant to their work, such as junk trees ).
Although ad hoc and goal-derived categories are ubiquitous
in everyday cognition, they have been the subject of relatively
little research. Much further study is needed to understand
their structure and role in cognition. Important issues include
the following: How do productive conceptual and linguistic
mechanisms produce ad hoc categories? How do these catego-
ries support goal pursuit during situated action? How do these
categories become established in memory through frequent use?
How does the acquisition of these categories contribute to exper-
tise in a domain?
Lawrence W. Barsalou
Ad hoc Categories Adjacency Pair
88
Static versus Dynamic Theories
Mundane observations make it abundantly clear that language
change is not punctual but gradual, and not categorical but
variable. Traditional views of language change imposed meth-
odological restrictions to avoid viewing change while it was pro-
gressing. Linguists had little confdence in their ability to discern
change precisely and accurately amid the apparent lawlessness
of social phenomena, as Edward Sapir incisively put it ( 1929 ,
213). However, in the 1960s, linguists began studying language in
its social context. By that time, economics, anthropology, sociol-
ogy, and other social sciences were well established, and linguis-
tic studies belatedly admitted sociolinguistics, its social science
adjunct (Chambers 2002b ).
Viewing language changes as they progressed entailed the
admission of coexisting linguistic entities as data. Linguists
were required to study the social distribution of, for example,
both sneaked and snuck as variants of the past tense of sneak
(as in the example to be discussed). Dealing coherently with
variables necessitated determining the distribution of vari-
ants with certain social factors, including the age of the speak-
ers. Uriel Weinreich, William Labov, and Marvin I. Herzog, in
the document that became the manifesto for viewing language
change in its social context, said: A model of language which
accommodates the facts of variable usage and its social and
stylistic determinants not only leads to more adequate descrip-
tions of linguistic competence, but also naturally yields a theory
of language change that bypasses the fruitless paradoxes with
which historical linguistics has been struggling for half a cen-
tury (1968, 99). Primary among those paradoxes, of course,
were concepts of change as punctual, categorical, and static. By
admitting social variables, it became possible to view change as
gradual, variable, and dynamic, consistent with commonsense
observations.
Real-Time Change
Te study of the progress of change can be carried out in real time
by revisiting survey sites at intervals and observing changes in the
social distribution of variants from one time to the next. Because
changes will not necessarily be completed in the interval but will
be continuing, their progress must be calculated by some kind
of proportional measure, such as the percentage of the variants,
their relative frequency, or their probabilities. Te proportional
diferences from one visit to the next provide a quantitative mea-
sure of the progress of the change.
Studying change in real time has a number of methodologi-
cal disadvantages. Most obvious is the time required before
attaining a result. Locating subjects on subsequent visits also
poses obvious problems because of mobility, cooperation, or
death. Furthermore, subsequent visits require the addition of
new subjects at the youngest age group each time. Tis is nec-
essary because of the manner in which linguistic innovations
typically difuse (see diffusion ) throughout the population.
Instead of spreading outward from the source and afecting
almost everybody in their sphere of infuence, as infectious
diseases do in epidemics and technological adoptions do
when, say, spin dryers replace washboards, linguistic inno-
vations tend to be stratifed . Under ordinary circumstances,
people acquire their accents and dialects in their formative
(2) A: Whats the time now? ( Question 1 )
B: Dont you have a watch? (Question 2)
A: No. (Answer 2)
B: I think its around three. (Answer 1)
(iii) Tey are organized as a frst and a second part, that is,
they are nonreversible. Tis is the case, incidentally, even in
ostensibly identical frst and second parts, such as the greet-
ing-greeting pair in (1), where reversing the order results in
an aberrant sequence.
(iv) Tey are ordered, so that a particular frst part requires a
relevant second part (e.g., greetings do not follow questions).
Te fact that the second part is conditionally relevant on the frst
part does not mean that only one option is available; in fact, cer-
tain frst parts typically allow for a range of possible second parts.
If two (or more) options are possible, one will be the more socially
acceptable, preferred response, the other(s) being dispreferred;
this phenomenon is known as preference organization , as in:
(3) A: Have a piece of cake ( frst part )
B1: Great thanks I will (preferred second)
B2: Ehm actually Ive just eaten but thanks anyway (dispre-
ferred second)
As illustrated in (3), dispreferred second parts tend to be structur-
ally diferent from preferred seconds (B2 being indirect, includ-
ing an explanatory account, and containing hesitation markers,
unlike B1).
Another, related phenomenon that merits mention here is
presequencing : Certain adjacency pairs can be introduced or
foreshadowed by a preceding exchange, as in:
(4) A1: Do you sell fresh semiskimmed milk?
B1: We sure do
A2: Ill have two bottles then please
B2: OK
Tis whole exchange forms one unit, in the sense that the occur-
rence of the question-answer pair A1-B1 is only interpretable
given the subsequent request-compliance adjacency pair A2-B2.
Phenomena such as this are indicative of a level of sequential
organization in conversation beyond two-turn sequencing (see
Scheglof 2007 ).
Ronald Geluykens
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Sacks , Harvey. 1992 . Lectures on Conversation . Oxford : Blackwell .
Sacks , Harvey , Emanuel A. Scheglof , and Gail Jeferson . 1974 . A sim-
plest systematics for the organization of turn-taking in conversation.
Language 50 : 696 735.
Scheglof , Emanuel A. 2007 . Te Language of Turn and Sequence .
Cambridge : Cambridge University Press .
AGE GROUPS
Age is one of the primary independent variables in
sociolinguistics , along with social class, sex, ethnicity, and
region. Age is the primary social correlate for determining that
language is changing and for estimating the rate of change and
its progress.
Age Groups
89
currency. It gained currency steadily thereafter, however, and
accelerated most rapidly in the speech of the 50-year-olds,
people born in 193342, increasing by some 25 percent and
becoming the variant used by almost 70 percent of them. In
subsequent decades, it was adopted by ever-greater numbers.
In the 1980s, the formative years for people born in 19738, the
teenagers in this survey, snuck virtually eliminated sneaked as
a variant.
Changes normally take place beneath the level of conscious-
ness. Young people seldom have a sense of the history of the
variants. In this case, people under the age of 30 often con-
sider the obsolescent form sneaked to be a mistake when they
hear someone say it. Tere is no communal sense that sneaked
was the historical standard and accepted form for centuries.
Occasionally, changes become self-conscious in the early stages
if teachers, writers, or parents openly criticize them. Such criti-
cisms almost never succeed in reversing trends, though they
may slow their momentum. When the incoming variant gains
enough currency, usually around 2030 percent, its use accel-
erates rapidly. It then slows again as it nears completion . Te
graphic pattern is known as the S-curve in innovation difusion ,
with relatively slow (or fat) movement up to 2030 percent, a
rapid rise through the middle stages, and fattening again in the
fnal stages.
Lack of communal consciousness of linguistic changes
in progress is a consequence of its social stratifcation (see
inequality, linguistic and communicative ). Changes
generally progress incrementally, so that diferences between
the most proximate age groups are small and barely noticeable.
In Figure 1, 30-year-olds difer from 40-year-olds by about 10
percent and from 20-year-olds almost not at all. Te difer-
ence between 30-year-olds and 70-year-olds, by contrast, is
over 60 percent. Social relations are closest among age-mates,
and the gradation of diferences so that proximate age groups
are most like one another blunts the perception of genera-
tion gaps within the community. By minimizing awareness of
changes as they progress, social gradation is a unifying force in
communities .
years, between 8 and 18, and maintain them throughout their
lives. People who grow up saying sneaked as the past tense
of sneak tend to use that form all their lives, even after they
come to know that younger people in their region say snuck
instead. Because of this stratifcation, the progress of a linguis-
tic change is not measurable in the life span of an individual
or one age group but only in comparison between individuals
whose formative years are not the same, that is, between dif-
ferent age groups .
Change in Progress
Correlating linguistic variants with their use by age groups in
the community as the change is taking place is a way of measur-
ing its progress. Inductively, change is evident when a linguistic
variant occurs with greater frequency in the speech of younger
people than in the speech of their elders. Tere are some excep-
tions (such as age grading), but the inference of change can be
made with reasonable confdence when the frequency of the
variant is stratifed from one age group to the next (Labov 2001 ).
Tat is, linguistic changes are almost never bimodal, with one
variant occurring in the speech of younger people and a difer-
ent one in the speech of older people. Instead, the variants are
typically dispersed along the age continuum in a progressive
gradation.
Figure 1 provides a case study. Te variable is the past tense of
the verb sneak , with variants sneaked and snuck . Te community
is the Golden Horseshoe, the densely populated region in south-
ern Ontario, Canada. Figure 1 shows the percentage of people
who say snuck , not sneaked , correlated with their age, from octo-
genarians to teenagers. Te correlation shows a progression from
18 percent in the oldest group to 98 percent in the youngest, with
gradation in the intermediate age groups (29 percent of 70s, 42
percent of 60s, and so on).
Other things being equal, it is possible to draw historical
inferences from apparent-time displays like Figure 1. The sur-
vey from which the data are drawn took place in 1992. Among
people born 80 or more years prior, that is, before 1913,
sneaked was the standard variant and snuck had very little
100
90
80
70
60
50
40
30
20
0
v
e
r
8
0
7
0
7
9
6
0
6
9
5
0
5
9
4
0
4
9
3
0
3
9
2
0
2
9
1
4
1
9
10
0
Figure 1. Percentage of people in different age groups who
say snuck , not sneaked , as past tense of sneak in the
Golden Horseshoe, Canada (Chambers 2002a , 36466).
Age groups
90
of how languages change, who the agents of change are, and how
changes difuse throughout communities. Te sociolinguistic
perspective on language change as dynamic, progressive, and
variable represents an advance in language studies. Te prin-
cipal theoretical construct, the apparent-time hypothesis, pro-
vides a comprehensive view of historical sequences from a single
methodological vantage point.
J. K. Chambers
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chambers , J. K. 2002 a. Patterns of variation including change. In Te
Handbook of Language Variation and Change , ed. J. K. Chambers ,
Peter Trudgill, and Natalie Schilling-Estes , 34972. Oxford : Blackwell
Publishing .
. 2002 b. Studying language variation: An informal epistemol-
ogy. In Te Handbook of Language Variation and Change , ed.
J. K. Chambers , Peter Trudgill, and Natalie Schilling-Estes , 314.
Oxford : Blackwell Publishing .
. 2009 . Sociolinguistic Teory: Linguistic Variation and Its Social
Signifcance . 3
rd
ed. Oxford : Blackwell .
Labov , William. 2001 . Principles of Linguistic Change: Social Factors .
Oxford : Blackwell Publishing .
Sapir , Edward. 1929 . Te status of linguistics as a science . Language
5 : 207 14.
Weinreich , Uriel , William Labov , and Marvin I. Herzog. 1968 . Empirical
foundations for a theory of language change. In Directions for
Historical Linguistics: A Symposium , ed. Winfred P. Lehmann and
Yakov Malkiel, 95188. Austin : University of Texas Press .
AGING AND LANGUAGE
It is well-documented that in healthy aging, some aspects of lin-
guistic ability, for example, phonology, syntax, and vocab-
ulary, remain generally preserved into very old age. However,
other domains of language, for example, naming, comprehen-
sion and spoken discousre, undergo declines, albeit only at very
late stages in the adult life span. It is interesting to note that
although these linguistic changes occur generally in the older
adult population as a group, there are many older adult individu-
als who do not experience these defcits but, rather, continue to
perform as well as younger individuals. Tus, the fnding of great
inter-individual variability should be considered.
Tis entry presents a detailed review of the three main linguis-
tic domains experiencing decline with age. Additionally, a review
of language changes due to cognitive deterioration, for example,
dementia, is presented.
Naming
A common complaint among the healthy aging population is the
increased frequency of word-fnding problems in their everyday
speech. A subset of these naming problems is often colloqui-
ally described as the tip-of-the-tongue (TOT) phenomenon. In
terms of cognitive models of lexical access, TOTs are described
as a type of word retrieval failure whereby individuals are able to
access the conceptual, semantic , syntactic , and even some
phonological /orthographic information (e.g., number of
syllables or initial sound/letters) of the target word, not enough
information, however, to fully phonologically encode the word
Apparent-Time Hypothesis
Figure 1 draws historical inferences of change based on the
behavior of age groups surveyed at the same time. Te replace-
ment of sneaked by snuck is not directly observed as it would be
in real-time studies, in which researchers go to communities at
intervals and track the changes from one time to the next. Instead,
the inference of change is based on the assumption that under
normal circumstances, people retain the accents and dialects
acquired in their formative years. Tat assumption is known as
the apparent-time hypothesis . Common experience tells us that
the hypothesis is not without exceptions. People are sometimes
aware of expressions that they once used and no longer do, and
sometimes they will have changed their usage after their forma-
tive years. If such linguistic capriciousness took place throughout
the entire community, it would invalidate historical inferences
drawn from apparent-time surveys.
However, community-wide changes beyond the formative
years are rare. Real-time evidence, when available, generally
corroborates apparent-time inferences. In the case of sneaked/
snuck , for instance, earlier surveys made in the same region in
1950 and 1972 show proportional distributions of the variants
that approximate the apparent-time results. However, inferring
linguistic changes from the speech of contemporaneous age
groups is not a direct observation of that change. It remains a
hypothesis, and its validity must be tested wherever possible.
Age-Graded Changes
Deeper understanding of linguistic change in progress should
ultimately lead to predictable classes of deviation from the
apparent-time hypothesis. One known deviation is age-graded
change. Tese are changes that are repeated in each generation,
usually as people reach maturity (Chambers 2009 , 200206).
Age-graded changes are usually so gradual as to be almost
imperceptible, so that tracking their progress is challenging.
As an example, in all English-speaking communities, there is a
rule of linguistic etiquette that requires compound subject noun
phrases (NPs) to list the frst-person pronoun (the speaker) last.
Adults say Robin and I went shopping, and never say I and
Robin went shopping. Tere is no linguistic reason for this rule
(that is, the sentences mean the same thing either way), but
putting oneself frst is considered impolite (see politeness ).
Children, however, do not know this and tend to say Me and
Robin went shopping. At some as-yet-undetermined age, chil-
dren become aware of the rule and change their usage to con-
form to adult usage.
Age-graded changes like these violate the apparent-time
hypothesis because the variants used by young people do not
persist throughout their lifetimes. Instead, young people change
as they reach maturity and bring their usage into line with adults.
Te occurrence of age-graded changes does not refute the appar-
ent-time hypothesis, but they provide a well-defned exception
to it. Failure to recognize them as age-graded can lead to an erro-
neous inference that change is taking place .
Age and Language Change
Introducing age groups into linguistic analysis as an indepen-
dent variable yielded immediate insights into the understanding
Age groups Aging and Language
91
that older adults beneft more from context in noise than do their
younger cohorts.
Lexical comprehension is one aspect of language that is
largely preserved or even enhanced with age. Studies on vocabu-
lary comprehension in older adults using a picture selection task
for auditorily presented words show that they are comparable to
younger adults in this task (Schneider, Daneman, and Pichora-
Fuller 2002 ). However, lexical-semantic integration at the level of
sentence comprehension may be afected in older adults.
Sentence comprehension in the elderly is known to be poor
in comparison to younger listeners. A number of reasons, both
linguistic and cognitive, have been discussed, including decline
in auditory perceptual skills and lexical-semantic and syntactic
processing capacity, as well as working memory capacity, speed
of processing, and ability to process with competing stimuli and
inhibit noise (Wingfeld and Stine-Morrow 2000 ). Studies on syn-
tactic comprehension of language in older adults demonstrate
that they are slower at judging sentences that are syntactically
complex and semantically improbable (Obler et al. 1991 ). It has
also been found to be relatively more dif cult for this population
than for younger adults to take advantage of constraining context
in a sentence. Tis leads to dif culties in situations where older
adults need to rely on context but are not able to do so skillfully.
However, in other aspects of sentence comprehension such as
disambiguation, no age-related diferences has been reported.
It is apparent that sentence comprehension is largely medi-
ated by the ability to hold these sentences in working memory
while they are being processed. Working-memory decline has
been reported in older adults (Grossman et al. 2002 ). Moreover,
syntactic-processing dif culties have also been attributed to
reduction in working-memory capacity (Kemptes and Kemper
1997 ). Executive functions such as inhibition and task switching
have been reported as negatively afected in the elderly.
Comprehension problems eventually afect the older adults
discourse abilities. Often, the elderly fnd that it gets harder to
follow discourse with advancing age. Studies have shown that
older adults are signifcantly poorer than younger adults in fully
understanding and inferring complex discourse, and this dif -
culty is enhanced further with increased perceptual or cognitive
load (Schneider, Daneman, and Pichora-Fuller 2002 ), such as
noise and length or complexity of the material. A combination of
general cognitive decline and a deterioration of specifc linguis-
tic and sensory-perceptual processes contribute to the general
slowing observed in the elderly while they engage in conversa-
tion and discourse.
Spoken Discourse
Results from many types of tasks requiring sentence produc-
tion, such as cartoon description tasks (Marini et al. 2005 ), have
indicated that older adults tend to produce grammatically sim-
pler, less informative and more fragmented sentences than do
younger adults. Middle-aged and young elderly adults tend to
be better and more ef cient story constructors than younger
or older adults. However, older adults usually produce a larger
number of words in their narrative speech, but they can show
dif culties in morpho-syntactic and lexico-semantic processing
by making more paragrammatic errors and semantic parapha-
sias than younger adults.
for articulation . Research clearly supports the view that there is a
general breakdown in phonological encoding (James and Burke
2000 ). However, since this stage of processing involves various
substages (e.g., segment retrieval, syllable encoding) that occur
at very fast rates, as per Willem J. M. Levelt, Ardi Roelofs and
Antje S. Meyers ( 1999 ) model, behavioral methods are limited
in their ability to identify the locus of processing dif culty.
Evidence from priming studies has strongly demonstrated
that TOTs are due to a failure in transmission of available
semantic information to the phonological system, as explained
in the transmission defcit hypothesis (TDH; Burke et al. 1991 ).
Tis theory proposes that older people are especially prone to
word-retrieval problems due to weakened connections at the
phonological level. Te phonological level, as compared to
the semantic level, is particularly vulnerable to breakdowns
in retrieval because this level generally has fewer connections
(e.g., phoneme -sound), whereas the semantic system has mul-
tiple joined connections (e.g. many words/concepts linked to
a given word). Factors such as word frequency or recency of
use infuence the strength of phonological connections; that
is, the lower the word frequency and the less recently used the
word, the weaker the connections, leading to greater retrieval
dif culties.
Both younger and older individuals beneft from phonologi-
cal primes as opposed to unrelated primes during moments of a
TOT state, supporting the TDH model. Te fnding that priming
leads to better retrieval is consistent with the claim that priming
strengthens the inherently weak phonological connections and
thus facilitates resolution of the TOT state. Furthermore, stud-
ies using semantic versus phonological cues have demonstrated
that in both younger and older people, provision of phonological
information was more efective, as a retrieval aid, than semantic
cues (Meyer and Bock 1992 ). Tis illustrates that in older indi-
viduals, semantic information is intact, although there is some
degradation in the use of phonological information.
In summary, much evidence supports the TDH and the claim
that the locus of the breakdown in TOT states is at the phono-
logical stage. Te exact phonological substage responsible for
this problem still remains unclear; however, there are indica-
tions from phonological cuing studies, self-reports of individuals
experiencing TOT states (Brown 1991 ), and an electrophysiolog-
ical study of lexical retrieval in healthy younger and older adults
(Neumann 2007 ) that the frst two substages (segmental and syl-
labic retrieval) are particular points at which breakdowns occur.
Language Comprehension
It is relatively well known that comprehension in older adults
is compromised in comparison to younger adults. Tese prob-
lems in comprehension may arise from individual or combined
efects of decline in their sensory/perceptual, linguistic, or cog-
nitive domains. Current research is focused on disambiguating
the efects of these processes on phonological, lexical-semantic,
and syntactic aspects of language decline in the elderly.
Research shows that speech perception in older adults is espe-
cially afected by noise. Tis means that even with normal hear-
ing, older adults experience dif culties understanding speech
(e.g., sentences) under noisy conditions such as a cocktail party
or cafeteria. Experiments focusing on sentence perception show
Aging and Language
92
Garrard , Peter , Matthew A. Lambon Ralph , John R. Hodges , and Karalyn
Patterson. 2001 . Prototypicality, distinctiveness, and intercorrela-
tions: Analysis of the semantic attributes of living and nonliving con-
cepts. Cognitive Neuropsychology 18.2 : 125 74.
Grossman , Murray , Ayanna Cooke , Christian DeVita , David Alsop , John
Detre , Willis Chen , and James Gee. 2002 . Age-related changes in
working memory during sentence comprehension: An fMRI study .
NeuroImage 15 : 302 17.
James , Lori E. , and Deborah M. Burke. 2000 . Phonological priming
efects on word retrieval and tip-of-the-tongue experiences in young
and older adults. Journal of Experimental Psychology: Learning,
Memory, and Cognition 26.6 : 1378 91.
James , Lori E. , Deborah M. Burke , Ayda Austin , and Erika Hulme. 1998 .
Production and perception of verbosity in younger and older adults .
Psychology and Aging 13 : 355 67.
Kemper , Susan , Ruth E. Herman , and Chiung-Ju Liu . 2004 . Sentence
production by young and older adults in controlled contexts . Journal
of Gerontology Series B: Psychological Sciences and Social Sciences
59 : 220 4.
Kemptes , Karen A. , and Susan Kemper 1997 . Younger and older adults
on-line processing of syntactically ambiguous sentences. Psychology
and Aging 12 : 362 71.
Levelt , Willem J. M. , Ardi Roelofs , and Antje S. Meyer 1999 . A theory
of lexical access in speech production . Behavioral Brain Sciences
22 : 1 75.
Marini , Andrea , Anke Boewe , Carlo Caltagirone , and Sergio Carlomagno.
2005 . Age- related diferences in the production of textual descrip-
tions. Journal of Psycholinguistic Research 34 : 439 63.
Meyer , Anje S. , and Kathryn Bock. 1992. Te tip-of-the-tongue phe-
nomenon: Blocking or partial activation? Memory and Cognition
20 : 715 26.
Morris , John , Martha Storandt , J. Phillip Miller , Daniel W. McKeel , Joseph
L. Price , Eugene H. Rubin , and Leonard Berg. 2001 . Mild cognitive
impairment represents early-stage Alzheimers disease. Archives of
Neurology 58.3 : 397 405.
Neumann , Yael . 2007 . An Electrophysiological Investigation of the Efects
of Age on the Time Course of Segmental and Syllabic Encoding dur-
ing Implicit Picture Naming in Healthy Younger and Older Adults .
Publications of the Department of Speech, Language, and Hearing
Sciences. New York : City University of New York .
Nordlund , Arto , S. Rolstad , P. Hellstrm , M. Sjgren , S. Hansen , and A.
Wallin. 2004 . Te Goteborg MCI study: Mild cognitive impairment is
a heterogeneous condition . Journal of Neurology, Neurosurgery, and
Psychiatry 76 : 1485 90.
Obler , Loraine K. , Deborah Fein , Marjorie Nicholas , and Martin L. Albert
1991 . Auditory comprehension and aging: Decline in syntactic pro-
cessing. Applied Psycholinguistics 12 : 433 52.
Pekkala , Seija. 2004 . Semantic Fluency in Mild and Moderate Alzheimers
Disease. Publications of the Department of Phonetics 47, University of
Helsinki . Available online at: http://ethesis.helsinki.f/.
Reuter-Lorenz , Patricia A. , John Jonides , Edward E. Smith , Alan Hartlye ,
Andrea Miller , Christina Marshuetz , and Robert A. Koeppe. 2000 . Age
diferences in the frontal lateralization of verbal and spatial working mem-
ory revealed by PET . Journal of Cognitive Neuroscience 12 : 174 87.
Schneider , Bruce A. , Meredith Daneman , and M. Kathleen Pichora-
Fuller . 2002 . Listening in aging adults: From discourse comprehen-
sion to psychoacoustics. Canadian Journal of Experimental Psychology
56 : 139 52.
Vuorinen , Elina , Matti Laine , and Juha Rinne . 2000 . Common pattern
of language impairment in vascular dementia and in Alzheimers dis-
ease . Alzheimer Disease and Associated Disorders 14 : 81 6.
Wingfeld , Arthur , and Elizabeth A. L. Stine-Morrow . 2000 . Language
and aging. In Te Handbook of Cognition and Aging , ed. Craig and
Salthouse , 359416. Mahwah, NJ : Psychology Press .
Older adults also show a decreased ability to coherently link
adjacent utterances in a story. Adults older than 75 use a larger
number of missing or ambiguous referents and more units of
irrelevant content that afect the coherence of the narratives
(Marini et al. 2005 ). Older people show a huge variation in story-
telling abilities, and they can also compensate for their storytell-
ing dif culties due to their greater accumulated life experience,
which they can use to combine diferent themes and to empha-
size relevant details.
In spoken discourse, some changes in conversational-in-
teraction style can occur in young elderly people (6074 years),
but the most noticeable changes are likely to take place in older
elderly people (7788 years), who show excessive verbosity, fail-
ure to maintain topic, poor turn-taking, and unclear referencing
(James et al. 1998 ).
Dif culties in grammatical processing in the aging can be
attributed to cognitive deterioration involving reduced working
memory (Kemper, Herman, and Liu 2004) and inhibitory defcits
(James et al. 1998 ), but they can also be a sign of impaired access
to the lemma level during lexical retrieval (Marini et al. 2005 ).
In summary, there are clear diferences between younger and
older adults in their sentence production, storytelling, and con-
versational abilities. Tese diferences are manifest as changes
in morpho-syntactic and lexico-semantic processing, excessive
verbosity, and reduced informativeness and coherence .
Dementia
Dif culties in language production, as well as comprehension,
become obvious in dementia. Word-fnding dif culties are char-
acteristic of mild cognitive impairment (MCI), Alzheimers dis-
ease (AD), and vascular dementia (VaD).
Arto Nordlund and colleagues ( 2004 ) indicated that 57.1 per-
cent of the individuals with MCI had signifcantly lower scores
in diferent language tasks than did typical aging adults. In AD,
language-processing dif culties are early signs of the disease.
In particular, AD appears to cause a breakdown of the seman-
tic domain of language, which can be refected in the impaired
comprehension and use of semantic relations between words
(Garrard et al. 2001) and in the reduced semantic noun and verb
fuency performance (Pekkala 2004 ) in this population.
Comparative studies between AD and VaD have indicated that
cognitive and linguistic performance cannot clearly diferentiate
the two types of dementia from each other. Elina Vuorinen, Matti
Laine, and Juha Rinne ( 2000 ) indicated that both AD and VaD
involved similar types of semantic defcits early in the disease,
including dif culties in comprehension, naming, and produc-
tion of semantic topics in narrative speech, while word repeti-
tion, oral reading, and fuency of speech output were preserved
in both types of dementia.
Yael Neumann , Seija Pekkala , and Hia Datta
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brown , Alan S. 1991 . A review of the tip-of-the-tongue experience .
Psychological Bulletin 109.2 : 204 23.
Burke , Deborah M. , Donald G. MacKay , Joanna S. Worthley , and
Elizabeth Wade. 1991 . On the tip of the tongue: What causes word
fnding failures in young and older adults? Journal of Memory and
Language 30 : 542 79.
Aging and Language
93
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aoun , J. and D. Sportiche . 1983 . On the formal theory of government.
Linguistic Review 2 : 211 36.
Barlow , M. , and C. Barlow , eds. 1988. Agreement in Natural Language .
Stanford, CA : CSLI Publications .
Boeckx , C. , ed. 2006. Agreement Systems , Amsterdam : Benjamins .
Chomsky , N. 2000 . Minimalist inquiries: the framework. In Step by
Step, ed. R. Martin , D. Michaels, and J. Uriagereka , 89155 . Cambridge,
MA : MIT Press
Covington , M. 1984 Syntactic Teory in the High Middle Ages ,
Cambridge : Cambridge University Press .
AGREEMENT MAXIMIZATION
Maximizing, or optimizing, agreement between speaker and
interpreter is part of the principle of charity (see charity, prin-
ciple of ) that, according to philosophers of language in the
tradition of W. V. O. Quine and Donald Davidson, governs the
interpretation of the speech and thought of others and guides
the radical interpretation of a radically foreign language .
According to Davidson, correct interpretation maximizes truth
and coherence across the whole of the beliefs of a speaker. For an
interpreter, maximizing truth across the beliefs he ascribes to a
speaker necessarily amounts to maximizing agreement between
himself and the speaker: He can only go by his own view of what
is true.
Take one of Davidsons own examples: Someone says Tere
is a hippopotamus in the refrigerator, and he continues: Its
roundish, has a wrinkled skin, does not mind being touched.
It has a pleasant taste, at least the juice, and it costs a dime. I
squeeze two or three for breakfast ( 1968 , 100). Te simplest way
of maximizing agreement between this speaker and us will prob-
ably be interpreting his expression hippopotamus as meaning
the same as our expression orange. Davidson himself, however,
soon came to consider maximizing agreement as a confused
ideal ( 1984 , xvii) to be substituted with optimizing agreement
( 1975 , 169). Te idea here is that some disagreements are more
destructive of understanding than others ( 1975 , 169). Very
generally, this is a matter of epistemic weight; the more basic a
belief is, and the better reasons we have for holding it, the more
destructive disagreement on it would be: Te methodology of
interpretation is, in this respect, nothing but epistemology seen
in the mirror of meaning ( 1975 , 169). According to Davidson, it
is impossible to codify our epistemology in simple and precise
form, but general principles can be given: [A]greement on laws
and regularities usually matters more than agreement on cases;
agreement on what is openly and publicly observable is more
to be favored than agreement on what is hidden, inferred, or ill
observed; evidential relations should be preserved the more they
verge on being constitutive of meaning ([ 1980 ] 2004, 157).
Agreement optimization does not exclude the possibility of
error; speakers are to be interpreted as right only when plau-
sibly possible (Davidson 1973 , 137). In certain situations this
prevents the interpreter from ascribing beliefs of his own to
the speaker, for instance, perceptual beliefs about objects the
speaker is in no position to perceive. Moreover, if the speaker
has other beliefs that provide him or her with very good reasons
for believing something false, optimizing agreement across all
AGREEMENT
Agreement is a form of featural dependency between diferent
parts of a sentence: Te morphological shape of a word is a func-
tion of particular morphological features of a diferent, often dis-
tant, word. Since the Middle Ages, agreement was taken to be in
complementary distribution with government , and, hence, it
became important to determine both the context of each type of
relation and the reasons why this diference exists (Covington
1984 ). Tis rich tradition has survived in generative grammar, all
the way up to the minimalist program (see mimimalism ), where
it is embodied under the notion agree .
Depending on the features and occurrences in an expression,
agreement can be characterized as external or internal (Barlow
and Fergusson 1988). External agreement typically involves per-
son and number features, taking place between verbs and cor-
responding dependents. We can witness it in you are friendly
versus he is friendly . Internal agreement (concord) normally
involves gender and number features, and typically takes place
internal to nominal expressions, between adjectives or relative
clauses and the head noun, freely iterating. Concord is easily
observed in modifed nominal expressions in Spanish: atractiv as
dam as attractive ladies versus atractiv os caballer os attractive
gentlemen (agreeing elements are boldfaced). Genitive agree-
ment internal to nominal expressions normally falls within the
external (not the internal) rubric.
Te principles and parameters system concentrated on
external agreement, through the relation (head, specifer) (Aoun
and Sportiche 1983 ). However, since agreement is possible also
in situations where no such relation seems relevant, the mini-
malist program (Chomsky 2000 ) proposes a relation between
a probe and a goal. Te probe contains a value-less attribute in
need of valuation from a distant feature of the same type, which
the probing mechanism achieves. Te goal cannot be contained
within a derivational cycle (a phase ) that is too distant from the
probe. When the probe fnds an identical category within its com-
plement domain, it attempts to get its valuation from it, thereby
sanctioning the relevant agreement. To illustrate, observe the
Spanish example in (1); note also the internal agreement mani-
fested within the noun phrase:
(1) Parec en [haber quedado [l os loc os soldad os ] en la guarnicin]
seem-3rd/pl. have remained the-m./pl. crazy-m./pl.
soldiers-m./pl. in the garrison
(2) Probe
1
[ [Goal]]] (plus iteration of - os within the nominal)
( ) person pl. number
( ) number m. gender
Agreement adds a strange redundancy to the language faculty.
In languages where the phenomenon is overt, the extra layer of
manifest dependency correlates with difering surface orders. But
it is unclear whether that justifes the linguistic emergence of the
agreement phenomenon, particularly since the probe/goal mech-
anism can be present without overt manifestations. Tis results in
much observed variation, from the almost total lack of overt agree-
ment of Chinese to the poly-personal manifestation of Basque .
Juan Uriagereka
Agreement Agreement Maximization
94
. 1975 . Tought and talk. In Davidson 1984, 15570.
. [ 1980] 2004 . A unifed theory of thought, meaning, and action.
Problems of Rationality, 15166. Oxford : Clarendon Press.
. [ 1983] 2001 . A coherence theory of truth and knowledge.
Subjective, Intersubjective , Objective , 13753. Oxford : Clarendon Press .
. 1984 . Inquiries into Truth and Interpretation . Oxford : Clarendon
Press.
. [ 1991] 2001 . Tree varieties of knowledge. Subjective,
Intersubjective, Objective , 20520. Oxford : Clarendon Press .
. 2001 . Externalisms. Interpreting Davidson , ed. P. Kotatko , P.
Pagin , and G. Segal , 116. Stanford, CA : CSLI .
Gler , Kathrin . 2006 . Triangulation. Te Oxford Handbook of Philosophy
of Language , ed. E. Lepore and B. Smith , 100619. Oxford : Oxford
University Press .
Grandy , Richard . 1973 . Reference, meaning, and belief. Journal of
Philosophy 70 : 439 52.
Kripke , Saul . 1972 . Naming and Necessity . Cambridge : Harvard University
Press .
ALLITERATION
Linguistically, alliteration, also known as initial or head rhyme ,
is defned as the selection of identical syllable onsets within a
specifc phonological , morphosyntactic, or metrical domain.
It is usually coupled with stress, as in the three Rs in educa-
tion: reading, writing, and arithmetic. Etymologically, the term
alliteration (from L. ad- to + littera letter) includes the repe-
tition of the same letters at word beginnings; its dual association
with sounds and letters refects a common cognitive crisscross-
ing between spoken and written language in highly literate
(see literacy ) societies well illustrated by the famous phrase
apt alliterations artful aid, where the alliteration is primarily
orthographic .
Alliteration based on the sameness of letters is found in visual
poetry, advertising, and any form of playful written language.
Phonologically based alliteration is a frequent mnemonic and
cohesive device in all forms of imaginative language: Examples
from English include idioms ( beat about the bush ), reduplicative
word-formation ( rifraf ), binominals ( slowly but surely ), catch
phrases, refrains, political slogans, proverbs, and clichs. In
verse , alliteration serves both as ornamentation and as a struc-
tural device highlighting the metrical organization into feet, cola,
verses, and lines. Along with rhyme, alliteration is a common fea-
ture of folk and art verse in languages as diverse as Irish, Shona,
Mongolian, Finnish, and Somali .
Te most frequent type of alliteration requires identity of the
onsets of stressed syllables, which makes it a preferred poetic
device in languages with word-initial stress, such as the older
Germanic languages. Within the Germanic tradition, metrically
relevant alliteration occurs on the stressed syllables of the frst
foot of each verse (or half-line), where it is obligatory. For Old
English, the language of the richest and most varied surviv-
ing alliterative poetry in Germanic, the second foot of the frst
half-line may also alliterate. Alliteration is disallowed on the last
stressed syllable in the line.
In Old English verse, alliteration appears with remark-
able regularity: Only 0.001% of the verses lack alliteration and
less than 0.05% contain unmetrical alliteration (Hutcheson
1995 , 169). Alliteration is, therefore, a reliable criterion used by
of his or her beliefs might well require ascription of outright
mistakes.
Optimizing agreement provides an interpreter with a method
for arriving at correct interpretations because of the way belief
content is determined, Davidson holds. Te arguments for this
claim have changed over the years; initially, the idea was that a
belief is identifed by its location in a pattern of beliefs; it is this
pattern that determines the subject matter of the belief, what the
belief is about ( 1975 , 168). Later, however, the role played by
causal connections between objects and events in the world and
the beliefs of speaker and interpreter becomes more and more
prominent: In the most basic, perceptual cases, the interpreter
interprets sentences held true (which is not to be distinguished
from attributing beliefs) according to the events and objects in
the outside world that cause the sentence to be held true ([ 1983 ]
2001,150 ). In the later Davidson, the account of content deter-
mination underlying the method of charitable interpretation
takes the form of a distinctive, social and perceptual meaning
externalism : In the most basic cases, the objects of thought
are determined in a sort of triangulation as the shared causes
of the thoughts of two interacting persons, for instance, a child
and its teacher (cf. Davidson [ 1991 ] 2001, 2001 ). According to
Davidson, such triangulation is a necessary condition for thought
with empirical content; moreover, he derives a quite far-reaching
epistemic antiskepticism from it and claims that belief is in its
nature veridical ([ 1983 ] 2001, 146; cf. also [ 1991 ] 2001, 211f).
Probably the most infuential argument against the idea that
any kind of maximizing agreement results in correct interpreta-
tion derives from Saul Kripkes attack on description theories
of proper names. According to such theories, the referent of a
proper name, for instance, Gdel is determined by a descrip-
tion, or cluster of descriptions, held true by the speaker(s), for
instance, the description the discoverer of the incompleteness
of arithmetic. Kripke argued, among other things, that such
theories fail because all of the relevant descriptions, all of the rel-
evant beliefs that a speaker, or even a group of speakers, holds
about Gdel could turn out to be false (cf. 1972 , 83f). Kripke gave
an analogous argument for natural-kind terms such as gold or
tiger , and many philosophers today believe that these arguments
can be generalized even further. While it is quite clear, however,
that most of the descriptions a speaker associates, for instance,
with a name could turn out to be false when taken one by one,
it is far less obvious that all (or most, or a weighted majority) of
them could do so at the same time . According to Davidson, for
instance, optimizing agreement amounts to reference determi-
nation by epistemically weighted beliefs. A signifcant number
of these are very elementary beliefs like the belief that Gdel
was a man, that he was human, that he worked on logic, that he
lived on Earth, and so on. If a speaker did not believe any of these
things about Gdel, it has been argued with Davidson, it is far
less clear that this speaker was in fact talking about Gdel: Too
much mistake simply blurs the focus (Davidson 1975 , 168) .
Kathrin Gler
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Davidson , Donald. 1968 . On saying that. In Davidson 1984, 93108.
. 1973 . Radical interpretation. In Davidson 1984, 12539.
Agreement Maximization Alliteration
95
Danny saw the man with the telescope , either Danny used the
telescope to help him see the man (3) or the man whom Danny
saw had a telescope (4).
3. Danny [
VP
saw [
NP
the man] [
PP
with the telescope]]
4. Danny [
VP
saw [
NP
the man [
PP
with the telescope]]]
Referential ambiguity occurs when it is not clear which entity
in a context is being referred to by the given linguistic expression.
Although deictics (see deixis ), such as pronouns, are typical
sources of referential ambiguity, full noun phrases and proper
nouns can also give rise to it.
5. (at a boys soccer game) He kicked him ! Who kicked who?
6. (at a boat race) Tat boat seems to be pulling ahead. Which
one?
7. (in a university corridor) Im of to meet with Dr. Sullivan .
Chemistry or math? (Tere are two Dr. Sullivans in difer-
ent departments.)
Scopal ambiguity occurs when a sentence contains more than
one quantifed NP and the interpretation depends on the rela-
tive scopes of the quantifers. For example, Some children saw
both plays can mean that a) there exist some children such that
each of them saw both plays or b) both plays were such that each,
individually, was seen by some children but not necessarily the
same children. Phonetic ambiguity arises when a given sound
pattern can convey diferent words, for example, two ~ too ~ to;
new deal ~ nude eel.
Although people typically do not notice the ambiguities
that they efortlessly resolve through context, they are certainly
aware of the potential for ambiguity in language. In fact, such
awareness is a precondition for getting the joke in Abbott and
Costellos Whos on First? skit or in headlines like Iraqi Head
Seeks Arms.
Whereas ambiguity does not frequently hinder efective
communication among people, it is among the biggest hurdles
for the machine processing of language. Tis is not surprising if
one considers how much reasoning is required to resolve ambi-
guity and how much knowledge of language, the context, and
the world must underpin such reasoning. As an example of the
large scale of the task, consider the short sentence Te coach
lost a set, which you probably interpreted to mean the person
who is the trainer of some athletic team experienced the loss of
a part of a match in an athletic competition (whether the coach
was playing or the team was playing is yet another ambiguity).
Other interpretations are also valid, given specifc contexts. For
example, the person who is the trainer of some team might have
lost a set of objects (keys, golf clubs) or a railroad car might have
lost a set of objects (door handles, ball bearings). If this sentence
were used as input to an English-Russian machine translation
system that relied on a standard English-Russian dictionary,
that system would have to select from among 15 senses of coach ,
11 senses of lose , and 91 senses of set a grand total of 15,015
combinations, if no further knowledge were brought to bear. Of
course, all machine translation systems incorporate some heu-
ristic knowledge, and lexicons developed for natural language
processing typically do not permit the amount of sense splitting
found in dictionaries for people. On the other hand, it is common
modern editors to determine the boundaries of verses and lines,
though no such divisions exist in the manuscripts. Te rein-
vented alliterative tradition of fourteenth-century England also
uses alliteration structurally, while its ornamental function is
enhanced by excessive verse-internal and run-on alliteration .
As a cohesive device in verse, alliteration refers to the under-
lying distinctions in the language and relies on identity of pho-
nological categories. Te interpretation of onset identity for the
purpose of poetic alliteration varies from tradition to tradition
and can include whole clusters, optionally realized segments,
and even the whole syllable up to the coda. In Germanic, all
consonants alliterated only with one another, the clusters st-,
sp-, sk- could not be split, and all orthographic stressed vowels
alliterated freely among themselves, most likely because their
identity was signaled by the presence of a prevocalic glottal stop
in stressed syllable onsets .
Donka Minkova
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Fabb , Nigel . 1997 . Linguistics and Literature: Language in the Verbal Arts
in the World . Oxford : Blackwell .
Hutcheson , Bellenden Rand . 1995 . Old English Poetic Metre.
Cambridge : D. S. Brewer .
Minkova , Donka . 2003 . Alliteration and Sound Change in Early English .
Cambridge : Cambridge University Press .
AMBIGUITY
Ambiguity refers to the potential of a linguistic expression to have
more than one meaning. Although many expressions ( words ,
phrases, and even sentences ) are ambiguous in isolation, few
remain so when used in a particular context. In fact, people typi-
cally resolve all ambiguities without even detecting the potential
for other interpretations. Ambiguity does not imply vagueness ;
rather, ambiguity gives rise to competing interpretations, each of
which can be perfectly concrete. Although ambiguity is pervasive
and unavoidable in natural languages, artifcial languages devel-
oped for mathematics, logic, and computer programming strive
to eliminate it from their expressions .
Ambiguity can be lexical , structural, referential, scopal,
or phonetic. Te examples of these phenomena that follow
include well-known classics in English.
Lexical ambiguity refers to the fact that some words, as writ-
ten or spoken, can be used in diferent parts of speech (see
word classes ) and/or with diferent meanings. For example,
duck can be used as a noun or a verb and, as a noun, can refer to
a live animal or its meat. Structural ambiguity arises when dif-
ferent syntactic parses give rise to diferent interpretations. For
example, in addition to being lexically ambiguous, Tey saw her
duck is also structurally ambiguous:
1. Tey saw [
NP
her duck] (the bird or its meat belongs to her)
2. Tey saw [
NP
her] [
VP
duck] (the ducking is an action she carries
out)
A common source of structural ambiguity involves the attach-
ment site for prepositional phrases, which can be at the level
of the nearest noun phrase (NP) or the clause. In the sentence
Ambiguity
96
cortical visual processing stream. Hence, through its extensive
connectivity with sensory processing regions, the amygdala is
ideally located to infuence perception based on emotion.
Language
In order to survive in a changing environment, it is especially
important for the organism to remember events and stimuli
that are linked with emotional consequences. Furthermore, it is
important to be vigilant of emotional stimuli in the environment
in order to allow for rapid evaluation of and response to these
emotional stimuli. In humans, emotional cues are transmitted
linguistically, as well as through body posture, voice, and facial
expression (see gesture ).
In the frst imaging study to examine language and the
amygdala, a modifed Stroop task was utilized, along with a
high-sensitivity neuroimaging technique, to target the neu-
ral substrate engaged specifcally when processing linguistic
threat (Isenberg et al. 1999 ). Healthy volunteer subjects were
instructed to name the color of words of either threat or neutral
valence, presented in diferent color fonts, while neural activity
was measured by using positron emission tomography. Bilateral
amygdalar activation was signifcantly greater during color nam-
ing of threat words than during color naming of neutral words
(see Color Plate 1 ). Associated activations were also noted in
sensory-evaluative and motor-planning areas of the brain. Tus,
our results demonstrate the amygdalas role in the processing
of danger elicited by language. In addition, the results reinforce
the amygdalas role in the modulation of the perception of, and
response to, emotionally salient stimuli. Tis initial study further
suggests conservation of phylogenetically older mechanisms of
emotional evaluation in the context of more recently evolved lin-
guistic function. In a more recent study that examines the neural
substrates involved when subjects are exposed to an event that is
verbally linked to an aversive outcome, activation is observed in
the left amygdala (Phelps et al. 2001, 43741).
Tis activation correlated with the expression of the fear
response as measured by skin conductance response, a periph-
eral measure of arousal. Te laterality of response may relate to
the explicit nature of the fear, as well as to the fact that the stim-
ulus is learned through verbal communication. Fears that are
simply imagined and anticipated nonetheless have a profound
impact on everyday behavior. Te previous study suggests that
the left amygdala is involved in the expression of fear when antic-
ipated and conveyed in language. Another study that sought to
examine the role of the amygdala in the processing of positive as
well as negative valence verbal stimuli also demonstrated activ-
ity in the left amygdala (Adolphs, Baron-Cohen, and Tranel 2002,
126474). During magnetic resonance (MR) scanning, subjects
viewed high-arousal positive and negative words and neutral
words. In this study, activity was found in the left amygdala while
subjects viewed both negative and positive words in comparison
to neutral words. Taken together, these studies suggest that the
amygdala plays a role in both positive and negative emotional
responses. Furthermore, they suggest that the left amygdala may
be preferentially involved in the processing of emotion as con-
veyed through language .
Lesion studies have generally suggested that the amygdala
is not essential for recognizing or judging emotional and social
for sentences to contain upward of 20 words, in which case there
is still the threat of combinatorial explosion .
Marjorie J. McShane
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Cruse , D. A. 1986 . Lexical Semantics . Cambridge : Cambridge University
Press. Includes features of and tests for ambiguity.
Small , Steven , Garrison Cottrell , and Michael Tanenhaus , eds. 1988 .
Lexical Ambiguity Resolution: Perspective from Psycholinguistics,
Neuropsychology and Artifcial Intelligence . San Mateo, CA : Morgan
Kaufmann.
Zwicky , Arnold M. , and Jerrold M. Sadock. 1975 . Ambiguity Tests and
How to Fail Tem. In Syntax and Semantics , ed. J. Kimball, IV: 136.
New York : Academic Press. Discusses tests to distinguish ambiguity
from lack of specifcation.
AMYGDALA
Studies in animals have established a clear role for the amygdala
in social and emotional behavior, especially as related to fear
and aggression (Le Doux 1996 ). Human studies, including lesion
studies, electrophysiology, and functional neuroimaging ,
have further elucidated the role of the amygdala in the process-
ing of a variety of emotional sensory stimuli, as well as its rela-
tionship to behavioral and cognitive responses (Adolphs 2001 ).
Tese responses not only guide social behavior but also aid in the
acquisition of social knowledge. Te focus of this entry is on the
amygdala and its role in the processing of language, in particu-
lar language relevant to social and emotional behavior (see also
emotion and language and emotion words ) .
Te amygdala is an almond-shaped group of neurons located
in the rostral medial temporal region on both left and right sides
of the brain (see left hemisphere language process-
ing and right hemisphere language processing ). It
has reciprocal connections to regions, such as the hypothala-
mus, that are important for coordinating autonomic responses
to complex environmental cues for survival, as well as premo-
tor and prefrontal areas that are necessary for rapid motor and
behavioral responses to perceived threat. Visual, somatosen-
sory, and auditory information is transmitted to the amygdala
by a series of indirect, modality-specifc thalamocorticoamygda-
lar pathways, as well as by direct thalamoamygdalar pathways.
Within the amygdaloid complex, information processing takes
place along numerous highly organized parallel pathways with
extensive intraamygdaloid connections. Te convergence of
inputs in the lateral nucleus enables stimulus representations
to be summated. Specifc output pathways from the central
nucleus and amygdalohippocampal area mediate complemen-
tary aspects of learning and behavioral expressions connected
with various emotional states. Te amydgala is thus well posi-
tioned to play a role in rapid cross-modal emotional recognition.
It is important for the processing of emotional memory and for
fear conditioning .
In addition, anatomical studies of the primate amygdala dem-
onstrate connections to virtually all levels of visual processing in
the occipital and temporal cortex (Amaral 2003 ). Terefore,
the amygdala is also critically placed to modulate visual input,
based on afective signifcance, at a variety of levels along the
Amygdala
97
In the aftermath of analogical reasoning about a pair of cases,
some form of relational generalization may take place, yielding a
schema for a class of situations (Gick and Holyoak 1983 ) .
Psychological Research
Within psychology, work in the intelligence tradition focused on
four-term or proportional analogies, such as ARM: HAND :: LEG: ?
Charles Spearman ( 1946 ) reviewed studies that found high cor-
relations between performance in solving analogy problems and
the g factor (general intelligence). Te ability to solve analogy-
like problems depends on a neural substrate that includes sub-
areas of the prefrontal cortex (Bunge, Wendelken, and Wagner
2005 ; see frontal lobe ). Although there have been reports of
great apes being successfully trained to solve analogy problems,
these results are controversial (Oden, Tompson, and Premack
2001 ). Complex relational thinking appears to be a capacity that
emerged in homo sapiens along with the evolutionary increase
in size of the frontal cortex. Te ability to think relationally
increases with age (Gentner and Rattermann 1991 ). Greater
sensitivity to relations appears to arise with age due to a com-
bination of incremental accretion of knowledge about relational
concepts (Goswami 1992 ), increases in working memory
capacity (Halford 1993 ), and increased ability to inhibit mis-
leading featural information (Richland, Morrison, and Holyoak
2006 ). Analogy plays a prominent role in teaching mathematics
(Richland, Zur, and Holyoak 2007 ).
Dedre Gentner ( 1983 ) developed the structure-mapping the-
ory of analogy, emphasizing that analogical mapping is guided
by higher-order relations relations between relations. Keith
Holyoak and P. Tagard ( 1989 ) proposed a multiconstraint the-
ory, hypothesizing that people fnd mappings that maximize
similarity of corresponding elements and relations, structural
parallelism, and pragmatic importance for goal achievement.
Several computational models of human analogical thinking
have been developed. Two infuential models are SME (Structure
Mapping Engine; Falkenhainer, Forbus, and Gentner 1989 ),
based on a classical symbolic architecture, and LISA (Learning
and Inference with Schemas and Analogies; Hummel and
Holyoak 2005 ), based on a neural-network architecture. LISA has
been used to simulate some efects of damage to the frontal and
temporal cortex on analogical reasoning (Morrison et al. 2004 ) .
Analogy and Language
Analogy is related to metaphor and similar forms of symbolic
expression in literature and everyday language. In metaphors,
the source and target domains are always semantically distant
(Gentner, Falkenhainer, and Skorstad 1988 ). Rather than simply
comparing the source and target, the target is identifed with the
source (Holyoak 1982 ), either directly (e.g., Juliet is the sun)
or by applying a predicate drawn from the source domain to the
target (e.g., Te romance blossomed) . As a domain-general
learning mechanism linked to human evolution, analogy ofers
an alternative to strongly nativist views of language acquisition
(Vallauri 2004 ; see innateness and innatism ). Gentner and
L. L. Namy ( 2004 ) review evidence that analogical comparison
plays important roles in speech segmentation, word learning,
and possibly acquisition of grammar .
Keith Holyoak
information from explicit, lexical stimuli, such as stories (see
narrative, neurobiology of ) (Amaral 2003, 33747).
However, while recognition of emotional and social information
may be relatively preserved in amygdalar damage, the aware-
ness that unpleasant emotions are arousing appears to be lost. In
a lesion study, normal subjects judge emotions such as fear and
anger to be both unpleasant and highly arousing; however, patient
S. M. 046, who sustained early amygdalar damage, judged these
same stimuli to be unpleasant but of low arousal. For example,
when told a story about someone driving down a steep mountain
who had lost the car brakes, she identifed the situation as unpleas-
ant but also gave a highly abnormal judgment that it would make
one feel sleepy and relaxed. It is interesting to note that S. M. 046
was able to judge arousal normally from positive emotions .
Te human amygdala is important both for the acquisition
and for the online processing of emotional stimuli. Its role is dis-
proportionate for a particular category of emotional information,
that pertaining to the evaluation of potential threat in the envi-
ronment. Te amygdalas role in enhanced, modality-specifc
processing required for the rapid evaluation and response to
threat is highlighted. Furthermore, this review suggests conser-
vation of phylogenetically older limbic mechanisms of emotional
evaluation in the context of more recently evolved language .
Nancy B. Isenberg
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Adolphs , R. 2001 . Te neurobiology of social cognition . Current
Opinions in Neurobiology 11 .2: 231 9.
Adolphs , R. , S. Baron-Cohen , and D. Tranel . 2002 . Impaired recognition
of social emotions following amygdala damage. Journal of Cognitive
Neuroscience 14.8 : 1264 74.
Amaral , D. G. 2003 . Te amygdala, social behavior, and danger detection.
Annals of the New York Academy of Sciences 1000 (Dec.): 337 47.
Freese , J. L. , and D. G. Amaral. 2006 . S ynaptic organization of projections
from the amygdala to visual cortical areas TE and V1 in the macaque
monkey . Journal of Comparative Neurology 496.5 : 655 67.
Isenberg , N. , D. Silbersweig , A. Engelien , S. Emmerich , K. Malavade , B.
Beattie , A. C. Leon , and E. Stern . 1999 . Linguistic threat activates the
human amygdala . Proceedings of the National Academy of Sciences
USA 96 . 18 : 10456 9.
Le Doux , J. 1996 . Te Emotional Brain: Te Mysterious Underpinnings of
Emotional Life . New York : Touchstone .
Phelps , E. A. 2006 . Emotion and cognition: Insights from studies of the
human amygdala . Annual Review of Psychology 57 : 27 53.
Phelps , E. A. , K. J. OConnor , J. C. Gatenby , J. C. Gore , C. Grillon , and M.
Davis . 2001 . Activation of the left amygdala to a cognitive representa-
tion of fear . Nature Neuroscience 4.4 : 437 41.
ANALOGY
Two situations are analogous if they share a common pattern of
relationships among their constituent elements, even though
the elements are dissimilar. Often one analog, the source , is more
familiar or better understood than the second analog, the target
(see source and target ). Typically, a target situation serves
as a retrieval cue for a potentially useful source analog. A map-
ping , or set of systematic correspondences aligning elements
of the source and target, is then established. On the basis of the
mapping, it is possible to derive new inferences about the target.
Analogy
98
b2, b3 (etc.). Tere is some relation R between a1, a2, and a3,
expressed as R(a1,a2,a3) , just as there is another such relation
S between b1, b2, and b3, expressed as S(b1,b2,b3) . For A and
B to be analogous, it is required that R and S be exemplifca-
tions of the same abstract structure X, as evidenced by a map-
ping between a1/a2/a3 and b1/b2/b3. Tis is what is meant by
saying that analogy (e.g., between A and B) is a structural sim-
ilarity, or a similarity between relations (e.g., R and S). It is not
a material similarity, or a similarity between things (e.g., a1
and b1). More and more abstract analogies are constituted by
similarities between similarities between relations between
things.
In its purely synchronic use (see synchrony and dia-
chrony ), analogy is understood to be the centripetal force that
holds the units of a structure together. To simplify an example
given by N. S. Trubetzkoy ([1939] 1958 , 6066), in a structure
containing just /p/, /b/, /t/, and /d/, the phoneme /p/ acquires
the distinctive features voiceless and bilabial by being contrasted
with, respectively, (voiced) /b/ and (dental) /t/. Te relation
between the pairs /p/ & /b/ and /t/ & /d/ is the same , and so is
the relation between /p/ & /t/ and /b/ & /d/, which means by
defnition that there is in both cases an analogy between the
two pairs. Tis type of analogy-based analysis applies to any well-
articulated structure, linguistic or nonlinguistic: A unit is what
the other units are not (as /p/ is neither /b/ nor /t/ nor, of course,
/d/); and this otherness is based on corresponding oppositions
(like voiceless vs. voiced and bilabial vs. dental).
Synchronic analogy may be characterized as analogy-as-
structure. Its counterpart is analogy-as-process,that is, discovery,
manipulation, or invention of structural similarity. Traditionally,
language acquisition was thought to be based on analogy: From
innumerable sentences heard and understood [the child] will
abstract some notion of their structure which is defnite enough
to guide him in framing sentences of his own (Jespersen [1924]
1965 , 19). After a period of neglect, this traditional view has
again become fashionable in some quarters (Pinker 1994 , 417;
Tomasello 2003 , 1639).
Only if analogy-as-process leaves a permanent trace that devi-
ates from the current norm is there reason to speak of language
change, the province of diachronic linguistics. Traditionally, the
term analogical change was restricted to morphology, or to cases
where irregularities brought about by sound change are elimi-
nated so as to achieve, or to approach, the goal of one meaning
one form . However, this same goal is involved in such large-scale
changes as have generally been ascribed to a need for harmony
or symmetry . In syntactic change, analogy consists in extending a
reanalyzed structure to new contexts (Anttila 1989 , 1024) .
Esa Itkonen
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anttila , Raimo . 1989 . Historical and Comparative Linguistics .
Amsterdam : Benjamins .
Jespersen , Otto . [1924] 1965 . Philosophy of Grammar . London: Allen and
Unwin .
Pinker , Stephen . 1994 . Te Language Instinct . New York : Morrow .
Tomasello , Michael . 2003 . Constructing a Language. A Usage-Based
Teory of Language-Acquisition . Cambridge: Harvard University
Press .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bunge , Silvia, C. Wendelken , and A. D. Wagner . 2005 . Analogical rea-
soning and prefrontal cortex: Evidence for separable retrieval and inte-
gration mechanisms . Cerebral Cortex 15 : 239 49.
Falkenhainer , Brian , K. D. Forbus , and D. Gentner . 1989 . Te Structure-
mapping engine: Algorithm and examples. Artifcial Intelligence
41 : 1 63.
Gentner , Dedre . 1983 . Structure-mapping: A theoretical framework for
analogy . Cognitive Science 7 : 155 70.
Gentner , Dedre , B. Falkenhainer , and J. Skorstad . 1988 . Viewing meta-
phor as analogy. In Analogical Reasoning: Perspectives of Artifcial
Intelligence, Cognitive Science, and Philosophy , ed. D. Helman , 1717.
Dordrecht, the Netherlands : Kluwer .
Gentner , Dedre , K. J. Holyoak , and B. N. Kokinov , eds . 2001 . Te Analogical
Mind: Perspectives from Cognitive Science . Cambridge, MA : MIT Press .
Tis book contains survey articles on topics in analogy.
Gentner , Dedre , and L.L. Namy . 2004 . Te role of comparison in chil-
drens early word learning. In Weaving a Lexicon , ed. D. Hall and S.
Waxman , 53368. Cambridge, MA : MIT Press .
Gentner , Dedre , and M. Rattermann . 1991 . Language and the career of
similarity. In Perspectives on Tought and Language: Interrelations
in Development , ed. S. Gelman and J. Byrnes , 22577.
Cambridge : Cambridge University Press .
Gick , Mary , and K. J. Holyoak . 1983 . S chema induction and analogical
transfer . Cognitive Psychology 15 : 1 38.
Goswami , Usha . 1992 . Analogical Reasoning in Children . Hillsdale,
NJ : Erlbaum .
Halford, Graeme . 199 3. Childrens Understanding: Te Development of
Mental Models . Hillsdale, NJ : Erlbaum .
Holyoak , Keith . 1982 . An analogical framework for literary interpreta-
tion . Poetics 11 : 105 26.
Holyoak , Keith , and P. Tagard . 1989 . Analogical mapping by constraint
satisfaction . Cognitive Science 13 : 295 355.
. 1995 . Mental Leaps: Analogy in Creative Tought . Cambridge,
MA : MIT Press . Tis book provides a broad introduction to the nature
and uses of analogy.
Hummel , John , and K. J. Holyoak . 2005 . Relational reasoning in a neu-
rally-plausible cognitive architecture: An overview of the LISA Project .
Current Directions in Cognitive Science 14 : 153 7.
Morrison , Robert , D. C. Krawczyk , K. J. Holyoak , J. E. Hummel , T. W.
Chow , B. L. Miller , and B. J. Knowlton . 2004 . A neurocomputational
model of analogical reasoning and its breakdown in frontotemporal
lobar degeneration . Journal of Cognitive Neuroscience 16 : 260 71.
Oden , David , R. K. R. Tompson , and D. Premack . 2001 . Can an ape rea-
son analogically? Comprehension and production of analogical prob-
lems by Sarah, a chimpanzee ( Pan Troglodytes ). In Gentner , Holyoak ,
and Kokinov 2001, 47197.
Richland , Lindsey , R. G. Morrison , and K. J. Holyoak . 2006 . Childrens
development of analogical reasoning: Insights from scene analogy
problems . Journal of Experimental Child Psychology 94 : 249 71.
Richland , Lindsey , O. Zur , and K. J. Holyoak . 2007 . Cognitive supports
for analogy in the mathematics classroom . Science 316 : 1128 9.
Spearman , Charles . 1946 . Teory of a general factor . British Journal of
Psychology 36 : 117 31.
Vallauri , Edoardo . 2004 . Te relation between mind and language: Te
innateness hypothesis and the poverty of the stimulus . Linguistic
Review 21 : 345 87.
ANALOGY: SYNCHRONIC AND DIACHRONIC
Analogy involves two (or more) systems , A and B, which are con-
stituted by their respective parts , that is, a1, a2, a3 (etc.) and b1,
Analogy: Synchronic and Diachronic
99
expressing it. Tus, our current linguistic conventions do not
make it true. Tird, our frst example certainly does say some-
thing about the world. Indeed, it says something about every
object in the universe. If it is an uncle, then either it is married
or it has a sibling.
Some philosophers have called the sort of analyticity dis-
cussed so far, where things are said to be true in virtue of mean-
ing, metaphysical analyticity . Tis is distinguished from epistemic
analyticity. A statement is epistemically analytic when under-
standing it suf ces for being justifed in believing it (Boghossian
1996 ). While these considerations make trouble for metaphys-
ical analyticity, they allegedly leave its epistemic counterpart
untouched.
Te purpose of introducing epistemic analyticity is similar to
that of its older ancestor. Te hope is that mathematical, logical,
and conceptual truths can be designated as a priori without pos-
tulating a special faculty of reason or intuition . Tis is done by
building certain kinds of knowledge in as preconditions for pos-
sessing or understanding concepts. If part of what it is to under-
stand the word uncle is to be disposed to accept that all uncles
are either married or have a sibling, then it could be argued
that once we understand that sentence, we know that it is true.
No experience (beyond what is required for understanding) is
necessary.
Te best candidates for epistemically analytic truths are sim-
ple truths of logic. But even the most obvious logical truths are
not immune to challenge. For example, a few philosophers and
logicians have claimed that some statements can be both true
and false (Preist 1987 ) and that modus ponens is invalid (McGee
1985 ). Yet these sophisticated theoreticians certainly understand
the meanings of their own words. Terefore, acceptance of some
specifc truth of logic is not necessary for understanding any log-
ical concept. And since we might someday have good reason to
reject any particular truth of logic while continuing to understand
our logical concepts, understanding some logical concept is not
suf cient for being justifed in believing any particular truth of
logic. And if logic is not epistemically analytic, nothing is .
Tese considerations make the existence of analyticity dubi-
ous. But there still appears to be a deep diference between the
two examples. If there is really a diference, it is not one of true in
virtue of meaning versus true in virtue of reality, but one of nec-
essary and a priori versus contingent and empirical .
Michael Veber
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
BonJour , L. 1998 . In Defense of Pure Reason . Cambridge : Cambridge
University Press .
Boghossian , P. 1996 . Analyticity reconsidered . Nous 30 . 3 : 360 91.
Kripke , S. 1980 . Naming and Necessity . Cambridge : Harvard University
Press .
McGee , V. 1985 . A counterexample to modus ponens . Journal of
Philosophy 82 : 462 71.
Priest , G. 1987 . In Contradiction . Boston: Kluwer.
Quine , W. V. O. 1953 . Two dogmas of empricism . In From a Logical
Point of View , 2046. Cambridge : Harvard University Press .
Veber , M. 2007 . Not too proud to beg (the question): Why inferential-
ism cannot account for the a priori . Grazer Philosophische Studien
73 : 113 31. A critique of Boghossian 1996 and similar views.
Trubetzkoy , N. S. [1939] 1958 . Grundzge der Phonologie . Gttingen:
Vandenhoeck and Ruprecht.
ANALYTICITY
Analyticity is a property that a statement has when its truth is
in some special way determined by its meaning. Many believe
that no such property exists. Led by W. V. O. Quine ( 1953 ), phi-
losophers complain that no one has been able to defne the con-
cept of analyticity in a way that is precise and that also fulflls the
purposes to which it is typically put .
To some people, it seems that all uncles are either mar-
ried or have a sibling is true just because of the meaning of its
constituent words , most prominently because of the meaning
of uncle . But all uncles are less than eight feet tall is true not
because of meanings but because of how the world has turned
out to be. Te frst sort of statement is said to be analytic, the
second synthetic.
Tis distinction has far-reaching interest and application.
Empiricists have always had dif culty accounting for the seem-
ingly obvious fact that the truths of logic and mathematics are
both necessary (i.e., they could not be false) and a priori (i.e., they
are known independently of sensory experience). For most of the
twentieth century, it was agreed that analyticity could explain
this obvious fact away as a merely linguistic phenomenon .
Te idea that all necessity could be explained away by ana-
lyticity fell out of fashion when S. Kripke ( 1980 ) convinced most
philosophers that some necessary truths are neither analytic nor
a priori (e.g., water = H
2
O). But as L. BonJour ( 1998 , 28) points
out, many still think that the unusual modal and epistemic sta-
tus of logic and mathematics is due to a special relation between
truth and meaning.
However we apply the concept, trouble for analyticity begins
when we remind ourselves that every truth depends, to some
extent, on the meanings of its constituent terms. If the word
uncle had meant oak tree, then both previous examples would
be false. In response, it is said that an analytic truth is one whose
truth depends solely on the meanings of its terms. Our linguistic
conventions alone make the frst example true, and the second is
true party because of meaning and partly because of the way the
world is. But how can linguistic convention alone make some-
thing true?
We can distinguish the sentence all uncles are either mar-
ried or have a sibling from the proposition that this sentence
now expresses. Meaning or linguistic convention alone makes
this sentence true in the following way. Given that this sentence
expresses the proposition that it does (i.e., given our current lin-
guistic conventions), it is true. Tis cannot be said of our other
example. Tat this sentence means what it does is not suf cient
to determine its truth or falsity (see also sentence meaning ).
Te world plays a part .
Tere are three serious problems. First, if this is what it is for
a sentence to be true solely in virtue of its meaning, then it is just
another way of saying that it expresses a necessary truth, and that
tells us nothing about how we know it. Tus, appeal to analyticity
cannot explain the necessity and a priority of logic, mathemat-
ics, or anything else. Second, the proposition now expressed
by our frst example would be true no matter how we ended up
Analyticity
100
restrictive conditions, some of these expressions may yet allow a
free logophoric interpretation .
It is an important task to arrive at a detailed understanding
of the ways in which languages encode interpretive relations
between their expressions and of the division of labor between
the components of the language system involved .
Eric Reuland
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , Noam . 1981 . Lectures on Government and Binding . Dordrecht,
the Netherlands : Foris.
Heim , Irene . 1982 . Te semantics of defnite and indefnite noun phrases .
Ph.D. diss., University of Massachusetts at Amherst.
Reinhart , Tanya . 1983 . Anaphora and Semantic Interpretation .
London : Croom Helm .
Reuland, Eric. 2010. Anaphora and Language Design. Cambridge: MIT
Press.
Safr , Ken . 2004 . Te Syntax of Anaphora . Oxford : Oxford University
Press.
ANIMACY
Languages often treat animate and inanimate nouns diferently.
Animacy can afect many aspects of grammar, including word
order , and verbal agreement . For example, in Navajo, the
more animate noun must come frst in the sentence (Hale 1973 ),
and in some Bantu languages, a more animate object must come
before a less animate object. Verbs are more likely to agree with
more animate nouns (Comrie 1989 ). Animacy can also afect the
choice of case , preposition, verb form, determiner (article), or
possessive marker (Comrie 1989 ).
What counts as animate difers cross-linguistically. Te gram-
matical category of animates may include certain objectively
inanimate things, such as fre, lightning, or wind. Languages may
make additional distinctions among pronouns, proper nouns,
and common nouns, or between defnites and indefnites, and
these are sometimes viewed as part of an animacy hierarchy by
linguists (Comrie 1989 ). person and number distinctions may
also be included in an animacy hierarchy in this broader sense .
For example, according to Michael Silverstein ( 1976 ), subjects
with features at the less animate end of the Animacy Hierarchy in
(1) are more likely to be marked with morphological case, while
the reverse holds of objects.
(1) Animacy Hierarchy
1pl > 1sing > 2pl > 2sing > 3human.pl > 3human.sing >
3anim.pl > 3anim.sing > 3inan.pl > 3inan.sing
Dyirbal exhibits this pattern in that only third person sub-
jects have morphological case, whereas all human objects do .
Silverstein ( 1976 ) postulates that the function of such diferen-
tial case marking is to fag less animate subjects and more ani-
mate objects to avoid ambiguity. It is interesting to note that
the patterns of such diferential animacy marking are far more
complex and diverse cross-linguistically for objects than for sub-
jects (Aissen 2003 ). Tis may be traced to a relation between ani-
macy and object shift, which produces an associated change in
case or verbal agreement (Woolford 2000 , 2007 ). Te less diverse
animacy efects on subject case, which do not afect agreement
ANAPHORA
Languages have expressions whose interpretation may involve
an entity that has been mentioned before: Subsequent reference to
an entity already introduced in discourse approximates a general
defnition of the notion of anaphora (Safr 2004 , 4). Tis works
well for core cases of nominal anaphora as in (1) (Heim 1982 ):
(1) a. Tis soldier has a gun. Will he shoot?
b. Every soldier has a gun. ??Will he shoot?
He in (1a) can be interpreted as the same individual as this soldier .
In (1b) every soldier is quantifcational, hence does not denote an
entity he can refer to, which makes anaphora impossible. Possible
discourse antecedents are as diverse as soldiers, water, beauty,
headaches, dissatisfaction, and so on. In addition to nominal
expressions, sentences , verb phrases, prepositional phrases,
adjective phrases, and tenses also admit anaphoric relations .
Tus, the notion discourse entity must be broad enough to cap-
ture all these cases of anaphora yet restrictive enough to separate
them from quantifcational cases such as every soldier.
Te notion of anaphora is closely related to the notion of
interpretative dependency . For instance, in (2), he can depend for
its interpretation on every soldier , and here, too, it is said that he
is anaphorically related to every soldier .
2. Every soldier who has a gun says he will shoot .
However, (1) versus (2) shows that two diferent modes of inter-
pretation must be distinguished: i) directly assigning two (or
more) expressions the same discourse entity from the inter-
pretation domain (I
D
) as a value: co-reference as in (1a), and ii)
interpreting one of the expressions in terms of the other by gram-
matical means: binding (Reinhart 1983 ). Tis contrast is repre-
sented in (3).
Coreference in (3a) is restricted in terms of conditions on dis-
course entities, binding in (3b) in terms of grammatical confg-
uration. Expr
1
can only bind expr
2
if it c-commands the latter
(Reinhart 1983 ). Tis condition is met in (2), but not in (1b),
hence the contrast .
Virtually all languages have words and expressions that are ref-
erentially defective they cannot be used deictically (see deixis ).
In much of the linguistic literature, these are called anaphors , as
they appear specialized for anaphoric use. Examples vary from
English himself , Dutch zich(zelf) , Icelandic sig , Russian sebja ,
Chinese (ta) ziji , to Georgian tav tavis , and so on. Such expres-
sions cannot be assigned a discourse value directly. Rather, they
must be bound, often in a local domain approximately the
domain of the nearest subject but subject to variation in terms
of specifc anaphor type and language. Furthermore, under
Anaphora Animacy
101
within this feld (e.g., Bradbury and Vehrenkamp 1998 ) reveals
a great deal about the communicative behavior of many species
and its origins, but within essentially the same picture of what
constitutes communication in (nonhuman) animals, confned to
unitary signals holistically transmitted and interpreted. Little if
any of what we have come to know about human linguistic com-
munication fnds a place here.
Biologists have not, in general, paid much attention to the
specifcs of linguistic research (though their attention has been
caught by the notion that human language is importantly based
in human biology) and are often not as sophisticated as one
might wish about the complexity of natural language. But the
consequences may not be as serious as linguists are inclined to
think. In fact, the communicative behavior of nonhumans in gen-
eral is essentially encompassed within the simple signal-passing
model. Te complexities of structure displayed by human lan-
guage are apparently quite unique to our species and may not
be directly relevant to the analysis of animal communication
elsewhere .
What (Other) Animals Do
Communication in the sense of emission and reception of infor-
mative signals is found in animals as simple as bacteria ( quorum
sensing ). Most familiar, perhaps, are visual displays of various
sorts that indicate aggression, submission, invitations to mate,
and so on. In some instances, these may involve quite complex
sequences of gestures, reciprocal interactions, and the like, as in
the case of the nesting and mating behavior of many birds. In
others, a simple facial expression, posture, or manner of walking
may provide the signal from which others can derive information
about the animals intentions and attitudes.
Tese diferences of internal structure are, of course, crucial
for the correct expression and interpretation of a particular sig-
nal, but they play little or no role in determining its meaning. Tat
is, the individual components of the signal do not in themselves
correspond to parts of its meaning, in the sense that varying one
subpart results in a corresponding variation in what is signaled.
Animal signals, however complex in form (and however elaborate
the message conveyed), are unitary wholes. An entire courtship
dance, perhaps extending over several minutes or even longer,
conveys the sense I am interested in mating with you, providing
a nesting place, and care for our ofspring. No part of the dance
corresponds exactly to the providing care part of the message;
the message cannot be minimally altered to convey I am inter-
ested in mating but not in providing care for our ofspring, I was
interested in mating (but am no longer), and so on. Variations in
intensity of expression can convey (continuous) variations in the
intensity of the message (e.g., urgency of aggressive intent), but
that is essentially the only way messages can be modulated .
Te most widely discussed apparent exception to this gen-
eralization is the dance language of some species of honeybees.
Te bees dance conveys information about a) the direction, b)
the distance, and c) the quality of a food source (or potential
hive site), all on quasi-continuous scales and each in terms of
a distinct dimension of the dance. Although the content of the
message here can be decomposed, and each part associated
with a distinct component of the form of the signal, there is no
element of free combination. Every dance necessarily conveys
(Comrie 1991 ), may be purely morphological, markedness efects
(Woolford 2007 ) .
Ellen Woolford
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aissen , Judith . 2003 . Diferential object marking: Iconicity vs. economy .
Natural Language and Linguistic Teory 21 : 435 83.
Comrie , Bernard . 1989 . Language Universals and Linguistic Typology .
Oxford: Blackwell.
. 1991 . Form and function in identifying cases. In Paradigms , ed.
F. Plank, 4155. Berlin: Mouton de Gruyter.
Hale , Kenneth L. 1973 . A note on subject-object inversion in Navajo. In
Issues in Linguistics: Papers in Honor of Henry and Rene Kahane , ed. B.
B. Kachru et al., 300309. Urbana : University of Illinois Press .
Silverstein , Michael . 1976 . Hierarchy of features and ergativity. In
Grammatical Categories in Australian languages , ed. R. M. Dixon ,
11271. Canberra : Australian Institute of Aboriginal Studies .
Woolford , Ellen . 2000 . Object agreement in Palauan. In Formal Problems
in Austronesian Morphology and Syntax , ed. I. Paul , V. Phillips , and L.
Travis , 21545. Dordrecht, the Netherlands : Kluwer .
. 2007 . Diferential subject marking at argument structure, syn-
tax, and PF. In Diferential Subject Marking , ed. H. de Hoop and P. de
Swart , 1740 . Dordrecht : Springer .
ANIMAL COMMUNICATION AND HUMAN LANGUAGE
An understanding of the communicative capacities of other ani-
mals is important on its face both for an appreciation of the place
of human language in a broader context and as a prerequisite to
discussion of the evolution of language (see, for example, evo-
lutionary psychology ). On closer examination, however,
the diferences between human language and the systems of
other animals appear so profound as to make both projects more
problematic than they appear at frst.
In the 1950s and 1960s, ethologists like Konrad Lorenz and
Niko Tinbergen revolutionized behavioral biologists views of
the cognitive capacities of animals, but consideration of ani-
mal communication focused on the properties of quite simple
systems. A prime example of communication in the texts of
the time was the stickleback. A crucial component of the mat-
ing behavior of this common fsh is the pronounced red color-
ation of the males underbelly when he is in mating condition,
which furnishes a signal to the female that she should follow him
to his preconstructed nest, where her eggs will be fertilized. On
this model, communication was viewed as behavioral or other
signals emitted by one organism, from which another organism
(typically, though not always, a conspecifc) derives some infor-
mation. Te biological analysis of communication thus came to
be the study of the ways in which such simple signals arise in the
behavioral repertoire of animals and come to play the roles they
do for others who perceive them. Tose discussions make little if
any contact with the analysis of human language.
In the intervening half century, we have come to know vastly
more about the nature and architecture of the human lan-
guage faculty and to have good reason to think that much of it
is grounded in human biology. One might expect, therefore, to
fnd these concerns refected in the behavioral biology literature.
A comprehensive modern textbook on animal communication
Animal Communication and Human Language
102
this regard. In general, their song is learned on the basis of early
exposure to appropriate models, from which they in turn com-
pose their own songs. It is interesting to note there appear to be
quite close homologies in the neurophysiology of vocal learn-
ing and perhaps even in its underlying genetic basis between
birds and humans, although what is learned in birds is a uni-
tary, holistic signal like those in other nonhuman communica-
tion systems, rather than individual lexical items subject to free
recombination to produce diferent meanings .
Tere is much variation across bird species, but a clear gen-
eralization emerges: For each, there is a specifc range of song
structures that individuals of that species can learn. Experience
plays a role in providing the models on which adult song is based,
but (with the exception of a few very general mimics, such as the
lyrebird) this role is quite narrowly constrained by the song-
learning system of the individual species .
What Humans Do, and How It Is Different
Like the systems of communication of other animals, human
language is deeply embedded in human biology. Unlike others,
however, it provides an unbounded range of distinct, discrete
messages. Human language is acquired at a specifc point in
development from within a limited range of possibilities, simi-
lar to the acquisition of song in birds. Unlike the communicative
signals of other species, human language is under voluntary con-
trol, with its underlying neurobiology concentrated in cortical
structures, as opposed to the subcortical control characteristic of
those other species that have been studied in this regard.
Human language is structurally a discrete combinatorial sys-
tem, in which elements from a limited set combine in a recursive,
hierarchical fashion to make an unlimited number of potentially
novel messages (see recursion, iteration, and metarep-
resentation ). Te combinatorial structure of language is gov-
erned by two quite independent systems: A small inventory of
individually meaningless sounds combine to make meaningful
words , on the one hand ( phonology ), while these words are
combined by a quite diferent system to make phrases, clauses,
and sentences (see syntax ). Tese properties (discrete com-
bination, recursive hierarchical organization, and duality of pat-
terning ) are not simply idiosyncratic ornaments that could in
principle be omitted without afecting the overall communicative
capacity of the system. Rather, they are what make large vocabu-
laries practical and unbounded free expression possible. Contrast
the unlimited range of potentially novel utterances that any (nor-
mal) speaker of a language can produce, and another speaker of
the same language comprehend, with the strictly limited range of
meaningful signals available to other organisms. No other form of
communication found in nature has these properties. Although
song in some species of birds does display a limited amount of
phonological combinatoriality, there is no analog even here to
meaningful syntax. Human language, and especially its syntactic
organization, is quite unique in the animal world .
Furthermore, eforts to teach systems with these essential
properties to other animals have not succeeded. Despite wide-
spread claims to the contrary in the popular literature, there is
no evidence that any nonhuman animal is capable of acquiring
and using such a system. Tis should not be seen as particularly
surprising. If language is indeed embedded in human biology,
exactly these three things, and it is only the relative value on each
dimension that is variable. As such, the degree of freedom avail-
able to construct new messages is not interestingly diferent from
that involved in conveying diferent degrees of fear or aggression
by varying degrees of piloerection .
Visual displays do not at all exhaust the modalities in which
animal communication takes place, of course. Auditory signals
are important to many species, including such classics of the ani-
mal communication literature as frog croaks and the calls and
songs of birds (see birdsong and human language ). In
some species, portions of the auditory spectrum that are inac-
cessible to humans are involved, as in the ultrasound communi-
cation of bats, some rodents, and dolphins, and the infrasound
signals of elephants. Chemical or olfactory communication is
central to the lives of many animals, including moths, mice, and
lemurs, as well our pet cats and dogs. More exotic possibilities
include the modulation of electric felds generated (and per-
ceived) by certain species of fsh.
In some of these systems, the internal structure of the signal
may be quite complex, as in the songs of many oscine songbirds,
but the general point still holds: However elaborate its internal
form, the signal has a unitary and holistic relation to the message
it conveys. In no case is it possible to construct novel messages
freely by substitutions or other ways of varying aspects of the sig-
nals form.
In most animals, the relation of communicative behavior to
the basic biology of the species is very direct. Perceptual systems
are often quite precisely attuned to signals produced by con-
specifcs. Tus, the frogs auditory system involves two separate
structures (the amphibian papilla and the basilar papilla) that are
sensitive to acoustic signals, typically at distinct frequencies. Te
frequencies to which they are most sensitive vary across species
but are generally closely related to two regions of prominence
in the acoustic structure of that species calls. Mice (and many
other mammals) have two distinct olfactory organs, projecting
to quite distinct parts of the mouse brain. Te olfactory epithe-
lium is responsive to a wide array of smells, but the vomeronasal
organ is sensitive primarily to the pheremones that play a major
role in communication and social organization. In this case, as in
many, many others, the perceptual system is matched to produc-
tion in ways that optimize the organisms sensitivity to signals
that play a crucial ecological role in the life of the animal.
Te essential connection between a species system of com-
munication and its biology is also manifested in the fact that
nearly all such systems are innately specifed. Tat is, the ability
to produce and interpret relevant signals emerges in the individ-
ual without any necessary role of experience. Animal commu-
nication is not learned (or taught) but, rather, develops (in the
absence of specifc pathology, such as deafness) as part of the
normal course of maturation. Animals raised under conditions
in which they are deprived of exposure to normal conspecifc
behavior will nonetheless communicate in the fashion normal to
their species when given a chance .
Exceptions to this generalization are extremely rare, apart
from human language. Vocal learning, in particular, has been
demonstrated only to a limited extent in cetaceans and some bats
and, more extensively, in 3 of the 27 orders of birds. Te study of
birds, especially oscine songbirds, is particularly instructive in
Animal Communication and Human Language
103
emission tomography (PET), and functional MRI (fMRI) have
helped defne functions of core speech and language areas .
Aphasia Syndromes
Aphasia has traditionally been categorized into seven subtypes,
including Brocas, Wernickes, global, anomic, conduction,
transcortical sensory, and transcortical motor. Tese aphasia
variants are characterized by diferent patterns of speech fuency,
auditory comprehension, repetition, and naming. Although
patients may be classifed as having one type of aphasia in the
early period after a brain injury, this classifcation may change as
language problems resolve with time and treatment.
Brocas aphasia is characterized by a constellation of symp-
toms, including slow, halting speech with impaired grammar;
disturbed auditory comprehension for grammatically complex
phrases and sentences; and poor repetition. Word-fnding prob-
lems and dif culty with reading and writing are common. Motor
speech disorders, such as apraxia of speech, a disorder of articu-
latory planning or coordination, and dysarthria, an impairment
in muscle strength, tone, or coordination, very often co-occur.
Patients with Brocas aphasia often talk in a series of nouns and
verbs, as is the case in the following sample from a patient describ-
ing the picnic scene from the Western Aphasia Battery (WAB):
I know it tree house car man with uh woman kid
over here with fag i can know it nice sun shiny day
[unintelligible word]
Patients with Brocas aphasia can participate relatively well
in everyday conversation by using single words or phrases,
often combined with meaningful gestures and facial expres-
sions. Some patients use writing and drawing to compensate for
restricted verbal output. As is true with all aphasia syndromes,
there is a wide range of symptom severity. Tose with severe
Brocas aphasia may present with such profound verbal defcits
that their speech is limited to a single recurrent utterance (e.g.,
yeah, yeah). In these patients, comprehension is usually pre-
served for short, simple phrases, but is signifcantly impaired for
more complex information.
Classic aphasia models assume that lesions to brocas area
(See Figure 1) result in Brocas aphasia, but research has indi-
cated that this is not always the case. Reports as early as 1870
documented cases that did not support this linear relationship
(e.g., Bateman 1870 ; Marie 1906 ; Moutier 1908 ; Dronkers et al.
2007). Modern research has found that chronic Brocas aphasia
typically results from large lesions that encompass left frontal
brain regions, underlying white matter, the insula, and the ante-
rior parietal lobe. Lesions restricted to Brocas area tend to
cause transient mutism that spontaneously resolves within days
or weeks (Mohr 1976 ). In some cases, Brocas aphasia can occur
without damage to Brocas area (e.g., Basso et al. 1985 ; Mazzocchi
and Vignolo 1979 ) .
Patients with Wernickes aphasia present a reverse pattern of
symptoms when compared to those with Brocas aphasia: While
speech is fuent, comprehension is impaired. Patients speak in
a normal or rapid rate. However, they often use meaningless
words, jargon, or semantic paraphasias (e.g., using a related
word, bike for a target word, car ). Reading and writing may
be similarly disrupted. Te following exemplifes the speech
there is no reason to expect it to be accessible to organisms with
a diferent biological endowment, anymore than humans are
capable of acquiring, say, the echolocation capacities of bats, a
system that is equally grounded in the specifc biology of those
animals .
Conclusion
Human language is often considered as simply one more instan-
tiation of the general class of animal communication systems.
Indeed, like others it appears to be highly species specifc.
Although relevant experience is required to develop the system
of any particular language, the overall class of languages acces-
sible to the human learner is apparently highly constrained, and
the process of language learning is more like genetically gov-
erned maturation than like learning in general. Te structural
characteristics of human language are quite diferent from those
of other communication systems, and it is the freedom of expres-
sion subserved by those distinctive properties that gives language
the role it has in human life. (See also primate vocalizations
and grooming, gossip, and language ) .
Stephen R. Anderson
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Stephen R. 2004 . Doctor Dolittles Delusion: Animals and the
Uniqueness of Human Language . New Haven, CT : Yale University
Press .
Bradbury , J. W. , and Sandra Vehrenkamp . 1998 . Principles of Animal
Communication . Sunderland, MA : Sinauer .
Hauser , Marc D. 1996 . Te Evolution of Communication . Cambridge,
MA : MIT Press .
APHASIA
Aphasia is a language impairment caused by brain injury that
afects speech content, auditory comprehension, reading, and
writing to varying degrees. Mild aphasia may result in occasional
word-fnding problems, while more severe forms can cause pro-
found defcits in all language domains. Aphasia difers from
developmental disorders in that it occurs after a brain injury to a
person with otherwise normal language skills .
Typically, aphasia results from damage to the left hemi-
sphere of the brain due to stroke, traumatic brain injury, tumor,
or degenerative neurological disease. Nearly all right-handed
individuals and most left-handers are thought to have core lin-
guistic functions semantics, phonology, syntax, and
morphology lateralized to the left hemisphere, while other
aspects of language , specifcally prosody and pragmatics, are
associated with the right hemisphere . Approximately one
million people in the United States are af icted with aphasia,
which is a prevalence rate similar to that of Parkinsons disease.
Roughly 80,000 more acquire aphasia annually .
Recent advances in the study of language have provided greater
insight into aphasia syndromes. Modern neuroimaging tech-
nology has helped refne classic models of language localization,
as well as our understanding of aphasia treatment and recovery.
In particular, brain-imaging techniques such as magnetic reso-
nance imaging (MRI), computerized tomography (CT), positron
Aphasia
104
Initial reports that conduction aphasia arose from lesions
to the arcuate fasciculus (the white matter tract connecting
Wernickes and Brocas areas; see Figure 1) have been refned
over the years. Modern studies have shown that conduction
aphasia results most often from lesions to the posterior supe-
rior temporal gyrus (Dronkers et al. 1998), the auditory cortex
(Damasio and Damasio 1980 ), or periventricular white matter
underlying the supramarginal gyrus (Sakurai et al. 1998 ) .
Global aphasia, the most severe syndrome, is characterized
by profound impairments in all language modalities. Speech,
auditory comprehension, naming, repetition, reading, and writ-
ing are all afected, leaving the patient with very little functional
communication. Speech may be limited to single stereotyped
or automatic words and phrases (e.g., yes , no , I dont know ).
Auditory comprehension may be impaired for even simple yes/
no questions. Such a severe loss of language typically results
from a large cortical lesion, encompassing the frontal, temporal,
and parietal lobes. Patients often rely on preserved nonverbal
skills to aid in communication (e.g., the recognition of pictures
and gestures to support auditory comprehension and the ability
to draw or gesture to aid in expression) .
Anomic aphasia, the mildest of the syndromes, results in
word-fnding defcits (anomia), while other language skills are
typically well preserved. When attempting to fnd a target word,
patients with anomic aphasia may describe its function or use a
synonym. Speech may be slow and halting, due to anomia, but
grammar is unafected. Anomic aphasia can result from lesions
anywhere within the perisylvian region .
Te transcortical aphasias are rare and characterized by a
preserved ability to repeat, despite impairments in other lan-
guage domains. Transcortical motor aphasia (TCMA) is simi-
lar to Brocas aphasia, in that patients present with nonfuent
speech and relatively intact comprehension, but repetition skills
are markedly well preserved. Lesions typically spare core lan-
guage areas, are smaller than those that cause Brocas aphasia,
and are restricted to anterior and superior frontal lobe regions.
Although patients may be mute initially, their symptoms tend
to resolve quickly, resulting in anomic aphasia. Patients with
transcortical sensory aphasia (TCSA) present much like patients
with Wernickes aphasia, with empty, fuent speech and poor
comprehension, but they too retain a striking ability to repeat.
Lesions typically involve portions of the posterior temporal
content of a patient with Wernickes aphasia describing the
WAB picnic scene:
And the man and hers and Ill say I dont think shes working.
Teyre not doing the thing. Ten the ladder then the tree
and the /let/ [points to kite] and lady here [points to boy]
have to clean that.
In contrast to patients with Brocas aphasia, those with
Wernickes aphasia may understand very little in conversation
because of their impaired comprehension of single words. In
addition, successful communication is made challenging by ver-
bal output that is empty, coupled with an inability to monitor
speech content. Using visual information to compensate for com-
prehension defcits is often benefcial (e.g., providing pictures,
drawing, or writing key words during conversational exchanges).
Persisting cases of Wernickes aphasia are not caused by injury
to wernickes area alone but, rather, by much larger lesions
afecting most of the middle temporal gyrus and underlying white
matter (Dronkers, Redfern, and Ludy 1995 ; see Figure 1). Such
damage amounts to a poorer prognosis for recovery. Patients
with lesions confned to Wernickes area tend to have symptoms
of Wernickes aphasia that resolve, resulting in milder forms of
aphasia, most often conduction aphasia or anomic aphasia, if
the lesion spares the middle temporal gyrus .
Conduction aphasia is a fuent aphasia characterized by an
inability to repeat. Auditory comprehension is relatively pre-
served, and patients use speech that is largely understandable
but may be rife with phonemic paraphasias (substituting sounds
in words, e.g., netter for letter ). While high-frequency words
and short phrases may be repeated accurately (e.g., the tele-
phone is ringing), low-frequency items are more dif cult (e.g.,
frst British feld artillery). Patients may retain the meaning
of such phrases, owing to their preserved comprehension, but
the phonological trace is disrupted, thereby disturbing verba-
tim repetition. Te following typifes the speech of a patient with
conduction aphasia, again describing the WAB picnic scene:
Well theres a house near a clearing, evidently its on the water.
Further, theres a stick with a banner in the foreground
[referring to the fag]. I dont know what thats called a pier
a tier? Teres a bucket and a /kovel/. It looks like theres
someone playing in the water.
Figure 1. Several of the key brain regions affected
in aphasia. Areas depicted as typical lesions are
derived from patient data obtained at the Center
for Aphasia and Related Disorders.
Aphasia
105
Dronkers , N. F. , O. Plaisant , M. T. Iba-Zizen , and E. A. Cabanis . 2007 .
Paul Brocas historic cases: High resolution MR imaging of the brains
of Leborgne and Lelong . Brain 130 : 1432 41.
Dronkers N. F. , B. B. Redfern , C. Ludy , and J. Baldo . 1998 . Brain regions
associated with conduction aphasia and echoic rehearsal . Journal of
the International Neuropsychological Society 4 : 23 4.
Gorno-Tempini , M. L. , N. F. Dronkers , K. P. Rankin, et al. 2004 . Cognition
and anatomy in three variants of primary progressive aphasia . Annals
of Neurology 55 : 335 46.
Holland , A. L. , D. S. Fromm , F. DeRuyter , M. Stein . 1996 . Treatment ef -
cacy . Journal of Speech and Hearing Research 39.5 : S27 36.
Marie P. 1906 . Revision de la question de laphasie: La troisieme circon-
volution frontale gauche ne joue aucun role special dans la fonction du
langage . Semaine Medicale 26 : 241 7.
Mazzocchi , F. , and L. A. Vignolo . 1979 . Localization of lesions in
aphasia: Clinical CT-scan correlations in stroke patients . Cortex
15 : 627 54.
Mesulam , M. M. 1982 . S lowly progressive aphasia without generalized
dementia . Annals of Neurology 11 : 592 8.
Mohr , J. P. 1976 . Brocas area and Brocas aphasia. In Studies in
Neurolinguistics. Vol. 1. Ed. H. Whitaker and H. Whitaker, 20133. New
York : Academic Press .
Moutier , F. 1908 . LAphasie de Broca . Paris : Steinheil .
Sakurai , Y. , S. Takeuchi , E. Kojima , et. al. 1998 . Mechanism of short-
term memory and repetition in conduction aphasia and related cogni-
tive disorders: a neuropsychological, audiological and neuroimaging
study . Journal of Neurological Sciences 154.2 : 182 93.
Snowden , J. S. , D. Neary , D. M. Mann , et al . 1992. Progressive language
disorder due to lobar atrophy . Annals of Neurology 31 : 174 83.
AREAL DISTINCTNESS AND LITERATURE
Tere are two criteria for determining whether a linguistic prop-
erty is a universal. First, it must occur across languages with a
frequency greater than chance. Second, the presence of the prop-
erty in some of these languages should not have been caused by
its presence in other languages. In linguistics, the causal crite-
rion is often operationally specifed into two subcriteria genetic
and areal distinctness, which is to say, distinctness in origin and
in cross-language interaction.
Researchers in literary universals also adopt the pre-
ceding criteria. However, literature is diferent from language
in being more readily open to infuence. Specifcally, the opera-
tional criterion of areal distinctness becomes much more dif cult
to satisfy in the case of literature. Even a single work, transported
across continents, may produce signifcant changes in the recipi-
ent literature .
Tere are three ways of responding to this problem. Te frst is
to focus on literary works produced before the period of extensive
global interaction. Research of this sort must form the primary basis
for any serious study of literary universals. Moreover, such research
indicates that there are some signifcant universals, for example the
narrative universals of heroic, romantic, and sacrifcial tragi-
comedy. However, this approach to areal distinctness cannot be as
rigorous as one might like. Global interaction extends back through
the formation of all the major literary traditions .
Te second response involves a more nuanced approach for
isolating infuence from a source tradition to a recipient tradi-
tion. Here, we may distinguish between self-conscious and
implicit learning. Self-conscious learning can occur with a single
and parietal regions, but tend to be much smaller than those of
Wernickes aphasia. Acute symptoms usually resolve to produce
an anomic aphasia .
While aphasia most often occurs suddenly, as the result
of injury, a degenerative form of aphasia was frst described
over a century ago by Arnold Pick , a Czech neurologist, and
later expanded upon by Marsel Mesulam in a landmark paper
in which he described six patients who presented with lan-
guage defcits, in the absence of other behavioral abnormalities
(Mesulam 1982). Speech or language defcits remained the only
impairment for the frst two years in these patients, but as the
disease progressed, more generalized dementia emerged. Tis
progressive disorder was distinct from other dementias, such as
Alzheimers disease, because language problems, rather than
memory complaints, were the most salient symptoms.
Since then, numerous cases of what is now termed pri-
mary progressive aphasia (PPA) have been described, in which
patients present with both fuent and nonfuent variants of the
disorder (Snowden, et al. 1992; Gorno-Tempini et al. 2004 ).
Neuroimaging typically shows left perisylvian atrophy, encom-
passing frontal regions in progressive nonfuent aphasia and
anterior temporal and temporo-parietal regions in the more fu-
ent semantic dementia and logopenic variants. Tere are many
underlying pathologies that cause the clinical syndrome of PPA,
including Picks disease , progressive supranuclear palsy, corti-
cobasal degeneration, dementia lacking distinctive pathology,
and Alzheimers disease .
Treatment for Aphasia
Critical reviews of aphasia treatment studies (e.g., Bhogal,
Teasell, and Speechley 2003 ; Holland et al. 1996 ) have shown
that treatment can be efective in improving language skills past
the point that might be expected from spontaneous recovery
alone. Although it remains dif cult to predict the treatment that
will result in the greatest amount of change for an individual,
there are many options from which to choose.
Patients with aphasia are typically referred to speech lan-
guage pathologists for diagnostic testing aimed at developing
treatment goals. Terapy may focus on improving impaired skills
or developing compensatory strategies to overcome obstacles to
successful communication. Patient-specifc factors (e.g., aphasia
severity, cognitive ability, general health, and motivation) also
infuence treatment decisions. Research is inconclusive, how-
ever, as to the prognostic weight that these variables contribute
to recovery and treatment planning for an individual .
Nina F. Dronkers , Jennifer Ogar
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Basso , A. , A. R. Lecours , S. Moraschini , and M. Vanier . 1985 . Anatomo-
clinical correlations of the aphasias as defned through computerized
tomography: On exceptions . Brain and Language 26 : 201 29.
Bateman , F. 1870 . On Aphasia . London : Churchill .
Bhogal , S. K. , R. Teasell , and M. Speechley . 2003 . Intensity of aphasia
therapy, impact on recovery . Stroke 34.4 : 987 93.
Damasio , H. , and A. R. Damasio . 1980 . Te anatomical basis of conduc-
tion aphasia . Brain 103 : 337 50.
Dronkers , N. F. , B. B. Redfern , and C. A. Ludy . 1995 . Lesion localization
in chronic Wernickes aphasia . Brain and Language 51 : 62 65.
Aphasia Areal Distinctness and Literature
106
made from symbol systems that have one or more of the symp-
toms of the aesthetic: syntactic density, semantic density, syntac-
tic repletenesss, and exemplifcation. Tese notions are defned
later in this entry.
According to Goodman, A symbol system consists of a sym-
bol scheme correlated with a feld of reference ([1968] 1976 ,143).
Goodmans primary interest in defning a symbol system is to
diferentiate the notational from the non-notational schemes,
where notation is a technical notion to which his Chapter 4
is devoted. His concern about notations follows from a concern
about forgeries and fakes and with the fact that some types of
art (such as painting) can be faked while others (such as per-
formance of a specifc piece of music) cannot. Where a work is
defned by compliance to a score (i.e., it has a notation), it cannot
be faked; such works are called allographic. Where a work is not
defned by compliance to a score, as in the case of a painting, its
authenticity can be established only by tracing the history of its
production back to its origin, and this permits faking; such works
are called autographic.
A symbol system is built on a symbol scheme, which con-
sists of characters (and usually modes of combination for these
characters). For example, for a natural language, a character is a
class of marks, where marks might include anything from single
sounds or letters up to whole spoken or written texts, as in the let-
ter P, a character that is the class whose members are all the writ-
ings-down of the letter P. Symbol systems are either notations or
not notations. If the symbol system is a notation, the characters
of which its scheme is comprised must meet two conditions, as
follows:
(1) For a character in a notation, the members can be
interchanged, where diferent characters can be true cop-
ies of one another; this is called the condition of character-
indiference and is true, for example, of letters of the English
alphabet.
(2) Characters in a notation must be fnitely diferentiated
or articulate; for a mark that does not belong to two charac-
ters, it must be theoretically possible to determine that it does
not belong to at least one of them (this is explained further
shortly).
(1, 2) are the two syntactic requirements that defne a sym-
bol system as notational .
Characters in a scheme are correlated with things outside
the scheme. For example, the marks that make up a score are
correlated with elements in the performance of the score; the
mark that is a written word is correlated with a pronunciation of
that word; and the mark that is a written word is (also and inde-
pendently) correlated with that words referent. Goodman uses
the term complies and says that the performance complies
with the score, or the referent, or pronunciation, complies with
the written word. Te set of things that comply with an inscrip-
tion (e.g., the set of things named that can be denoted by a
name) is called the compliance class of the inscription . For the
symbol system to be a notation, it must frst include a symbol
scheme that is notational (i.e., that satisfes the two syntactic
conditions), and it must also satisfy three semantic conditions,
as follows.
exposure to salient features of a literary work. Implicit learning,
however, is likely to require many exposures, commonly while
immersed in the culture and language of the source tradition. In
isolating literary universals, then, we may take into account the
degree to which a particular property is likely to have been trans-
ported from one tradition to another by learning of either sort,
given the degree of contact between the traditions. For example,
the practice of dramatic performance may be transmitted from
one tradition to another through limited interaction, as this may
be learned through a single exposure. Te same point does not
hold for background imagery .
Finally, we may wish to expand our study of cross-cultural
patterns to actual borrowings. Here, too, it is crucial to distin-
guish diferent types of infuence. We may roughly divide infu-
ence into two categories hegemonic and nonhegemonic.
Hegemonic infuence occurs when the source tradition has
greater economic power (e.g., in the publication and distribu-
tion of literary works), more pervasive control of government
or education, or a higher level of prestige (due, for example, to
military strength), or when it is otherwise in a position of cul-
tural domination over the recipient society. Obvious cases are
to be found in colonialism . Common properties that result
from non-hegemonic infuences are not universals themselves.
However, they may tell us something about cross-cultural aes-
thetic or related propensities. Common properties that result
from hegemonic infuences, in contrast, may simply refect the
efects of power .
Patrick Colm Hogan
WORK CITED AND SUGGESTIONS FOR FURTHER READING
Comrie , Bernard. 1981 . Language Universals and Linguistic Typology .
Chicago : University of Chicago Press .
Hogan , Patrick Colm. 2005 . Literary universals and their cultural tra-
ditions: Te case of poetic imagery. Consciousness, Literature, and
the Arts 6.2 . Available online at: http://www.aber.ac.uk/cla/archive/
hogan.html.
ART, LANGUAGES OF
Languages of Art , a book by Nelson Goodman (190698), was
frst published in 1968, with a second edition in 1976. Te pres-
ent entry focuses solely on this book, which raises interesting
questions about language in a general sense and its role in
aesthetic experience. Tis entry does not attempt to contextu-
alize Goodmans book relative to his philosophy, for which see
Daniel Cohnitz and Marcus Rosenberg ( 2006 ) and Catherine Z.
Elgin ( 1992 ); Goodmans later and related Ways of Worldmaking
is also recommended (Goodman and Elgin 1978 ).
By languages (of art), Goodman means more generally sym-
bol systems ; natural language is one of the symbol systems, which
include, for example, musical notation or the symbol system of
cubist painting. Certain characteristics of symbol systems, when
used in an artwork, place cognitive demands on its audience,
which make the artwork good to think (to borrow Claude Lvi-
Strausss term). Te symbol systems from which artworks are
composed enable us to be exploratory, drawing on our cognitive
(including emotional) resources. Tis is because artworks are
Art, Languages of
107
of the outline of Mount Fuji. What makes the electrocardiogram
a diagram is that not every aspect of its form is relevant; the line
can vary in thickness or color without constituting a diferent
character. In contrast, every aspect of the form of the picture is
relevant; pictures have much fewer contingent features than dia-
grams, and pictures are thus said to be relatively (syntactically)
replete. Te diference between diagram and picture is a matter
of degree; repleteness is a relative characteristic.
Goodman concludes his book by using these notions to
develop four symptoms of the aesthetic. Objects have aes-
thetic symptoms when they use symbol systems that are syn-
tactically dense, semantically dense, and syntactically replete.
Te fourth symptom of the aesthetic is that aesthetic objects
exemplify. (In Ways of Worldmaking , Goodman explores the
notion of style and proposes that the style of an artwork is one
of the referents exemplifed by its symbols, where style consists
of those features of the symbolic functioning of a work that are
characteristic of author, period, place or school [1978, 35]. In the
same book, he introduces a ffth symptom of the aesthetic, which
is multiple and complex reference.) Note that the frst three of
these symptoms are characteristic of non-notational symbol sys-
tems; all three are associated with density in some more gen-
eral sense, which, Goodman says arises out of, and sustains, the
unsatisfable demand for absolute precision, thus engaging our
interest in aesthetic works ([ 1968 ] 1976, 253).
Goodman concludes his discussion by asking what gives an
aesthetic object its value, both relative to other aesthetic objects
and, more generally, to us: What makes us want to know it? He
argues that aesthetic objects invite our interest by asking us to
understand what they are, including how their symbol systems
operate, and what they exemplify; these tasks are made particu-
larly dif cult by the four symptoms of the aesthetic, which thus
particularly stimulate our interest in aesthetic objects. He sum-
marizes three criteria, drawn from general ideas about aesthet-
ics: Engagement with artworks improves our ftness to cope with
the world, just manifests our playfulness (i.e., homo ludens),
or communicates special kinds of knowledge to us. Tese are
partial insights into the primary purpose of our engagement with
aesthetic objects: Te primary purpose is cognition in and for
itself; the practicality, pleasure, compulsion, and communica-
tive utility all depend on this ([1968] 1976 , 258). Te symbol
systems or languages of art serve this purpose, allowing
for the possibility of producing symbolic objects that engage us.
Furthermore, the characteristic density of the symbolic systems
used in artworks, and the characteristic unparaphrasability of
what they express both permit a person to reenter the same art-
work and repeatedly to discover new things in it.
Nigel Fabb
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Cohnitz , Daniel , and Marcus Rosenberg . 2006 . Nelson Goodman .
Montreal : McGill-Queens University Press .
Elgin , Catherine Z. 1992 . Depiction. In A Companion to Aesthetics , ed.
David Cooper , 11316. Oxford : Blackwell .
. 1992 . Nelson Goodman. In A Companion to Aesthetics , ed. David
Cooper , 1757. Oxford : Blackwell .
Goodman , Nelson. [1968] 1976. Languages of Art . 2d ed.
Indianapolis : Hackett .
(3) Notational systems must be unambiguous; it must be
clear which object complies with each unique element of the
scheme.
(4) In a notation, compliance classes must be disjoint; for
example, a performance cannot comply with two diferent
scores.
(5) A notational system must be semantically fnitely difer-
entiated; for an object that does not comply with two char-
acters, it must be theoretically possible to determine that the
object does not comply with at least one of them.
Te notion of fnite diferentiation is important both in the
syntax and semantics of notational systems; fnite diferentiation
is articulation, and its lack constitutes density. As we will see,
though articulation is important for a notational system, density
is more generally important in defning works as aesthetic. Finite
diferentiation requires gaps between elements in the system
(between characters, or between compliants); if between two
adjacent elements a third can always be inserted, the scheme
lacks fnite diferentiation. For example, a scheme lacks fnite
diferentiation if it has two characters, where all marks not lon-
ger than one inch belong to one character and all longer marks
belong to the other, and where marks can be of any length.
Between a mark belonging to the character of marks not longer
than one inch and a mark belonging to the character of longer
marks, it is always (theoretically) possible to have a third that
falls between them (this ever-diminishing between-space is a
kind of Derridean mise-en-abme).
A symbol system is called a notation if it meets the fve condi-
tions. Goodman asks whether various types of symbol systems
that have been developed in the arts are notations. (A type of
artistic practice may be non-notational just because no notation
has been developed for it; in principle, notations might be devel-
oped for all of them, but in practice they have not been.) A tra-
ditional musical score is a character in a notational system. Te
compliants of the score are the performances, which collectively
constitute the work of music. Similar comments are made for
Labanotation, a scoring system developed for dance. A literary
work is a character in a notational scheme (but not a character
in a notational system): Like the language from which it is com-
posed, it meets the syntactic requirements for a notation, but not
the semantic requirements. A painting is a work that is in a sym-
bol system that is not notational.
Having developed these notions, Goodman uses them as
a way of defning a representation, a problem raised and not
solved in the frst part of the book, where, for example, he argues
that we cannot distinguish a representation by criteria such
as resemblance to the represented object. Representation for
Goodman is distinct from description (i.e., the term representa-
tion does not correspond to current cognitive science or linguistic
uses, in which propositions or tree structures are representa-
tions). A description uses a symbol scheme that is (syntactically)
articulate, whereas a representation uses a symbol system that is
dense (or uses symbols from a dense part of a symbol scheme).
He distinguishes between two types of dense (representational)
schemes, diferentiating a diagram from a picture. His example is
a pair of representations that are visually identical, consisting of a
peaking line, one an electrocardiogram and the other a picture
Art, Languages of
108
Sound is short-term variations or disturbances in ambient
air pressure. Tese pressure disturbances are created when air
moves from a region of high pressure to a region of low pressure.
Tere are three piston-like articulatory movements that can
create such pressure diferentials with respect to atmospheric
pressure. Tese, which J. C. Catford calls the initiation mecha-
nisms, are pulmonic, glottalic, and velaric. Tese mechanisms
can either create a positive pressure vis--vis atmospheric pres-
sure, in which case they are called egressive , or a negative pres-
sure, and then they are called ingressive .
Pulmonic egressive initiation is by far the most common. All
languages use it and most use it exclusively. Te chest cavity, by
virtue of decreasing its volume as in normal respiratory expira-
tion, compresses the air in the lungs, thus raising lung pressure
above that of the atmospheric pressure. Since speech neces-
sarily involves valves that impede the exiting airfow (e.g., the
adducted vocal cords and/or whatever articulations are made in
the oral cavity), the pulmonic or subglottal pressures developed
in speech are much larger than those seen in quiet respiratory
expiration. Because such initiation is so common, it is normally
not included in the usual phonetic descriptions; for example,
the [p] in French pre , which would otherwise be described as
pulmonic expiratory voiceless bilabial stop , is usually designated
simply as voiceless bilabial stop .
Pulmonic ingressive initiation (so-called ingressive voice)
is possible and is encountered in many cultures, notably in
Scandinavia and France, where short interjections, ja, oui, non,
can be uttered on ingressive voice (usually with some breathi-
ness), but although some sociolinguistic or pragmatic contrast
may be associated with this trait, no language documented so
far uses pulmonic ingressive initiation to make lexical contrasts.
Ingressive phonation may also be encountered as a (not very
efective) vocal disguise, and it is universally encountered as a
kind of coda to very young babies cries where the vocal cords are
still approximated but the respiration has shifted from expiratory
to inspiratory.
If the vocal cords are tightly closed and the larynx as a whole
is raised, acting like a piston, while there is a complete closure in
the oral cavity (and with the velum raised), a positive pressure
may be developed. Such sounds, glottalic egressives or ejectives ,
are not uncommon, being found in various African languages
(from diferent language families), in some languages of South
and Central America and the Pacifc Northwest (in the Americas),
and in the Caucasus. For example, Quechua bread is [tanta].
Glottalic ingressives or implosives involve the larynx most
commonly when the vocal cords are in voicing position being
lowered during the stop closure, thus creating a negative pres-
sure in the oral cavity or at least moderating the buildup of pos-
itive pressure. Historically, such stops often come from voiced,
especially geminated (long), stops, for example, Sindhi /pauni/
lotus plant fruit < Prakrit *paba. Enlarging the oral cavity helps
to maintain a positive pressure drop across the glottis, which
favors voicing. Although ejective fricatives are attested, there are
no implosive fricatives probably because the noise of a fricative
is generated when the air jet expands after leaving the narrow
constriction. Such expansion would occur inside the vocal tract
if made implosively, and the sound would be attenuated by the
oral constriction.
. 1988 . Reconceptions in Philosophy and Other Arts and Sciences .
London : Routledge .
Goodman , Nelson , and Catherine Z. Elgin. 1978 . Ways of Worldmaking .
Indianapolis : Hackett .
ARTICULATORY PHONETICS
Articulatory phonetics is that part of phonetics that studies how
speech is produced by the lips, tongue, velum (soft palate), lar-
ynx, and lungs to alter the air pressures and airfows and turbulent
noises in the vocal tract and to create the air spaces that yield the
resonances that diferentiate speech sounds. Te basic vocabulary
of articulatory phonetics is used for the taxonomic, that is, clas-
sifcatory, description of speech sounds. For example, the initial
sound in the French word pre , father, would be described as
a voiceless bilabial stop, symbolized using the IPA (International
Phonetic Alphabet) symbol [p]. Te initial sound in the English
word pear would be described as a voiceless aspirated bilabial
stop and symbolized using the IPA symbol [p
h
]. However, besides
this essential taxonomic function, articulatory phonetics stud-
ies the mechanisms (e.g., muscular, aerodynamic) that produce
speech and especially the how and why of variability in speech
sounds. Te following is a brief overview of the subject; for in-
depth accounts, the readings listed below should be consulted.
History
Te study of speech articulation and the development of a
descriptive terminology has an impressive history, with the frst
surviving instance being the Adhyy of the Sanskrit gram-
marian, Pini (ca. 500 b.p.e. ), who gave an articulatory account
of the relatively large sound inventory of Sanskrit. Other notable
achievements in the characterization of speech sounds were
given by many Greek grammarians, notably Dionysius Trax
(frst cent. b.p.e. ); the Arab and Persian grammarians al Khalil
Ibn Ahmad and Sbawaihi of the eighth century, who described
the Arabic of their times; the First Grammarian of Iceland
(twelfth cent.); and the work commissioned by and credited to
the Korean King Sejong (ffteenth cent.), which provided not only
an articulatory description of Korean as spoken then but also a
transcription, now the of cial orthography for Korean, hangul ,
which is partially iconic in its representation of how the sounds
are produced.
In Europe, the Baroque and modern eras saw dozens of
proposals for the description of speech sounds, for example,
by John Wilkins, Johan Conrad Amman , William Holder,
Francis Lodwick, Alexander J. Ellis, Robert Nares, Ernst Brcke,
Richard Lepsius, Alexander Melville Bell, Henry Sweet, and Otto
Jespersen . Although there is still some variation in the descrip-
tive terms, works such as Catford ( 1977 ) and Maddieson ( 1984 )
have helped to standardize the terminology.
The Basics
Speech articulations enable communication between speakers
and listeners because they create sound; it is the sound trans-
mitted to listeners and the perception of these sounds that are
the ultimate goal in speaking. Descriptions of articulation are
intended to capture the gestures that create these distinctive ele-
ments in the speech code.
Articulatory Phonetics
109
intense frication noise, because with no resistance to the airfow
at the glottis, the velocity of airfow will be greater at the oral con-
striction, and that also afects the degree and loudness from the
air turbulence .
Subcategories of sonorants include laterals , where the con-
striction is on one side of the palate, the other being open, for
example, the medial geminate (or long) alveolar lateral in Hindi
palla loose end of a sari used as a head covering [pla]. Nasals
are consonants made with a complete closure in the oral cav-
ity (at any place farther forward of the uvular region) but with a
lowered velum, for example, Tswana [ku] sheep with an ini-
tial velar nasal. Glides and approximants have nonlateral oral
constrictions that are not suf cient to generate turbulence, for
example, the labial-velar glide at the beginning of the English
word [wd]. Vowels are considered to have the least constriction
(descriptive terms follow) .
Place
Te primary places of articulation of speech sounds, proceeding
from the farthest back place to the farthest forward: glottal, pha-
ryngeal, uvular, velar, palatal, alveolar, dental, and labial. Some
of these places have already been illustrated. Finer place distinc-
tions can easily be made if necessary by appending the prefxes
pre - and post -, and multiple simultaneous constrictions can be
diferentiated by concatenating these terms as was done with the
labial-velar glide [w]. In most cases, these anatomical landmarks
on the upper side of the vocal tract are suf cient; if necessary, an
indication of the lower (movable) articulator can be specifed, for
example, the voiced labial-dental fricative [v] as in French voir
to see [vw] (as opposed to, say, the voiced bilabial fricative
[] in Spanish cerveza beer [sesa]).
State of the Glottis
In most cases, specifying whether the vocal cords are apart and
not vibrating (voiceless), lightly approximated and vibrating
(voiced), or tightly pressed together (glottal stop) is suf cient.
However, voicing itself occasionally needs to be further diferen-
tiated as breathy (a more lax type of voicing lacking energy in the
higher harmonics), tense (rich in higher harmonics), or creaky
(irregular staccato type of phonation, also with much energy in
the higher frequencies, though since it is irregular, one cannot
clearly identify harmonics as such). Many of the Indic languages
employ a distinctive breathy voice associated with voiced stops ,
for example, Hindi bhsh language [ b::], and many lan-
guages and many speakers voiced phonation changes to creaky
at a point of low F0 in intonation. (Low F0 is a low rate of vibra-
tion of the vocal cords due to lesser tension giving rise to low
pitch or note of the voice.) Creaky voice is also a common vari-
ant of glottal stop.
Vowels
Te descriptors for vowels deviate from those for consonants.
An imaginary quadrilateral space in the mouth (seen sagittally)
is posited, and vowels are said to have the high point of tongue
at regions in this space whose dimensions vertically are high
mid low and horizontally, front central back. (Tese may
also have further qualifers. In French the high front unrounded
vowel contrasts with a high front rounded vowel [i] and with a
If an air pocket is trapped between the tongue and palate or
the tongue dorsum and lips, and the tongue is lowered, a large
negative pressure can be generated, which, when released, can
create quite a loud sound. Such sounds, so-called clicks have
velaric ingressive initiation. Tey are common in cultures all over
the world as interjections, signals to animals, and so on. However,
they are used as speech sounds to diferentiate words only in a
few languages of southern and East Africa. Tey are very com-
mon in the Khoisan languages and a few neighboring Bantu lan-
guages. For example, Khoi one is [|ui] where the [|] is a dental
africated click (the sound often symbolized in Western orthog-
raphies as tsk or tut). In the Khoisan languages, clicks can be
freely combined with pulmonic and/or glottalic egressives either
simultaneously or in clusters. Velaric egressives, where a posi-
tive pressure is created by upward and forward movement of the
tongue, are not found in any languages lexicon but are used in
some cultures as a kind of exaggerated spitting sound, where the
positive pressure creates a brief bilabial trill.
In general, after the mechanism of initiation is specifed for
speech sounds, there are three main categories of terms to fur-
ther characterize them: place, manner, qualifers. For example,
the Russian [brat
j
] to take has in word-fnal position a voiceless
palatalized dental stop . Te manner is stop, the place is den-
tal, and voiceless and palatalized are qualifers.
Manners
Tere are two fundamental categories of manners, obstruent and
sonorant , each with subcategories. An obstruent is a sound that
substantially impedes the fow of air in the vocal tract to a degree
that turbulent noise is generated either as continuous frication or
as a noise burst. Obstruents may be stops or fricatives. (Ejectives,
implosives, and clicks are inherent stops.) Sonorants, which do
not impede airfow, are subdivided generally into laterals, glides,
approximants, nasals, and vowels.
Stops present a complete blockage of the airfow, for exam-
ple, the glottal stop in the name of the Hawaiian island Oahu
[oahu]. A special subclass of stops are africates , which are stops
with a fricative release, as in the initial and fnal consonants of
English judge [
PHON
(1)
criticized
SYNSEM
MONLOC . . .
. . .
LOC
LOC
CAT
CONT w(j )(i )
CONX
HEAD
SUBJ
COMS
VFORM fin
AUX
NP
normi
NP
normj
verb
Figure 1.
Habitus, Linguistic Head-Driven Phrase Structure Grammar
355
Te key valence features comp (plement) s and subj (ect)
encode the combinatorial requirements of particular lexical ele-
ments, making it possible to eliminate specifc phrase structure
rules in favor of broad schemata that correctly project phrases
from lexica regardless of the latters valence peculiarities. Each
such schema identifes a certain very general kind of structure,
which will be feshed out in detail, depending, in many cases,
on the particular lexical item that heads the structure. Two of the
most important conditions interacting with these schemata are
the valance principle and the head feature principle, which can
be stated roughly as follows:
Valence Principle: For any valence feature f, the value on the
mother is the list containing the value of f, on the head daugh-
ter minus the values corresponding to the head daughters
sisters .
Head Feature Principle (HFP): For any phrasal category,
the value of the head feature is identical to the value of the
categorys head daughters head feature .
Te valence principle specifes that the appearance of any
required valent in a local structural relationship to a select-
ing head removes the corresponding element from the must-
have list of the mother. Tis principle does not, of course, limit
the number of valents that may appear as the heads sisters.
But the schematic possibilities of English require a phrase of
this type to be lexically headed, and in order to satisfy both this
requirement and the valence principle, exactly the number
and type of complement sisters that the verb identifes on its
comps list (i.e., the list of descriptions that must be satisfed by
the heads selected sisters) must appear in the structure so as
to yield the empty complements lists on the mother. We thus
license the structure in (2)a, which can be abbreviated as (2)b
in Figure 2 . Te head-subject schema allows a completely satu-
rated verb (V) to have a phrasal head daughter with a subj list
of length one. Again, in conjunction with the valence principle,
this schema allows a verb phrase (VP) combining with a con-
stituent that exactly meets the description indicated in its subj
to appear as a structure of type head-subject under an S (i.e.,
clausal) node, that is, a V with both an empty comps and an
properties together with what are, in efect, lexical redundancy
rules, to account for all facts of the kind noted in the previous
paragraph and all syntactic dependencies, including parallelisms
between seemingly related construction types . Te following
discussion presents a more technically feshed-out instantiation
of this general approach to characterizing grammatical well-
formedness.
An Illustration of the System: Constituency and
Dependencies
As noted, local dependencies are uniformly accounted for in
HPSG as instances of nothing more elaborate than valence sat-
isfaction, or systematic relationships between valence speci-
fcations of related classes of lexica. Tis approach is suf cient
to account for even quite distant linkages, such as It continues
to appear to have been raining during the night , by means of a
feature subj encoding subject-selection possibilities on lexical
heads and their phrasal projections.
LOCAL DEPENDENCIES . Te best way to exhibit the nature of
HPSGs approach to an account of the patterns refected in natu-
ral language is to take an example of a single sentence and show
how it is licensed by the interaction of HPSG constraints. Assume,
for example, that the lexical entry for criticized comprises a num-
ber of separate (partial) descriptions that includes a description
roughly along the lines of (1) in Figure 1 .
As this (partial) lexical entry illustrates, signs in HPSG com-
prise a ramifed feature geometry that simultaneously specifes
phon (ology) and synt (tax)/ sem (antics), the latter a complex
information package revealing loc (al) and nonloc (al) prop-
erties, itemizing, respectively, the inherent morphosyntactic
and semantic properties of the category, on the one hand, and
the presence of elements within the sign with nonlocal link-
ages (the presence of a gap linked to a fller arbitrarily far away,
wh properties, etc.), on the other hand. Further subspecifca-
tions identify inherent morphosyntactic properties of the sign,
including head those necessarily shared with the heads
phrasal mother by virtue of the mother/head daughter rela-
tionship along with semantic and contextual information.
PHON
HEAD
SUBJ
SUBJ
COMPS
COMPS
AUX
AUX
VFORM fin
VFORM fin
verb
verb
HEAD
LOC
LOC
LOC
LOC
SYNSEM
SYNSEM
CAT
CAT
criticized
(a)
(2)
(b)
PHON criticized
criticized Leslie
Leslie
v NP
j
VP
NP
norm
i
NP
norm
i
NP
norm
j
NP
norm
i
CONT w(j )(i )
CONT w(j )(i )
Figure 2.
Head-Driven Phrase Structure Grammar
356
the percolation of wh properties (as in so-called pied-piping phe-
nomena), throughout arbitrarily large structures exist in natural
language and must be accounted for.
HPSG follows the central innovation introduced by Gerald
Gazdar a quarter of a century ago in treating fller/gap linkages
as a by-product of the difusion of a specifc feature slash , whose
value is correlated with that of the fller and which is driven
from mother to (at least one) daughter by the nonlocal feature
principle:
Nonlocal Feature Principle (NFP, simplifed): For any nonlo-
cal feature f the value of f on a phrase at any point below the
place in the structure where f was introduced is the union of
the values of f on the set of daughters.
Te NFL states, in essence, that below the point where the non-
local feature is introduced, at least one daughter in each two-
generation tree must bear the value of f specifed. In the case of
fller/gap constructions, the feature slash is identifed at the top
of the dependency with the relevant part (called the loc value,
illustrated in Figure 2 ) of the fller. Te descendents of the high-
est category with a nonempty slash feature preserve this value
in their own specifcations, until at the bottom of the depen-
dency, the slash value is cashed out as an empty category. An
example is illustrated in (3)a, with the relevant lexical entry for a
slash terminal category listed in (3)b in Figure 3 . Tis approach
to fller/gap dependencies comports well with familiar phenom-
ena in languages that mark such dependencies by local fagging
of extraction pathways; it also has considerable advantages over
movement-based approaches in the analysis of multiple gap
linkages to a single fller, such as parasitic and across-the-board
extractions in coordinate structures .
Remaining Issues
Tere are, of course, a number of other important aspects to
the grammar architecture incorporated in HPSG, some of them
involving major points of debate within the framework.
empty subj list. Tus, Robin criticized Leslie will be straightfor-
wardly licensed .
Note further that * Robin criticized and * Robin criticized Leslie
certain books will both be ruled out. Neither of them will satisfy the
head-complement schema under the constraint imposed by the
valence principle. In the frst case, the lack of a complement daugh-
ter will yield a result with a nonempty complements list, violating
the schema requirement. In the second case, the valence require-
ments on criticize will not equal the sum of those on the mother plus
the set of daughters that appear in the structure, as the full form of
the valence principle requires. Hence, both fail to be licensed.
Te valence feature lists do not in fact specify information
about the entire sister sign selected by the head. Lexical selection
is universally blind to phonological form and also to the descend-
ing constituent structure of any selected phrasal constituent;
therefore, it makes sense to restrict valence specifcations to syn-
sem values. But the latter contain enough information to imple-
ment virtually all local dependencies as simple expressions of
selection. Tus, it is possible to select not only for coarse-grained
information, such as the number of valents and their respective
lexical category types, but also for information such as their case,
or the infectional properties of their heads (on the assumption
that the latter are head features and will therefore be shared
between a phrasal valent and its lexical head daughter). Te mor-
phosyntactic dependency in English between the identity of aux-
iliary elements, on the one hand, and the infectional form of the
verb that immediately follows them, on the other hand, follows
simply and directly by specifying that, for example, each auxiliary
have selects as its complement a VP, preserving, in its own feature
specifcations, the infectional class of its head daughter (specif-
cally, the value psp , encoding past participial status).
Nonlocal Dependencies
Nonlocal dependencies, in contrast, do not depend on the
selectional possibilities of particular (classes of) lexical items.
Unbounded dependency constructions involving extraction, or
NP
LOC
Terry
NP
know
stories
V NP
N
PP
NP
LOC
SLASH
VP
SLASH
SLASH
SLASH
P
e
about
I
SLASH
PHON
SYNSEM
LOC
NONLOC|SLASH
S
S (a) (3)
(b)
1
1
1
1
1
1
1
1
1
Figure 3.
Head-Driven Phrase Structure Grammar
357
Because of the unique and circumscribed nature of his
1953 surgery, H. M. is probably the most studied patient in the
history of neuropsychology (Ogden and Corkin 1991): A neu-
rosurgeon inserted thin metal tubes above the eyes, and via
suction, removed parts of H. M.s hippocampus and directly
linked MTL structures. Tis operation greatly ameliorated H.
M.s life-threatening epilepsy, left H. M.s neocortex virtually
undamaged, and spared all neocortex with known links to lan-
guage comprehension. However, the operation caused a selec-
tive memory defcit , with normal recall of information familiar
to H. M. before his operation and used frequently since then,
but impaired recall of information newly encountered after his
operation and not massively repeated since then (see MacKay
et al. 2007).
H. M. has sentence -level language defcits that precisely
mirror his memory defcits. D. G. MacKay et al. (2007) tested
H. M.s sentence-level comprehension in six tasks. In one task,
participants identifed the grammatical versus ungrammati-
cal status of never previously encountered sentences that were
either grammatical or ungrammatical (see grammaticality ).
Here, H. M. responded with the correctly answer reliably less
often than controls matched for age, IQ, and education. Tis
comprehension defcit impaired a wide variety of syntactic struc-
tures, including ones that memory-normal participants fnd easy
to recall: H. M. exhibited equivalent comprehension defcits for
easy- and dif cult-to-recall sentences.
In a second task, H. M. again performed reliably worse than
controls in identifying grammatical sentences as grammatical
and in detecting, identifying, and repairing errors in sentences
containing incorrect and misordered words. A third task required
multiple-choice identifcation of who-did-what-to-whom in
novel sentences. Here, H. M. identifed the correct thematic
role of sentence constituents reliably less often than controls. A
fourth task required multiple-choice recognition of the appropri-
ate interpretation for sentences containing novel metaphors .
Here, H. M. chose the correct interpretation reliably less often
than controls, and his errors indicated failure to recognize that
the sentences were metaphoric. A ffth task required yesno
recognition of the appropriate interpretation for ambiguous
sentences. Here, H. M. responded correctly less often than con-
trols and sometimes responded yes- and -no despite repeated
requests to respond yes- or -no.
Consistent with several earlier results discussed next,
H. M.s ambiguity comprehension defcits were not due to
memory overload associated with multiple meanings: In the
ambiguity detection and description task of MacKay, Stewart,
and Burke (1998), H. M. took much longer than controls to
begin to describe the frst of two meanings in ambiguous sen-
tences, even when he never discovered the second meaning.
H. M. also discovered both meanings without experimenter
help less often than controls and often failed to understand
meanings that the experimenter had just explained. Research
isolated seven defcits in how H. M. described the sentence
meanings: Grammatically impossible interpretations, misread-
ings refecting failure to comprehend sentence-level mean-
ing, errors in pronoun use ( anaphora ), error correction
failures, free associative responses, self-miscomprehensions,
Te syntax/semantics interface issue is not fully resolved in
HPSG, in the sense that there are a number of competing
approaches to a variety of issues, ranging from the nature of
the objects in HPSG representations that map to semantic
representations to the mapping rules themselves. Diferent
answers to these questions typically correspond to major dif-
ferences in syntactic analyses.
Probably the most fundamental point of contention is the
disagreement within HPSG about the degree to which pat-
terns in natural languages refect, on the one hand, lexi-
cal properties and systematic mappings between lexical
descriptions, or restrictions on specifc constructions, as
encoded by constraints imposed on types belonging to very
elaborate ontologies, on the other hand. At one extreme,
HPSG includes a set of lexical heads with very abstract
properties and possibly no phonological realization, as in
earlier treatments of relative clauses. At the other extreme
are intricate multi-inheritance hierarchies in which prop-
erties of constructions are essentially posited as underived
primitives, or are derived by combining a number of such
underived primitives.
Tese and a number of other foundational issues are currently
under intense discussion and debate within the HPSG research
community, and there is no reason to expect a consensus on any
of these questions in the near future.
On the basis of the relatively brief history of the framework
so far, it seems very possible that divergences in approaches
to these fundamental matters will in time yield major schisms
within the theory, leading in the end to two or more major ver-
sions of the theory, in much the way that categorial grammar has
split into combinatory categorial grammar, on the one hand,
and the type-logical version based on the Lambek calculus, on
the other. It remains to be seen whether the theory will develop
along these lines.
Robert D. Levine
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Borsley , Robert . 1996 . Modern Phrase Structure Grammar .
London : Blackwells .
Levine , Robert, and Tomas Hukari . 2006 . Te Unity of Unbounded
Dependency Constructions . Stanford, CA : CSLI
Pollard , Carl, and Ivan Sag . 1994 . Head-Driven Phrase Structure Grammar .
Chicago : University of Chicago Press .
HIPPOCAMPUS
Te hippocampus is a bilaterally symmetric subcortical structure
adjacent to the lateral ventricle in the medial temporal lobe
(mtl) . Researchers in 1968 reported dramatic memory phe-
nomena associated with hippocampal-MTL damage, and data
reported from 1998 to 2006 have indicated parallel phenomena
for many other aspects of cognition, including language. I frst
discuss patient H. M., the initial source of data for the hippocam-
pal-memory and hippocampal-language links. I then discuss
related patient groups and the theoretical signifcance of the
hippocampal-language link.
Hippocampus
358
in reading novel sentences aloud (Friedman 1996; MacKay and
James 2001) .
Second, these parallels are dif cult to explain in current sys-
tems theories, in which independent systems process memory,
language comprehension, language production, and visual cog-
nition, and the hippocampus subserves only the memory sys-
tem (see, e.g., Schmolck, Stefanacci, and Squire 2000). Under
systems theories, hippocampal-MTL damage should yield
memory defcits without defcits in other cognitive systems, and
certainly without parallel defcits and parallel sparing across
supposedly independent systems for sentence comprehension,
sentence production, visual cognition, and episodic memory.
Tese predictions have failed, and major attempts to rescue
current systems theories from these failed predictions have like-
wise failed (see MacKay 2001, 2006; and MacKay, James, and
Hadley 2008).
Third, a new theoretical framework known as binding the-
ory (not to be confused with the anaphoric binding theory
in linguistics; see Jackendoff 2003, 15) readily explains and,
indeed, originally predicted the links between hippocampal-
MTL damage and parallel deficits and sparing in memory,
sentence-level language, and other aspects of cognition.
Under binding theory, hippocampal-MTL damage impairs
binding mechanisms for forming new internal representa-
tions in the cortex but does not affect mechanisms for activat-
ing already existing cortical representations (see, e.g., MacKay
et al. 2007 and James and MacKay 2001 for important theoret-
ical details regarding forgetting, frequency of use, and aging
and language ).
To illustrate in detail how binding theory explains his selec-
tive defcits, consider H. M.s sentence production in a standard
picture-description task requiring the incorporation of prespeci-
fed target words (MacKay et al. 2007): H. M. described the word-
picture stimuli signifcantly less accurately and completely than
eight controls, included fewer target words, and produced more
incomplete sentences (e.g., lacking a subject or verb), viola-
tions of agreement rules, non sequiturs, and run-on sentences
than the controls. Descriptions by H. M. (1a2a) versus controls
(1b2b) for the same word-picture stimuli illustrate some of these
diferences.
(1a) H. M. description : Because its wrong for her to be
and hes dressed just as this that hes dressed and the same
way.
(1b) Control description : Well, I think Ill take that one
although it looks wrong.
(2a) H. M. description : I want some of that pie either some
pie and Ill have some.
(2b) Control description : Uh, there are two people getting
pie, but theres only one piece of blueberry pie left, and so,
either one of them will have to have it.
Note that H. M.s picture-description problems in 1a and
2a were selective : Unlike agrammatic aphasics, H. M . did
not produce morphemes and nonsense words jumbled
together into morphological salads (Jackendoff 2003,
264). Moreover, he produced frequently used units, such as
its wrong, to be, the same way (1a), some of that, and
and failures to follow experimenter requests for clarifcation.
Research also indicated comprehension failure involving an
initial meaning for sentences, ambiguous or not.
To summarize, in a wide range of tasks involving many fun-
damental aspects of sentence comprehension, H. M. exhibited
defcits not caused by his memory problems (for corroborating
evidence on H. M.s comprehension defcits, see Corkin 1984;
Lackner 1974; and Schmolck, Stefanacci, and Squire 2000).
However, H. M.s comprehension defcits were selective rather
than across the board: Experiment six in MacKay et al. (2007)
demonstrated that H. M. comprehended familiar words and
phrases in isolation without defcit despite large defcits in com-
prehending these same stimuli when embedded within sen-
tences. Besides demonstrating selectivity, these results indicated
that H. M.s defcits were not attributable to low motivation, to
failure as a child to learn the meaning of the critical words and
phrases, or to failure to understand and follow instructions for
the task.
H. M. also exhibited signifcant production defcits when
describing the meanings of familiar words that he compre-
hended without defcit in MacKay et al. (2007): Judges blind to
speaker identity rated H. M.s meaning descriptions as reliably
more redundant, less coherent, less grammatical, and less com-
prehensible than those of controls. Tese fndings replicated
earlier results indicating defcits in H. M.s production of novel
or non-clich sentences (see MacKay, Stewart, and Burke 1998).
Again, however, H. M. exhibited selective production defcits that
mirrored his memory defcits, for example, spontaneously pro-
ducing clich phrases such as in a way (familiar from before his
surgery) without errors (ibid.).
H. M. also exhibited similar defcits and sparing in the seem-
ingly simple task of reading sentences aloud (MacKay and James
2001): He produced abnormal pauses at major syntactic bound-
aries unmarked by commas in the sentences, but normal pauses
at syntactic boundaries marked with commas, a prosodic marker
that H. M. had learned prior to his operation. H. M. also produced
abnormal pauses within unfamiliar phrases in the sentences, but
normal pauses within frequently used phrases. Tese and other
selective defcits indicated that he has dif culty with the process
of reconstructing novel aspects of sentence structure when read-
ing aloud .
H. M. also exhibited similar defcits and sparing in visual
cognition: When detecting target fgures hidden in concealing
arrays, he performed reliably worse than controls for unfamiliar
targets but not for familiar targets (MacKay and James 2000). In
short, H. M. exhibits similar selective defcits in visual cognition,
episodic memory, sentence-level comprehension, and sentence
production when speaking and reading aloud; impaired pro-
cessing of never previously encountered events, visual fgures,
phrases, and propositions; but spared processing of informa-
tion familiar to him before his lesion and used frequently since
then.
Why are these parallels important? One reason is that H. M.
is not unique: Other patients with hippocampal-MTL damage
exhibit identical parallels, reinforcing the links among hippocam-
pus-MTL, language, and memory. For example, other amnesiacs
exhibit defcits in detecting the two meanings in ambiguous sen-
tences (Zaidel et al. 1995) and make errors resembling H. M.s
Hippocampus
359
memory and sentence production . Journal of Experimental and
Clinical Neuropsychology 30 .3: 280300.
MacKay , D. G., L. E. James, J. K. Taylor, and D. E. Marian . 2007 .
Amnesic H. M. exhibits parallel defcits and sparing in language and
memory: Systems versus binding theory accounts . Language and
Cognitive Processes 22 .3: 377452.
MacKay , D. G., R. Stewart, and D. M. Burke . 1998 . H. M. revis-
ited: Relations between language comprehension, memory, and the
hippocampal system . Journal of Cognitive Neuroscience 10 : 377 94.
Ogden , J. A., and S. Corkin . 1991 . Memories of H. M. In Memory
Mechanisms: A Tribute to G. V. Goddard , ed. W. C. Abraham , M.
Corballis, and K. G. White , 195215. Hillsdale, NJ : Erlbaum .
Schmolck , H., L. Stefanacci, and L. R. Squire . 2000 . Detection and expla-
nation of sentence ambiguity are unafected by hippocampal lesions
but are impaired by larger temporal lobe lesions . Hippocampus
10 : 759 70.
Zaidel , D. W., E. Zaidel, S. M. Oxbury, and J. M. Oxbury . 1995 . Te inter-
pretation of sentence ambiguity in patients with unilateral focal brain
surgery . Brain and Language 51 : 458 68.
HISTORICAL LINGUISTICS
Historical linguistics is the study of how languages change over
time. We can approach the study of change in various ways. One
is by studying the histories of individual languages. An example
would be analyzing the changes that have taken place in English
over the last thousand years. A second approach involves com-
paring various related languages in order to draw inferences
about the types of changes that have occurred since the time
they split from their common ancestor. We can further study the
frequency and naturalness of the changes that the languages are
hypothesized to have undergone and the efects of change in
one area on other parts of the language. Finally, historical lin-
guistics is also concerned with fnding explanations for change,
including why languages change and how a particular change
was actuated in a particular circumstance. Historical linguistics
also intersects with other felds. For example, a linguist trying
to reconstruct the geographic extent of a proto-language may
also make use of data from both archaeology and historical
anthropology.
Te following discussion begins with an abbreviated history
of the feld, from classical times to the early twentieth century.
From there, we turn to contemporary research and conclude
with a look at possible future directions for the feld.
History of the Field
Modern historical linguistics was developed in the nineteenth
century, although there were both eighteenth- and seventeenth-
century scholars who practiced something that todays linguists
would recognize. In contrast, Classical Greek and Roman lin-
guists had little to say about language history despite consider-
able sophistication in their synchronic descriptive techniques
(see synchrony and diachrony ). Te study of language
change in the Graeco-Roman world was largely confned to
etymology that is, to claims about the origin of individual lexi-
cal items. For example, in Platos Cratylus , the word anthropos is
said to derive from the contraction of the phrase anathro:n h
po:pen (looking up at the things hes seen). Similar methods
were employed by Latin linguists such as Varro, who claimed that
Ill have some (2a), without errors. Under binding theory,
separately stored syntactic units and rules serve to activate
already formed internal representations so that words and
phrases become produced in the appropriate order. Because
H. M.s syntax-based activation mechanisms are intact and
frequently used since his lesion, H. M. therefore produces
familiar words, phrases, and propositions such as its wrong
and Ill have some, without errors. However, he lacks already
formed internal representations for propositions that he
has used repeatedly before and after his lesion to describe the
MacKay et al. (2007) word-picture stimuli. The word-picture
stimuli, therefore, triggered familiar units that H. M. simply
concatenated without forming complete, appropriate, and
coherent utterances (see 1b, 2b).
In conclusion, the pressing problem for future research is to
test new binding theory predictions for relations among brain,
language, memory, and other aspects of cognition (see MacKay
et al. 2007) .
Donald G. MacKay
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Corkin , S . 1984 . Lasting consequences of bilateral medial tempo-
ral lobectomy: Clinical course and experimental fndings in H. M.
Seminars in Neurology 4 : 249 59.
Friedman , R. B . 1996 . Phonological text alexia: Poor pseudo-word
reading plus dif culty reading functors and af xes in text Cognitive
Neuropsychology 13 : 869 85.
Jackendof , R . 2003 . Foundations of Grammar: Brain, Meaning, Grammar,
Evolution . Oxford : Oxford University Press .
James , L. E., and D. G. MacKay. 2001 . H. M., word knowledge and
aging: Support for a new theory of long-term retrograde amnesia .
Psychological Science 12 : 485 92.
Lackner , J. R . 1974 . Observations on the speech processing capabilities
of an amnesic patient: Several aspects of H. M.s language function .
Neuropsychologia 12 : 199 207.
MacKay , D. G . 2001 . A tale of two paradigms or metatheoreti-
cal approaches to cognitive neuropsychology: Did Schmolck,
Stefanacci, and Squire demonstrate that Detection and explanation
of sentence ambiguity are unaffected by hippocampal lesions but
are impaired by larger temporal lobe lesions ? Brain and Language
78 : 265 72.
. 2006 . Aging, memory and language in amnesic H. M .
Hippocampus 16 : 491 4.
MacKay , D.G., D. M. Burke, and R. Stewart . 1998 . H. M.s language pro-
duction defcits: Implications for relations between memory, seman-
tic binding, and the hippocampal system . Journal of Memory and
Language 38 : 28 69.
MacKay , D. G., and L. E. James . 2000 . Binding processes for visual cog-
nition: A hippocampal amnesic (H. M.) exhibits selective defcits in
detecting hidden fgures and errors in visual scenes. Poster presented
to the Cognitive Neuroscience Society, San Francisco .
. 2001 . Te binding problem for syntax, semantics, and
prosody: H. M.s selective sentence-reading defcits under the
theoretical-syndrome approach . Language and Cognitive Processes
16 : 419 60.
. 2002 . Aging, retrograde amnesia, and the binding problem for
phonology and orthography: A longitudinal study of hippocampal
amnesic H. M . Aging, Neuropsychology, and Cognition 9 : 298 333.
MacKay , D. G., L. E. James, and C. Hadley . 2008 . Amnesic H. M.s per-
formance on the Language Competence Test: Parallel defcits in
Hippocampus Historical Linguistics
360
William Jones to the Royal Asiatic Society in 1786, in which he
noted similarities between Sanskrit and the languages of Europe
and hypothesized that they may come from a common ances-
tor: [N]o philologer could examine the Sanskrit, Greek, and
Latin, without believing them to have sprung from some com-
mon source, which, perhaps, no longer exists. Tere is a similar
reason, though not quite so forcible, for supposing that both the
Gothic and the Celtic had the same origin with the Sanskrit.
(Te speech is quoted in almost all introductory textbooks for
historical linguistics; see, for example, Campbell 2004 and Trask
1996.) However, as we have seen, elements of comparative and
historical linguistic methods predate Jones by several hundred
years.
Te nineteenth century saw an explosion of work on his-
torical linguistics and the reconstruction of language history;
in fact, many methods developed during this period are still in
use. It is this period that gives us the idea of the correspondence
set: a set of words in related languages that are descended from
a common proto-form and which exhibit regular phonologi-
cal correspondences. For example, English three, Latin trs,
Greek tris, Sanskrit tryas , and Gothic rs all refect proto-
Indo-European *treyes. Furthermore, the correspondences
among phonemes in these languages are regular. For example,
English th corresponds to Latin t in cognate words (cf. fa th er
and pa t er, bro th er and fra t er , among others; for the method and
further reconstructions, see Trask 1996). Te frst detailed dis-
cussion of such correspondence sets dates to A. Turgot ([1756]
1961), and the method was systematized by Rasmus Rask
(1818). We also, shortly afterwards, fnd the frst reconstructions
of Indo-European. Tese are in the work of August Schleicher
(1848), who not only reconstructed lexical items but also con-
structed an Indo-European fable. Schleicher also introduced
the Stammbaum, or family tree, model of linguistic relationship
(see language families ) .
Neither Schleicher nor his contemporaries placed much
weight on the importance of regularity in correspondences,
however. Scholars of the following generation, including Karl
Brugmann and Berthold Delbrck (the Neogrammarians or
Junggrammatiker ), were the frst to recognize the importance
of regularity in sound change for the comparative method
and to use it as a tool for discriminating between inheritance
and analogy . Te recognition of regularity in sound change
is the pillar of historical reconstruction and forms the
basis of much modern work on historical linguistics . Without
a conception of sounds changing regularly in particular pho-
netic environments, it is impossible to identify irregularities,
to reconstruct proto-forms, and thereby to form a reliable idea
of linguistic relationships. However, arguments about the uni-
versal regularity of sound change continue. One area involves
the paradox between apparent regularity at the macro level
and irregularity when a language at a particular stage in time
is examined. Tat is, a language is not homogeneous at any
stage because of the amount of dialectal variation among
speakers. A second point of debate is the applicability of mod-
els relying on regularity of sound change to languages outside
of Europe, for instance, those spoken by huntergatherer com-
munities in Australia.
anas duck is related to the verb nre to swim. Furthermore,
although numerous similarities between Latin and Greek were
noted, it was assumed that all such words were direct borrow-
ings into Latin from Greek. Shared common ancestry from a lan-
guage no longer spoken was never considered (see extinction
of languages ). Such an assumption was in keeping with the
strong cultural debt that the Roman world owed to the Greek
(see Law 2003) .
Te classical etymological method continued to be employed
throughout the Middle Ages, where it was joined by theories of
language change and diversity built on the biblical story of the
destruction of the Tower of Babel (Gen. 11). A summary of the
theory can be found in Dantes de Vulgari Eloquentia (for one
English translation, see Shapiro 1990). Such a theory of diversity
specifes both a cause of language change and a partial model
of the origin of modern linguistic diversity. Work within a model
of change laid the foundation for much later linguistic scholar-
ship, for it led to questions about the language that was spoken
by those erecting the Tower of Babel (and, therefore, what the
frst human language was) and exactly how modern attested lan-
guages related to one another.
Te de Vulgari Eloquentia is also a founding discussion of
relationships among the vernacular languages of Europe. Te
rise of the study of Romance vernaculars led to an examination
of systematic diferences among those languages, as well as com-
parison with Latin (for example, why Latin de is a preposition
meaning from, but in French and Italian it marks possession).
As R. H. Robins (1968, 100 f) notes, it was this examination that
allowed the development of an adequate framework for dia-
chronic linguistics because of a chance to study change where
the parent language was already well understood.
In the early Middle Ages, there was also a highly sophisticated
Arabic comparative linguistic tradition. Ibn Hazm (9941064)
noted regular correspondences among Hebrew, Arabic, and
Syraic, and in the Ihkam Ibn Hazm further identifed changing
pronunciation and language contact as driving forces in the cre-
ation of linguistic diversity (see contact, language ).
In the sixteenth to eighteenth centuries, we begin to see the
study of language change linked to other branches of linguistics,
such as typology . Scholars such as Konrad Gessner ([1555]
1974), Joseph Justice Scaliger ([1599] 1610), and later Andreas
Jger (1686) and Peter Simon Pallas (1786) collected and com-
pared vocabularies of the languages available to them and made
hypotheses about linguistic relationships on this basis. However,
the comparisons are unsystematic and based mostly on very few
features. For example, Scaliger divides the languages of Europe
into four major classes, depending on whether their word for
god is based on deus, theos, gott , or bog (roughly correspond-
ing to Latin/Romance, Greek, Germanic, and Slavic, respec-
tively). Gessner ([1555] 1974, 110) deduces that Armenian is
closely related to Hebrew because of the similarity of words such
as lezu tongue (Hebrew laschon in Gessner) and hhatz cross
(Hebrew etz or hetz ). Moreover, until G. W. Leibniz, there is no
conception that languages could have been descended from a
language that is no longer attested .
Te beginning of modern historical linguistics and the com-
parative method is often said to date from a speech given by Sir
Historical Linguistics
361
work on historical syntax concerns word order change in the syn-
chronic analysis of ancient languages or the causes of syntactic
change, rather than reconstruction per se. Infuential here has
been the work of David Lightfoot (e.g., 1999), who has developed
a theory of change that assigns the primary cause of change to
a childs acquisition of the language. Since children are exposed
to linguistic data that is slightly diferent from what their parents
were exposed to, they therefore draw slightly diferent conclu-
sions about the syntactic structure of their language. We see this
refected in the historical record as a syntactic change. Others are
less comfortable in ascribing change solely to grammar change at
acquisition and argue that syntactic changes also occur in adult
speakers as a result of exposure to new languages and dialects,
changing prestige , and other factors. Tere is ongoing debate
over the extent to which a person may spread change as an adult,
since there are also clearly generational diferences in linguistic
production (see age groups ) .
Te study of reconstructions within historical phonology,
morphology, and syntax may also be used to classify languages
into genetic families. Language classifcation must also take ideas
of language contact into account. Extensive contact between two
unrelated languages may over time lead to enough similarities
that it is dif cult to tell whether they are related or not. Several
languages have been misclassifed on this basis. For example,
Armenian was originally classifed as an Indo-Iranian language
rather than as its own branch of Indo-European because of the
number of loans it exhibits .
Tere are other less widely accepted methods of investigat-
ing linguistic prehistory. One is lexicostatistics, which involves
estimating genetic relatedness by comparing the percentage of
vocabulary common to pairs of languages. Underlying the method
is the assumption that languages that share more common mate-
rial are likely to be more closely related. Glottochronology uses
the estimations from lexicostatistics to estimate the time depth
of a particular family. Mass comparison (e.g., Greenberg 1987)
involves using large-scale word lists to reconstruct further back
than the strict application of the comparative method allows, by
granting more exceptions to the regularity of sound change and
greater latitude in semantics . In each case, the methods are not
widely accepted. For example, gross similarity between two
languages may be caused by several factors apart from common
genetic inheritance, including chance and borrowings, and only
detailed reconstruction of correspondences by the comparative
method allows us to choose between them .
Future Prospects
Currently, our knowledge of the history of diferent language
families in the world is very uneven. Some families includ-
ing Indo-European, Finno-Ugric, and Algonquian have been
reconstructed in detail. In other cases, we are not even sure
which languages belong to the family, let alone what the proto-
language looked like.
Tere are still many pressing concerns and active areas of
research. Te frst is in the reconstruction of various language
families. Basic original reconstruction research is needed for
much of the world. Secondly, there is an ever-increasing concern
with questions about how and why languages change. We have
long moved away from arguments of language change involving
A further methodological coup is due to Ferdinand de
Saussure ([1915] 1972). Saussure hypothesized, on the basis of
both internal evidence of root structure and varied correspon-
dences among vowels in Indo-European languages such as Greek
and Sanskrit, that there had once been a further set of laryngeal
consonants (possibly /h/, // and /
w
/) that had disappeared in
all environments in attested languages. Saussures contribution
is very important to historical linguistics for two reasons: First, it
is the foundation of internal reconstruction , that is, the
hypothesis of reconstructions based on synchronic patterns in
one language, rather than a direct comparison of forms between
languages. Second, it demonstrates very clearly the power of the
comparative method and the importance of regularity in corre-
spondence sets. Te subsequent decipherment of inscriptions in
Hittite and Luwian confrmed Saussures hypothesis since two of
the laryngeals are preserved in these languages precisely where
we should expect to fnd them.
Historical linguists in the twentieth century made progress in
the reconstruction of families outside Europe, in the growing use
of quantitative methods in modeling and describing language
change, and in historical syntax (see syntactic change ). Most
of the methods used today were developed through the recon-
struction of proto-Indo-European; however, the methods have
also been successfully applied to other families. Much early his-
torical work was done on Finno-Ugric languages (e.g., Sajnovics
[1770] 1968; Gyarmathi 1799), and more recently, there has been
considerable progress in the reconstruction of Austronesian
(Pawley and Ross 1993), Niger-Congo (cf. Hombert and Hyman
1999), and numerous families in North America (see, for exam-
ple, Campbell 1997).
While the family tree model has been very infuential in his-
torical linguistics, other models of language change are also
used. Perhaps the most commonly cited is the wave theory of
J. Schmidt and Jules Guilliron (see Guilliron 1921), who pro-
posed that sound changes difuse through the lexicon, gradually
afecting more and more instances of phonemes in a given envi-
ronment. Regularity in correspondences is thus epiphenomenal
and only appears once a change is complete. Another common
appeal to wave theory is in subgrouping, where it is argued that
the family tree is not an accurate representation of language split-
ting. Rather, linguistic diferentiation occurs through the gradual
building up of isoglosses and changes afecting individual lexical
items .
Current State of Research
Te methods of historical linguistics can be applied to all areas
of language study. Within historical phonology, there has been
a great deal of work on types of sound change, the plausibility
of diferent changes, and the mechanisms by which change is
spread throughout a speech community. (For discussion and
diferent approaches, see Ohala 1993, Blevins 2004, and Labov
2001.) Te study of morphological change was particu-
larly important in the nineteenth century, and the reconstruc-
tion of morphemes and paradigms is still an important area of
research , as is grammaticalization theory .
Historical syntax has received rather less attention than histor-
ical morphology or phonology mostly because of the dif culties of
applying the comparative method to syntactic constructions. Most
Historical Linguistics
362
Labov , William . 2001 . Principles of Linguistic Change .
Oxford : Blackwell .
Law , Vivian . 2003 . Te History of Linguistics in Europe: From Plato to
1600 . Cambridge : Cambridge University Press .
Lightfoot , David . 1999 . Te Development of Language: Acquisition,
Change and Evolution . Oxford : Blackwell .
McMahon , April, and Robert McMahon . 2006 . Language Classifcation by
Numbers . Oxford : Oxford University Press .
Ohala , John . 1993 . Te phonetics of sound change. Historical
Linguistics: Problems and Perspectives , ed. Charles Jones , 23778.
London : Longman .
Pallas , Peter Simon . 1786 . Linguarum totius orbis vocabularia compara-
tiva . St. Petersburg .
Pawley , Andrew, and Malcolm Ross . 1993 . Austronesian histori-
cal linguistics and culture history . Annual Review of Anthropology
22 : 425 59.
Rask , R . 1818 . Undersgelse om det gamle nordiske eller islandske sprogs
oprindelse . Copenhagen.
Robins , R. H. 1968 . A Short History of Linguistics . Bloomington : Indiana
University Press .
Sajnovics , Jnas . [1770] 1968. Demonstratio idioma Ungarorum et
Lapponum idem esse , ed. Tomas Sebeok . Bloomington : Indiana
University Press .
Saussure , Ferdnand de . [1915 ] 1972 . Course in General Linguistics
[Cours de linguistique general] . Trans. Roy Harris. Peru, IL : Open Court
Classics .
Scaliger , Joseph Justus . [1599] 1610. Diatriba de Europaeorum linguis. In
Opuscula varia antehac non edita . Paris .
Schleicher , August. 1848. Sprachvergleichende Untersuchungen. / Zur
vergleichenden Sprachgeschichte . 2 vols. Bonn : H. B. Koenig .
Shapiro , Marianne . 1990 . De Vulgari Eloquentia , Dantes Book of Exile.
Lincoln and London : University of Nebraska Press .
Tomason , Sarah Grey, and Terrence Kaufman . 1988 . Language Contact,
Creolization, and Genetic Linguistics . Berkeley : University of California
Press .
Trask , R. L . 1996 . Historical Linguistics . London : Arnold .
Turgot , A . [1756] 1961. tymologie. Brugge : De Tempel.
HISTORICAL RECONSTRUCTION
Like all complex things in nature, languages change over
time. Historical reconstruction is a process of inference by
which changes are undone so as to recover certain aspects
of historically nonattested linguistic structure and content in
hypothetical form. Although it can be applied to languages
for which written documentation is available (e.g., the recon-
struction of proto-Romance next to the extant records of Latin)
or to single languages by means of internal reconstruction, his-
torical reconstruction typically is directed to prehistoric lan-
guages and requires the simultaneous comparison of multiple
witnesses.
Te set of procedures used in historical reconstruction is
called the comparative method . After the elimination of
chance, borrowing, and universals as plausible causes of cross-
linguistic similarity, it can be shown that the resemblances
between two or more languages must result from a common
origin (usually called genetic relationship ), coupled with diver-
gent descent. In this respect, historical reconstruction in lin-
guistics shows striking parallels to the study of the evolutionary
history of natural species, and much of its terminology is con-
ceptually modeled on that of biological taxonomy. Once it has
sloppy speech or linguistic degeneration by ignorant speakers.
Instead, research has focused on the relative importance of lan-
guage acquisition in language change versus social factors, such
as peer pressure, prestige, and diffusion . Linguistic research is
also important for the study of prehistory and ancient population
movement.
Not everyone is convinced that the methods discussed in
this entry are generally applicable to all languages and lan-
guage families in the world. As already noted, the family tree
has been the predominant model for 150 years. However, some
have pointed out its reliance on transmission from parents
to children, that ignores other types of transmission which
can lead to rapid language change, such as creolization (see
creoles) and the formation of mixed languages (Tomason
and Kaufman 1988) .
Finally, historical work increasingly involves computational
modeling and the integration of techniques used in computa-
tional biology. Te last 10 years have seen an increasing amount
of sophisticated statistical analysis and computational modeling
in research (for an overview, see McMahon and McMahon 2006).
We also see work that aims at estimating time depth and rates of
language change. It remains to be seen, however, how successful
this work will be. No matter how sophisticated the techniques for
statistical analysis, any estimates of time depth need also to take
into account sophisticated theories of language change. At this
point, we have no idea why languages change and split at difer-
ent rates, although such difering rates are clearly observable in
the historical record .
Claire Bowern
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Blevins , Juliette . 2004 . Evolutionary Phonology. Cambridge : Cambridge
University Press .
Campbell , Lyle . 1997 . American Indian Languages: Te Historical
Linguistics of North America . Oxford : Oxford University Press .
. 2004 . Historical Linguistics: An Introduction . Cambridge, MA : MIT
Press .
Gessner , Konrad . [1555] 1974. Mithridates: de Diferentiis Linguarum
tum Veterum tum quae Hodie apud Diversas Nationes in Toto Orbe
Terrarum in Usu Sunt . Neudruck der Ausgabe Zrich, Aalen: Scientia
Verlag .
Greenberg , Joseph . 1987 . Language in the Americas. Stanford,
CA : Stanford University Press .
Guilliron , Jules . 1921 . Pathologie et thrapeutique verbales .
Paris : Champion.
Gyarmathi , Smuel . 1799 . Affinitas linguae Hungaricae cum linguis
Fennicae originis gram matice demonstrata. Vocabularia dialec-
torum Tataricarum et Slavicarum cum Hungarica comparata .
Gttingen .
Harris , Alice, and Lyle Campbell . 1995 . Historical Syntax in Cross-
Linguistic Perspective . Cambridge : Cambridge University Press .
Hock , Hans H . 1991 . Principles of Historical Linguistics . Berlin :
Mouton .
Hombert , Jean-Marie, and Larry Hyman . 1999 . Bantu Historical
Linguistics: Teoretical and Empirical Perspectives. Stanford,
CA : CSLI .
Jger , Andreas . 1686 . De Lingua Vetustissima Europae, Scytho-Celtica et
Gothica . Wittinberg .
Historical Reconstruction Historical Linguistics
363
dispersal, or homeland, of a language family (or subgroup
ancestor) and, hence, gives rise to hypotheses about direction
of migration that can be tested against the evidence of other
scholarly disciplines, such as archaeology or population genet-
ics. However, it has been recognized since at least the 1870s
that not all processes of linguistic diferentiation are treelike,
and it is widely accepted that both family tree and wave mod-
els accurately describe the process of language split, the former
under conditions of sharp social separation and the latter under
conditions of gradual diferentiation of independent languages
from a dialect complex .
Te reconstruction of the proto-Indo-European case system
in the frst half of the nineteenth century marked the beginning
of work on comparative morphosyntax, but many would argue
that since this involves the identifcation of cognate af xes, it is
a variant of lexical reconstruction. In recent years, greater atten-
tion has been paid to problems of reconstruction in other areas
of syntax , such as word order, and in semantics . It is note-
worthy that the models for such work almost invariably derive
from typological approaches to synchonic linguistic structure,
rather than from formal theories of syntax (see synchrony and
diachrony ).
A proto-language inevitably presents a very incomplete pic-
ture of the language that must actually have existed. Nonetheless,
the comparative method, which is generally thought to permit
reconstruction of languages up to about 6,000 years old, is a
powerful tool that allows a variety of inferences about prehis-
toric language communities and their cultures. Te potential
use of linguistic reconstruction for inferences about culture his-
tory was recognized in the second half of the nineteenth century
and labeled linguistic palaeontoloy by Ferdinand de Saussure.
However, little use was made of this potential until relatively
recent times. Since roughly the 1970s there has been increasingly
fruitful interdisciplinary cooperation, especially between histor-
ical linguists and archaeologists, in exploring Holocene human
prehistory. Tis has led to a renewed inquiry into the antiquity of
the Indo-European settlement of Europe and has been a power-
ful force in understanding the prehistoric human settlement of
the Pacifc .
Robert Blust
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Blust , Robert . 1987 . Lexical reconstruction and semantic recon-
struction: Te case of Austronesian house words . Diachronica
4 : 79 106.
Campbell , Lyle . 1999 . Historical Linguistics: An Introductio n . Cambridge,
MA : MIT Press .
Harris , Alice C ., and Lyle Campbell . 1995 . Historical Syntax in Cross-
Linguistic Perspective . Cambridge : Cambridge University Press .
Kiparsky , Paul . 1995 . Te phonological basis of sound change. In Te
Handbook of Phonological Teory, ed. John A. Goldsmith , 64070.
Oxford : Blackwell .
Renfrew , Colin . 1998 . Archaeology and Language: Te Puzzle of Indo-
European Origins . London : Cape.
Ross , Malcolm, Andrew Pawley, and Meredith Osmond , eds. 1998.
The Lexicon of Proto Oceanic: The Culture and Environment
of Ancestral ceanic Society . Vol. 1. Material Culture. Pacific
Linguistics C-152. Canberra : Department of Linguistics, Research
been determined that languages are genetically related, a more
exact picture of their historical connection can be achieved by
the reconstruction of a proto-language or hypothetical common
ancestor.
Although some progress has been made with other aspects
of historical reconstruction, it is widely agreed that the
comparative method has been successfully applied only in
phonology . For reconstruction to begin, it is necessary to
identify a corpus of cognate morphemes , that is, morphemes
that have a common historical origin, identified inductively
as forms of similar meaning that exhibit recurrent sound
correspondences. Distinct sound correspondences that are
found in the same or closely similar environments normally
must be attributed to different proto- phonemes , and the
inventory of proto-phonemes so inferred forms a hypothesis
about the sound system of the proto-language of a language
family . Since proto-phonemes can only be reconstructed in
lexical forms, phonological and lexical reconstructions
are inextricably bound together. Reconstructed phonemes
and reconstructed words (proto-forms) are preceded by an
asterisk to indicate their hypothetical status. Some linguists
take the position that the phonetic substance of such sym-
bols is beyond recovery and that proto-phonemes are, there-
fore, little more than abstract formulas used to summarize
sound correspondences (the formulaic position). The major-
ity view (the realist position) is more sanguine; although the
phonetic nature of some proto-phonemes clearly is contro-
versial, many others permit little latitude in interpretation.
A related, though distinct, issue concerns the structure of
reconstructed phonological systems, since some of these
have violated implicational universals in typology .
Where this occurs, most historical linguists today would
question the validity of the reconstruction, using the typo-
logical generalizations of the present as a guide to inferences
about the past .
Once lexical reconstructions are available, it becomes pos-
sible to determine sound changes in a large number of daugh-
ter languages. No topic in linguistics has a longer history than
the study of sound change, which commenced during the frst
quarter of the nineteenth century with the pioneering studies of
Rasmus Christian Rask and Jacob Grimm . A major point of con-
troversy in the study of sound change is the issue of regularity.
It is now generally agreed that the strict Neogrammarian posi-
tion, which ruled out the possibility of unconditioned phone-
mic splits, is overly restrictive. A second issue, which is yet to be
resolved, is whether all sound change is phonetically (or phono-
logically) motivated.
An examination of sound changes leads not only to theoret-
ical models of how (and why) this process occurs but also to
evidence for subrelationship within a language family. Sound
changes that are exclusively shared ( exclusively shared inno-
vations ) form the basis for linguistic subgroups. Subgrouping
allows linguists to move beyond the mere recognition of a
language family as an internally undiferentiated collection
of related languages to the reconstruction of a family tree that
defnes the historical order of splits of major and minor groups
of languages within the family. Te structure of a family tree,
in turn, supports inferences about the most likely center of
Historical Reconstruction
364
Theoretical Approaches
At the onset, it is important to understand the possible
claims about grammatical knowledge during the holophras-
tic period. The most conservative claim would be, basically,
what you see is what you get, that is, that children only pro-
duce single words because that is all they know at that point.
This approach has come to be known as a lean interpretation
of childrens knowledge. At the other extreme, one can claim
that childrens grammatical knowledge is greater than what is
reflected in single-word productions. This rich interpretation
is based on the fact that language acquisition is rapid and that
children must have innate language learning mechanisms
that enable them to determine the grammatical character-
istics of their language at a very early age (see innateness
and innatism ).
Tese two positions can be demonstrated by again looking at
a childs production of the words doggie and eat . A lean interpre-
tation would be that the child has some initial semantic cat-
egorization of word meaning , such that doggie represents
an emerging category of animate objects, and eat represents
an emerging category of actions. A rich interpretation would
propose that innate principles enable the child to establish
explicit grammatical knowledge from these semantic categories.
One principle would be that languages universally categorize
things as nouns and actions as verbs. Another principle would
be that categories like noun and verb are heads of larger units
or phrases, that is, noun phrases (NPs) and verb phrases (VPs).
Another principle would be that NPs and VPs are semantically
connected through semantic relations such as agent and action.
Seeing the doggie eating would lead to the establishment of the
sentence category. Tis process of building grammatical knowl-
edge from semantics has been called semantic bootstrapping
(Pinker 1984).
How, then, does one decide between these two positions? Te
answer depends on the importance placed on supporting chil-
drens grammatical advances with observable changes in their
linguistic behavior. Te rich interpretation deals with grammati-
cal development as a logical problem. Since the complexities of
language are acquired so rapidly, it seems reasonable to assume
that linguistic principles must be at work at the very onset of
language acquisition. Researchers who study childrens actual
comprehension and production of language, however, set an
additional requirement. Claims about childrens abilities must
be supported in the way children are comprehending and pro-
ducing language. Te remainder of the present entry discusses
the evidence for grammatical knowledge during the holophras-
tic stage, based on studies of childrens comprehension and
production .
Language Comprehension
Studying the language comprehension of one-year-olds can be
dif cult since they are not capable of responding to the tasks typ-
ically used for older children and adults. Researchers, however,
have come up with several clever ways to get at least a general
idea of childrens understanding at this age. If childrens knowl-
edge of language is limited to single words, then one would pre-
dict that they should do as well or better responding to utterances
School of Pacific and Asian Studies, The Australian National
University .
. 2003 . Te Lexicon of Proto Oceanic: Te Culture and Environment
of Ancestral Oceanic Society . Vol. 2. Te Physical Environment . Pacifc
Linguistics 545. Canberra : Department of Linguistics, Research
School of Pacifc and Asian Studies, Te Australian National
University .
HOLOPHRASTIC STAGE, THE
Despite extensive individual variation, children on average
begin to produce words around the end of the frst year of life.
Initially, there is period of slow word learning of several months,
when the rate of learning is usually not much more than eight or
so words a month. Toward the end of this period of slow word
learning, two changes take place. First, word learning begins to
increase, such that the rate eventually increases to between one
and two new words a day. Second, children begin to combine
words to form their frst sentences (see two-word stage ).
Tis period of word acquisition, demarked at the start by the
frst words and at the end by the onset of word combinations,
is commonly referred to as the holophrastic stage of language
acquisition .
An alternative would be to refer to this period as the one-
word stage . Tis option would be descriptively adequate, in
that it captures the fact that childrens productions are limited
to a single word. Lois Bloom (1973), in fact, in an extensive
study of her daughters frst words, made this choice, entitling
her book One Word at a Time . Te term holophrastic , how-
ever, is making a somewhat diferent claim about childrens
language acquisition when they are producing single words.
Holophrastic can be defned as a single word expressing the
ideas of a phrase or sentence. Te term more explicitly states
that something more than single word production is taking
place. When adults produce single words, such as doggie or
eat , it is usually with the intent to express the meaning of the
individual word. When a child around age one says doggie
(or probably a little later eat), however, he or she combines
the meaning of the individual words with the communica-
tive intention of the utterance, intending a broader mean-
ing, such as there is a doggie, and I can see it or there is an
apple, and I want to eat it.
Te decision to call the period of slow word learning the one-
word stage versus the holophrastic stage is more than that of
personal taste. Te former choice refects a more conservative
view of the childs grammatical knowledge at this point is lan-
guage acquisition. Te latter choice, on the other hand, repre-
sents the position that childrens knowledge of grammar may
be greater than that directly seen in single-word productions.
Te present entry provides an overview of these two viewpoints
and discusses the nature of childrens emerging grammatical
knowledge during the time in which they produce one-word
utterances. It suggests that grammatical development begins
during this period, as refected in childrens understanding of
sentences and in some of the patterns of their single-word pro-
ductions, particularly as they near the point of making word
combinations.
Holophrastic Stage , The
365
understand the nature of multiword utterances. Tey can pro-
cess relations between at least two words in some sentences and
are aware that sentences contain both stressed lexical words
and unstressed words. Tey may not be aware of the nature of
the latter words, but they know that sentences require them to
be well formed .
Language Production
Te claim that preliminary knowledge of grammar takes place
during the holophrastic stage would be strengthened if evidence
could also be found in childrens spoken language. At frst glance,
this would seem impossible since holophrastic children are only
producing a single word at a time. Tere are, however, aspects
of childrens spoken language at this stage that, taken together,
indicate emerging grammatical knowledge as well.
It is well known that holophrastic children are not very intel-
ligible, the result of the fact that they are limited in their pho-
netic skills and are, in most cases, mixing their single-word
productions with babbling . A. Peters (1983) also pointed out
that some children are not exclusively single-word producers.
He identifed children who do not just attempt single words
but attempt to imitate and repeat longer sentences. Tese lon-
ger utterances are often hard to interpret and may be identi-
fed in many cases as some form of jargon, that is, attempts
to produce sentence-length productions without meaning.
Peters found, however, that some of these jargon productions
may be meaningful, though the meaning may be missed by
their parents. Tese productions do not represent evidence
that holophrastic children know grammar, but they support
the previously stated results on well-formed commands, that
is, that the children know that sentences consist of more than
single words.
Peters (1983) and others also drew attention to the fact that
even single-word productions are not exclusively single words.
With the advent of advanced tape recorders, researchers found
in their phonetic transcriptions that words often were pre-
ceded by brief phonetic material that was often dif cult to hear
or interpret. For example, a child who was saying book, was
actually saying something like uh book or uhm book. Tese
brief phonetic instances have been called several terms, such as
fller syllables , phonetically consistent forms , and presyntac-
tic devices . Te last term refects the opinion of many research-
ers that these fller syllables are not yet syntactic units, such
as articles or auxiliaries, but they are evidence that children
are taking notice that they exist and noting their distributional
characteristics.
A further characteristic of holophrastic speech is that sin-
gle words are produced in sequences. Bloom (1973) exam-
ined the successive single-word utterances of her daughter to
see if these sequences refected later word combinations. For
example, does a child who says eat cookie when she begins
word combinations show earlier single-word sequences, such
as eat, cookie, during the holophrastic stage? Such cases
would provide evidence of grammatical knowledge during
the holophrastic stage. Bloom found that her daughters early
sequences did not show these relations, in that each word had
its own context, that is, distinctly associated action. Later, how-
ever, toward the end of the holophrastic period, sequences on
of one word than to those of multiple words. Te results of sev-
eral studies, however, have shown that children can respond to
multiword utterances and that they have an awareness that the
words are related in a way suggestive of the relations between
words in a sentence.
E. Shipley, C. Smith, and L. Gleitman (1969) examined
young childrens responses to commands directed toward
them by their mothers while the children played. Te com-
mands were a single word, for example, ball!, two words,
for example, throw ball!, and well-formed commands, for
example, throw me the ball! Te results showed that children
around the end of the holophrastic period and beginning of
word combinations actually responded most often when they
heard the well-formed commands. While not directly show-
ing that the children understood the well-formed commands,
the results indicated that children were aware that such com-
mands met the characteristics of English sentences, while the
other two do not.
Other studies have examined more directly whether very
young children can diferentiate multiword utterances on the
basis of the specifc words in them. One potential problem in
testing children on this aspect is that they may give the appear-
ance of understanding a multiword utterance when they are only
doing what they typically would do. For example, a child who
throws a ball when told throw the ball may do so just because
that is what children typically do with a ball. To avoid this prob-
lem, J. Sachs and L. Truswell (1978) used novel combinations
that were not likely to be part of childrens experiences. Test sen-
tences included unusual commands, such as smell truck and
kiss truck. Te results indicated that young children were able
to respond correctly to such novel commands, suggesting that
they were aware of at least two-word relations in the sentences
they heard. Other research by J. Miller and colleagues (1980) has
examined the range of sentence types that are understood by
children during the holophrastic stage. Te results indicate that
this range is limited. Children did best on responding to sen-
tences that communicated an actionobject relation (e.g., kiss
the shoe) and to those that communicated a possessor to pos-
sessed relation (e.g., mamas shoe). Tey did less well on other
relations, such as agent action (e.g., make the horsey kiss).
Tese studies examined relations between lexical words in
sentences. A further question would be whether or not holo-
phrastic children are also becoming aware that sentences
contain smaller functional words as well, such as articles and
auxiliaries. Te fact that young children preferred the well-
formed commands suggests that this may be the case. N. Katz,
E. Baker, and J. Macnamara (1974) explored this issue by show-
ing children pictures that contained either a single instance of
a nonsense fgure (e.g., an odd-shaped form called zav) or a
picture of more than one instance. Te children were then asked
either show me zav or show me the zav. It was found that the
children tended to indicate the picture of the single instance in
the former case and the picture of the multiple instances in the
latter case. Te children in this study were a bit older than holo-
phrastic children, but they were not yet using articles in their
spontaneous speech.
In summary, a variety of studies on children in or around
the holophrastic stage indicate that they are beginning to
Holophrastic Stage , The
366
only by locating them within complexes of relations or structures.
One simple and infuential technique of structuralist analysis,
developed by Claude Lvi-Strauss , is the isolation of homologies .
A homology is a simple binary opposition mapped onto another
binary opposition as its structural equivalent. One model here is
phonology , where, for example, voiced/unvoiced pairs (e.g.,
b/p and d/t) may be understood as structurally equivalent with
respect to voicing. Lvi-Strauss takes homologies to manifest
important structural relations across a range of higher linguistic
levels, including, for example, narrative. Tus, we might say that
in Shakespeares Hamlet , Hamlet is to Laertes as Claudius is to
Hamlet.
In a way, this particular homology is self-evident. But why
does it work? Hamlet killed Laertes father, just as Claudius killed
Hamlets father. Moreover, Hamlet was in love with Laertes sis-
ter, just as Claudius was in love with Hamlets mother. Finally,
Hamlet was partially responsible for Laertes losing his sister,
as Claudius was responsible for Hamlet being, to some degree,
separated from his mother .
As this suggests, homologies operate through larger com-
plexes of relations. In his four-volume Introduction to the Science
of Mythology (see, for example, Lvi-Strauss 1969), Lvi-Strauss
systematically explored structural analysis beyond homologies
through the concept of transformation sets. A transformation set
is a series of multiplace structures that map onto one another.
Te mapping is defned by transformation rules, which are trig-
gered by specifable conditions, often cultural conditions. For
example, there may be a transformational relation between the
myths of two groups say, agriculturalists and fshers such that
when the myths of one group refer to earth, the parallel myths of
the other group refer to water .
One model here is morphology . For instance, in English,
the plural morpheme is pronounced s after unvoiced nonsi-
bilants (as in cats), z after voiced nonsibilants (as in dogs),
and z after sibilants (as in bushes). S , z , and z form a trans-
formation set, and the contextual trigger defning the transfor-
mation is phonological.
Clearly, the situation is more complex and the analysis less
straightforward with higher-level structures, such as literary
works. Consider Hamlet . In addition to the structures already iso-
lated, Hamlet and Ophelia may be mapped onto each other in
losing their fathers due to an older relative/older lover, in feigning
madness/going mad, in contemplating suicide/committing sui-
cide, and so on. Te establishment of such a transformation set
raises intriguing questions. For example, are the Hamlet/Ophelia
structures diferentiated by a simple gender context (i.e., male
vs. female)? If so, how is this related to the mapping of Hamlets
mother onto Laertes sister in the Hamlet/Laertes transformation
set? Lvi-Straussian analysis allows us to recognize such complex
patterns and, perhaps, begin to understand them as well .
Patrick Colm Hogan
WORK CITED
Lvi-Strauss , Claude . 1969 . Te Raw and the Cooked. Trans. John and
Doreen Weightman . New York : Harper . Tis is the frst volume of
Introduction to the Science of Mythology .
single words with a shared event began to emerge. For exam-
ple, one sequence of words involving up, neck, zip was
all produced in the context of her daughter wanting her mother
to zip up her coat.
Te last piece of evidence to suggest that early words are holo-
phrastic comes from returning to the original sources of the term.
Early diary studies many years ago by parents on their childrens
language learning recorded observations that their children
used single words with difering communicative contexts. One of
the most famous was the study by W. Leopold (193949) on his
daughter, Hildegard. Leopold noted that Hildegard used her sin-
gle words to express distinct communicative intents. Some utter-
ances were intended to show her noticing something, some to
make a request for the parent to perform some action, and others
to demand something such as a toy out of reach. Tese diferent
functions were distinguished by variations in the prosody of the
word and by the childs gestures .
Summary
Children go through a period of language acquisition of six
months or so during the second year of life when they produce
single words at a time. Tis time of acquisition has been referred
to as the holophrastic stage. Te reason is that the single words
used often communicate the idea of a sentence, that is, the
meaning of the word expressed and the communicative intent
of the utterance. Te term is suggestive that preliminary gram-
matical acquisition is taking place during this period. Research
on childrens comprehension and production during this stage
suggests that this may be the case .
David Ingram
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bloom , L . 1973 . One Word at a Time . Te Hague: Mouton .
Ingram , D . 1989 . First Language Acquisition. Cambridge : Cambridge
University Press .
Katz , N., E. Baker, and J. Macnamara . 1974 . Whats in a name? A study of
how children learn common and proper names . Child Development
45 : 469 73.
Leopold , W . 1939 49. Speech Development of a Bilingual Child: A
Linguists Record . 4 vols. Evanston, IL: Northwestern University
Press .
Miller , J., R. Chapman, M. Bronston, and J. Reichle . 1980 . Language
comprehension in sensorimotor stages V and VI . Journal of Speech
and Hearing Research 23 : 284 311.
Peters , A . 1983 . Te Units of Language Acquisition . Cambridge: Cambridge
University Press .
Pinker , S . 1984 . Language Learnability and Language Acquisition .
Cambridge: Harvard University Press .
Sachs , J., and L. Truswell. 1978 . Comprehension of two-word instruc-
tions by children in the one-word stage . Journal of Child Language
5 : 17 24.
Shipley , E., C. Smith, and L. Gleitman . 1969 . A study in the acquisition of
language: Free responses to commands . Language 45 : 322 42.
HOMOLOGIES AND TRANSFORMATION SETS
A fundamental principle of structuralism is that one cannot
understand elements in isolation. One can understand elements
Homologies and Transformation Sets
367
association of general ideas, which operates to cause the sym-
bol to be interpreted as referring to its object ( 1935a , 249). Tere
are numerous examples of symbols, from the various words and
utterances in human languages to such things as road signs.
What is crucial in the case of symbols is that there exists some
underlying convention, agreement, habit, or law that means
that invoking some symbol invokes its associated object. For
instance, a red traf c lights being symbolic of a lack of priority
at a road junction works because we have all agreed (by habit, by
convention, and by imposing traf c regulations) to use red traf c
lights this way .
Troughout his life, Peirce made numerous alterations to his
account of signs (see, for instance, Short 2004 ), but the broad
division among icons, indices, and symbols tends to fnd a place
throughout. Tere are, of course, some subtleties to Peirces
account. For instance, it is not clear that there are very many
examples of signs that are purely iconic, indexical, or symbolic
that is, which do not overlap with one or both of the other ele-
ments of the trichotomy. As an example, take a painted portrait
as a sign of the person it depicts. Tis sign is an icon in that it sig-
nifes its object in virtue of the qualities it shares with that object
the skin and hair color of the depicted person are replicated in
the painting. But, of course, many of the things that make a por-
trait a successful depiction of its sitter are due to particular con-
ventions governing paintings and how particular blocks of color
in two dimensions can stand for some subject. Tis seems to
make the painting look as though it has symbolic elements, too.
Similar considerations hold for indices such as barometers
although such signs indicate their objects in virtue of a causal
and physical connection with their object; conventions about
how we should interpret this physical connection also seem to
play a part in signifcation. Whats more, there are clear instances
of symbols that have some iconic element. Obvious examples
might include forms of writing, such as Chinese, that involve
pictograms, at least partially. Even onomatopoeic words such as
cuckoo present clear cases of symbols with a strong iconic ele-
ment the phonic qualities of the object are aped by the phonic
qualities of the word.
Peirce was aware of the various overlaps among icons, indi-
ces, and symbols, and at some point proposed to call icons and
indices with symbolic elements hypo-icons and subindices as
a way of acknowledging this. However, in any case where more
than one of the three elements is present, one will be most
prominent. Consequently, we can think of Peirces trichotomy
as dividing signs according to whether they are predominantly
iconic, indexical, or symbolic .
Te main infuence of Peirces division is in semiotics,
where his work is considered foundational. However, the icon,
index, symbol distinction has had some infuence in philoso-
phy, particularly through the work of Arthur Burks ( 1949 ), and
has even been used in such diverse areas as literary theory (see,
for example, Sherif 1989 ), flm theory (see, for example, Wollen
1969 ; see also film and language ), and musicology (see Turino
1999 ; see also music, language and ). Te use and relevance of
this distinction to linguistics are similarly diverse, but it features
most prominently in analyses of the relation between animal
communication and human language and in some expla-
nations of the evolution of language .
I
ICON, INDEX, AND SYMBOL
Te nineteenth-century American philosopher C. S. Peirce devel-
oped extensive sign theories in order to explain reference ,
meaning, communication , and cognition. One of the central
and most innovative features of his theories was the icon, index,
symbol classifcation of signs.
A crucial aspect of understanding Peirces icon, index, sym-
bol division is his account of sign structure. According to Peirce,
any instance of signifcation consists of three interrelated parts: a
sign, an object, and an interpretant. For the sake of simplicity,
we can think of the sign as the signifer, for example, a written
word or an animals footprint. Te object, on the other hand,
is whatever is signifed, for example, the object denoted by the
written word or the animal that left the print. Te interpretant is
the understanding or interpretation that the sign/object relation
generates, for example, that the word or utterance is meant to
refer to its object or that the animal track signifes the presence
of the animal that made it. Te importance of the interpretant
for Peirce is that signifcation is not a simple dyadic relationship
between sign and object: A sign signifes an object only if it can
be interpreted as such.
With this structure in mind, Peirce was interested in classi-
fying the various ways in which the sign/object relation might
generate an interpretant. In particular, he thought that a sign
might come to signify its object, and so generate an interpretant,
in three possible ways. First, a sign may be understood as signify-
ing in virtue of similarities or shared qualities between it and its
object. As Peirce says, I call a sign which stands for something
merely because it resembles it an icon ( 1935b , 362). His own
preferred examples of icons are portraits or mathematical dia-
grams indeed, he thought icons were especially important to
mathematical thought. However, we can also include examples
such as color swatches, sculptures, and so on. What is central to
iconic signifcation is that the qualities of the sign are also quali-
ties of the signifed object and that this sharing of qualities is cru-
cial in enabling the sign to signify.
Te second way in which a sign might be understood as signi-
fying is in virtue of some physical or causal connection between
it and its object. Such a sign is an index. Peirces own description
of an index is as a sign which refers to the object that it denotes
by virtue of being really efected by that object ( 1935a , 248).
Again, there are numerous and wide-ranging examples, includ-
ing demonstratives and indexical expressions, weather vanes,
barometers, fever as a sign of an underlying illness, or smoke
as a sign of fre. What is crucial to indices is that the object has
a causal efect upon the sign (as in the case of fre causing the
smoke that indicates it) or has some spatio-temporal proximity
to its sign, which can be used to aid an interpreter of the sign
to grasp that object (as in the case of pointing to some nearby
object).
Te third way in which a sign might be understood as signify-
ing is in virtue of some convention or law that connects it to its
object. Peirces own description of a symbol is as a sign which
refers to the object that it denotes by virtue of a law, usually an
Icon, Index, and Symbol
368
allows for the symbolic communication typical of human lan-
guage is never attained, and vervet monkey calls and chimpan-
zee symbol manipulation never rise above the level of indexical
communication .
Albert Atkin
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brinck , Ingar , and Peter Grdenfors . 2003 . Co-operation and communi-
cation in apes and humans . Mind and Language 18 : 484 501.
Burks , Arthur . 1949 . Icon, index and symbol . Philosophy and
Phenomenological Research 9 : 673 89.
Deacon , Terence . 1997 . Te Symbolic Species: Te Co-evolution of
Language and the Human Brain . New York : Norton .
Liszka , James Jacob . 1996 . A General Introduction to the Semiotic of
Charles S. Peirce . Bloomington : Indiana University Press .
Peirce , C. S. 1935 a. Te Collected Papers of Charles S. Peirce . Vol 2.
Cambridge : Harvard University Press .
. 1935 b. Te Collected Papers of Charles S. Peirce . Vol. 3.
Cambridge : Harvard University Press .
Seyfarth , R. M. , D. L. Cheney , and P. Marler . 1980 . Monkey responses
to three diferent alarm calls: Evidence for predator classifcation and
semantic communication . Science 210 : 801 3.
Sherif , John K. 1989 . Te Fate of Meaning: Charles Peirce Structuralism
and Literature . Princeton, NJ : Princeton University Press .
Short , Tomas . 2004 . Te development of Peirces theory of signs.
In Te Cambridge Companion to Peirce , ed. Cheryl Misak , 21440.
Cambridge : Cambridge University Press .
Turino , Tomas . 1999 . Signs of imagination, identity, and experience: A
Peircian semiotic theory for music . Ethnomusicology 43 : 221 55.
Wollen , Peter . 1969 . Signs and Meaning in Te Cinema . London : Secker
and Warburg/British Film Institute .
IDEAL SPEECH SITUATION
Tis term was coined by the German social theorist and philoso-
pher Jrgen Habermas to refer to the conditions necessary for free
and transparent communication and discussion. Te concept of
ideal speech situation plays a key part in his early formulations of
a theory of communicative action and of universal prag-
matics (Habermas 1979 , 168; 1984 ; 1987 ). In his later writings,
the term has tended to be replaced by Karl-Otto Apels notion
of an unrestricted communication community (Apel 1980 ;
Habermas 1990 , 88).
An ideal speech situation may be understood as the condi-
tions that would allow for open discussion between free and
equal participants, who strive to come to an agreement upon
any topic purely through the force of better argument. Tus, the
participants enter a discussion assuming that their ideas may be
challenged by any other participant, but that only those ideas
and arguments that are rationally formulated and supported
by relevant and persuasive evidence will survive interrogation.
Te personality, status, power, or rhetorical abilities of the per-
son holding the idea will be rendered irrelevant in the course of
debate.
Te idea of an ideal speech situation has its origins in the work
of the American pragmatist philosopher Charles Sanders Peirce .
In his philosophy of science, Peirce proposed the notion of an ideal
community of scientists. He recognized that scientifc research is
a necessarily communal enterprise. Typically, scientists work in
In explaining animal communication, the distinction is espe-
cially useful since it allows us to classify various cases of ani-
mal language without treating all such instances as uniform.
Consequently, a diverse range of animal camoufage or cases
of mimicry can be classifed as iconic instances of communica-
tion. For example, the harmless milk snakes mimicking of the
poisonous coral snakes red, black, and yellow coloring in order
to avoid predation is easily explained as an instance of iconic
communication these colors mean poisonous! As for indexical
communication, a well-discussed case is vervet monkey warning
calls (see Seyfarth, Cheney, and Marler 1980 ; see also primate
vocalizations ). In such an example, the calls are classifable
as indexical since they rely upon a causal and physical connec-
tion with particular predators in order to refer the calls are made
in response to the snakes, eagles, or leopards whose presence is
perceived. And this is all in contrast to human language, which
is predominantly symbolic and can enable communication even
if the objects referred to are not present. Ingar Brinck and Peter
Grdenfors ( 2003 ) make compelling use of the icon, index, sym-
bol trichotomy in explaining animal communication where they
discuss the role of such communication in cooperation .
Te most prominent use and interesting development of
Peirces icon, index, symbol trichotomy is Terence Deacons
( 1997 ) account of the coevolution of human language and
brains . According to that account, language evolution is to be
explained by seeing iconic, indexical, and symbolic communi-
cation and reference as related to one another in a hierarchy.
What this means is that in order to master symbolic communi-
cation, we must frst master indexical communication. And in
order to master indexical communication, we must frst master
iconic communication. For instance, a predators inability to
distinguish the milk snakes coloring from that of a coral snake
is suggestive of iconic reference it is manifest in the preda-
tors inability to distinguish one type of snake from the other.
However, this iconic communication needs to be in place in
order for the predator to take the coloring of those snakes as
an indexical signifer of the poisonous status of the snake red,
yellow, and black banding are an index of a venomous snake.
Other instances of indexical reference work in just this way. It is
because the vervet monkey sees the eagle above as being quali-
tatively similar to previously experienced eagles (that is, as an
icon of a recognized predator) that it is able to produce a warn-
ing cry (an indexical reference) when that predator is present.
Symbolic reference requires the presence of indexicals but also
requires that the indexical relationship between words/sounds
and their objects has become ingrained, habitual, and appropri-
ately interconnected with other symbols so that reference and
communication are maintained even if the stimulus to indexical
reference is lost or removed.
Once this symbolic threshold is achieved, complex relation-
ships between words develop, allowing words to signify other
words and explain the relationships that exist among them. Such
a model is useful for explaining various diferences between
cases like vervet monkey warning calls, captive chimpanzee
symbol manipulation, and human language learning in the
two former cases, the connection between sign and object is lost
when the object is absent for sustained periods. Consequently,
the habituation and interconnectedness of indexical signs that
Icon, Index, and Symbol Ideal Speech Situation
369
manipulation of one another, Habermas argues, they all presup-
pose, with a rather studied navet, that the other participants
are telling the truth and being sincere in their participation in
the conversation, and that they have the right to speak and act
as they do. In practice, these assumptions can quickly be over-
turned. However, Habermass point is that a person could not
enter into a conversation assuming that the other participants
were lying or systematically trying to deceive. Every utterance
would be treated with suspicion, and ultimately no fxed mean-
ing could be attributed to it. While some form of social interac-
tion might continue, it would not be a true conversation or what
Habermas understands as communicative action. Tat is to say,
the participants would not be seeking to reach a mutual agree-
ment (constrained only by the force of better argument). Rather,
each would be trying to manipulate the other (in what Habermas
calls strategic action; 1982 , 266) .
Andrew Edgar
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Apel , Karl-Otto . 1980 . Towards a Transformation of Philosophy .
London : Routledge and Kegan Paul .
Habermas , Jrgen . 1970 . On systematically distorted communication .
Inquiry 13 : 36075.
. 1979 . Communication and the Evolution of Society . Trans. Tomas
McCarthy. Boston : Beacon .
. 1982 . A reply to my critics. In Habermas: Critical Debates , ed.
John B. Tomas and David Held , 21983. London : Macmillan .
. 1984 . Te Teory of Communicative Action. Vol 1. Reason
and the Rationalisation of Society . Trans. Tomas McCarthy.
Cambridge : Polity .
. 1987 . Te Teory of Communicative Action. Vol. 2. Life World and
System: A Critique of Functionalist Reason . Trans. Tomas McCarthy.
Cambridge : Polity .
. 1990 . Moral Consciousness and Communicative Action . Trans.
Christian Lenhardt and Shierry Weber Nicholsen. Cambridge, MA : MIT
Press .
IDENTITY, LANGUAGE AND
Linguistic Identity
Our identities who we are are bound up with how we speak,
write, and sign. Whether or not we intend it or are even aware of
it, other people interpret clues from our use of language in order
to assign to us identity categories of all sorts, including gender,
race, ethnicity, nationality, the region, or even the precise local-
ity we come from, age or generation, sexual orientation, religion,
level of education, and that vague complex of factors that bundle
together as class. All this is in addition to and part and parcel
with the decisions they make about our intelligence, likeability,
and trustworthiness and whether or not to believe what we are
telling them.
Although self-identity has long been given a privileged role,
the identities we construct for ourselves and others are not difer-
ent in kind, only in the status we accord to them. Te gap between
the identity of an individual and of a group seems most like a true
diference of kind, with group identities more abstract than indi-
vidual ones. Brazilianness, after all, does not exist separately
from the Brazilians who possess it, except as an abstract concept.
teams, but even if they work in individual isolation, they will still
be required to submit their research results to a process of peer
review. Scientifc hypotheses are thus formulated, refned, and
fnally accepted as (provisionally) true only through a process a
collective debate, criticism, and defense. Peirce was aware that
in practice, scientifc debate falls well short of any ideal process
of rational scrutiny. Imperfections will occur in part due to prac-
tical limitations. Certain evidence may be unavailable due, for
example, to lack of suf ciently refned experimental technology
to test a hypothesis rigorously. More importantly for Habermass
use of Peirce, however, is the distorting role that hierarchies of
status and power may play within the scientifc community. Te
opinions of certain fgures within the scientifc community will
carry more weight than those of others. Te opinions of a senior
researcher trump those of a laboratory assistant. Peirces con-
cern is that open and rational debate is then being compromised
by hierarchies of power and status within the scientifc commu-
nity. Te senior researchers opinions count not because of his or
her greater rationality or insight but simply because he or she is
in a position of power. Junior scientists may feel unable to raise
their criticisms in debate or may believe that their opinions have
no place in the debate. Peirces notion of an ideal community of
scientists can then be understood, in potential at least, as a crit-
ical tool that draws attention to the defciencies of real scientifc
communication .
Habermas may, therefore, be seen to use the notion of the
ideal speech situation, particularly in his early formulations of
it, in a similarly critical manner. It encapsulates a perhaps unre-
alizable standard against which actual communication and dis-
cussion can be measured. Real communication will be distorted,
perhaps because of a lack of relevant information, perhaps
because of defciencies in the participants ability to recognize
good argument and evidence, but also, crucially, because some
will exercise power over the discussion. Power can be used to
introduce topics, to silence certain forms of criticism or sup-
press evidence, and to silence certain potential contributors.
Power can be exercised openly through, for example, threats and
intimidation. More subtly, it may be exercised through rhetorical
means so that the other participants fail to recognize that weak
arguments have been given or that relevant evidence has not
been presented. Perhaps most signifcantly, Habermas argues
that power diferentials may be so ingrained in a culture that
participants take their inferiority or superiority for granted and,
as such, do not notice the infuence of power on discussion. For
example, in a patriarchal society, women will typically have less
opportunity to raise topics in conversation or to challenge and
interrupt other (male) speakers (see gender and language ).
Such implicit and unacknowledged power is characterized as
systematically distorted communication (Habermas 1970 ) .
Te images of an ideal community of scientists and an ideal
speech situation have a utopian ring about them, suggesting per-
fect societies at the end of human history. Habermas is keen to
reject such utopian interpretations of the ideal speech situation
(Habermas 1982 , 261f). Tis becomes clearer in his later formu-
lations of the argument. Te ideal speech situation is under-
stood as a counterfactual assumption made by all participants
in conversation and discussion. Upon entering a conversation
where the participants strive for mutual agreement rather than
Ideal Speech Situation Identity, Language and
370
statistically based examination of social networks to more inter-
pretative examination of communities of practice, defned
as an aggregate of people who come together around mutual
engagement in an endeavor (Eckert and McConnell-Ginet
1992 , 464). In the course of this endeavor, shared beliefs, norms,
and ideologies emerge, including, though not limited to, linguis-
tic and communicative behavior. Tis line of research is thus
continuous with another one that has focused more directly
on the normative beliefs or ideologies by which national and
other group identities are maintained (see Verschueren 1999 ;
Blommaert 1999 ; Kroskrity 2000).
Other features of recent work on language and identity
include the view that identity is something constructed, rather
than essential, and performed, rather than possessed features
that the term identity itself tends to mask, suggesting as it does
something singular, objective, and reifed. Each of us performs
a repertoire of identities that are constantly shifting and that we
negotiate and renegotiate according to the circumstances .
Co-constructing National Identity and Language
Within these repertoires, any particular identity can become the
salient one in a given context. None inherently matters more
than the rest. However, national identity requires a separate
discussion because of its unique impact on views about what
a language is. Modern nationalism has been grounded in a
belief that the best proof of a peoples historical authenticity and
right to self-determination is the possession of a language that
is uniquely theirs. Hence, one of the frst obstacles to be over-
come in establishing a national identity is the nonexistence of a
national language.
Te nation-state myth that basic view of the world as con-
sisting naturally of autonomous states, each corresponding to an
ethnically unifed nation assumes that national languages are
a primordial reality. Dantes treatise De vulgari eloquentia (ca.
1306) lays out the process by which he claimed to discover, not
invent, the national language of a nation, Italy, that would take
fve and a half centuries to emerge politically . Tis all seems a
fction, a pretense of discovery in what will actually be Dantes
invention of an illustrious vernacular which will, in turn, cam-
oufage how much of it is actually based on his native Tuscan.
But Dantes volgare illustre became the template upon which
other modern European standard languages were modeled.
Once the national languages existed, their invention was
promptly forgotten. Te people for whom they represented
national unity inevitably came to imagine that the language had
always been there and that such dialectal diference that existed
within it was the product of recent fragmentation when, in fact, it
had preceded the unifcation by which the national language was
forged. By the early nineteenth century, this nationalist mythol-
ogy would lead to Romantic theorizations of national political
identities being grounded in a primordial sharing of language.
One of the strongest expressions was that of Johann Gottlieb
Fichte ([1808] 1968, 1901):
Te frst, original, and truly natural boundaries of states are
beyond doubt their internal boundaries. Tose who speak the
same language are joined to each other by a multitude of invis-
ible bonds by nature herself, long before any human art begins;
Yet combinations of such abstractions are what our individual
identities are made up of, and group identity frequently fnds
its most concrete manifestation in a single, symbolic individual.
Group identities nurture our individual sense of who we are but
can also smother it.
Recent work on the evolution of language has suggested that it
came about to fulfll something more than the two purposes tra-
ditionally ascribed to it, communication and representation.
Language also exists for the purpose of reading other people, in
order to discriminate useful allies from potential competitors.
sociolinguistic inquiry into identity and language is con-
cerned with the way people read each other, in two senses. First,
how are the meanings of utterances interpreted, not just follow-
ing idealized word senses and rules of syntax as recorded in
dictionaries and grammars, but in the context of who is address-
ing whom in what situation? Secondly, how are speakers them-
selves read, in the sense of the social and personal identities their
listeners construct for them based on what they say and how they
say it? Tis is a complex process because speakers output is usu-
ally shaped in part by how they have read their listeners and their
expectations. Every day, each of us repeatedly undertakes this
process of constructing our reading of the people we encounter,
in person, on the telephone, on the radio or the screen, or in writ-
ing, including on the Internet, on the basis of their language, at
least in part and in some of the media just mentioned, on the
basis of that alone .
Targeting Identity in the Analysis of Language
Modern linguistics has moved slowly but steadily toward embrac-
ing the identity function as central to language. Te impediment
has been the dominance of the traditional outlook that takes
representation alone to be essential, with even communication
relegated to a secondary place. Although signifcant develop-
ments within linguistics (surveyed in Joseph 2004 ) pushed it in
the direction of attending to identity over the course of the twen-
tieth century, a crucial prompting came from social psychology,
where one approach in particular needs to be singled out: social
identity theory , developed in the early 1970s by Henri Tajfel (see
ethnolinguistic identity ). Tis approach was novel in not
being concerned with power but in the status we give ourselves
as members of in-groups and out-groups. Tis would come
into even greater prominence in the self - categorization theory
that developed as an extension of the original model, notably in
the work of Tajfels collaborator J. C. Turner (see Turner et al.
1987 ) .
Partly under the infuence of such work, many sociolin-
guists reoriented their object of investigation. L. Milroy ( 1980 )
reported data from studies she conducted in Belfast showing
that the social class of an individual did not appear to be the
key variable allowing one to make predictions about the forms
that the person would use. Rather, the key variable was the per-
sons social network, a concept borrowed from sociology, and
defned as the informal social relationships contracted by an
individual (Milroy 1980 , 174). Where close-knit localized net-
work structures existed, there was a strong tendency to maintain
nonstandard vernacular forms of speech.
Over the next two decades, sociolinguistic investigation
of groups ideologically bound to one another shifted from
Identity, Language and
371
Fichte , J. G. [1808] 1968. Addresses to the German Nation . Trans.
R. F. Jones and G. H. Turnbull , ed. G. A. Kelly . New York : Harper Torch
Books .
Fishman , J. A. , ed. 1999. Handbook of Language and Ethnic Identity .
Oxford : Oxford University Press .
Hobsbawm , E. J. 1990 . Nations and Nationalism since 1780: Programmes,
Myth, Reality . Cambridge : Cambridge University Press .
Holmes , J. , and M. Meyerhof , eds. 2003. Te Handbook of Language and
Gender . Malden, MA, and Oxford : Blackwell .
Joseph , J. E. 2004 . Language and Identity: National, Ethnic, Religious .
Houndmills : Basingstoke; New York: Palgrave Macmillan .
Kroskrity , P. V. , ed. 2000. Regimes of Language: Ideologies, Polities, and
Identities . Santa Fe, NM : School of American Research Press .
Milroy , L. 1980 . Language and Social Networks . Oxford and New
York : Blackwell .
Renan , E. 1882 . Quest-ce quune nation? Confrence faite en Sorbonne,
le 11 mars 1882 . Paris : Calmann Lvy .
Smith , A. D. 1998 . Nationalism and Modernism: A Critical Survey
of Recent Teories of Nations and Nationalism . London and New
York: Routledge .
Tajfel , H. 1978 . Social categorization, social identity and social com-
parison. In Diferentiation between Social Groups: Studies in the
Social Psychology of Intergroup Relations , ed. H. Tajfel , 6176.
London : Academic Press .
Turner , J. C. , M. A. Hogg , P. J. Oakes , S. D. Reicher , and M. J. Wetherell .
1987 . Rediscovering the Social Group: A Self-Categorization Teory .
Oxford : Blackwell .
Verschueren , J. , ed. 1999. Language and Ideology: Selected Papers from
the 6th International Pragmatics Conference . Antwerp : International
Pragmatics Association .
IDEOLOGY AND LANGUAGE
Te concept of ideology refers broadly to the ways in which a
persons beliefs, opinions, and value systems intersect with the
wider social and political structures of the society in which he
or she lives (cf. politics of language ). Many linguists, espe-
cially those working in the traditions of critical linguistics (e.g.,
Fowler et al. 1979 ) and, more recently, critical discourse
analysis , take the view that language or more exactly, a range
of language practices are infuenced by ideology. From this
perspective, all texts, whether spoken or written, are seen as
being inexorably shaped and determined by a mosaic of political
beliefs and sociocultural activities. Te critical linguistic position
on language is, therefore, one that challenges directly the liberal
construal of texts as natural products of the free communicative
interplay between individuals in society. For critical linguists,
texts are anything but neutral or disinterested, and so it falls to
close linguistic analysis to help us understand how ideology is
embedded in language and, consequently, to become aware of
how the refexes of dominant or mainstream ideologies are
sustained through textual practices .
Although coined in the early 1800s by the French philoso-
pher Destutt de Tracy , the term ideology is normally associated
with Karl Marx , particularly with his treatise on Te German
Ideology, a project developed in 18456, but published, in
various languages and installments, from the 1930s onward (see
Marx [1933] 1965 ). Over the intervening years, the concept has
been adopted more widely (and without any necessary adher-
ence to Marxist doctrine) to refer to the belief systems that are
they understand each other and have the power of continuing to
make themselves understood more and more clearly; they belong
together and are by nature one and an inseparable whole.
Fichte was writing in order to rouse the German nation
to repel the advance of Napoleon. However, in 1870, the shoe
was on the other foot when the Franco-Prussian War led to the
German annexation of Alsace, a German-speaking province that
had been part of France for more than two centuries and whose
population was mainly loyal to France in spite of their linguistic
diference. Tis provoked a sharp turn away from the Fichtean
view on the part of French linguists, such as Ernest Renan ( 1882 ),
who formulated a new view of national identity as based not in
any primordially determining characteristic such as language
but on a shared will to be part of the same nation, together with
shared memories.
Te nation, in other words, exists in the minds of the people
who make it up. Tis is the conception that B. Anderson ([1983]
1991, 6) would return to in defning the nation as an imagined
political community. Te legacy of memories to which Renan
pointed would dominate future philosophical and academic
attempts to analyze national identity. M. Billig, a colleague and
collaborator of Tajfel, has explored how the continual acts of
imagination on which the nation depends for its existence are
reproduced (1995, 70), sometimes through purposeful deploy-
ment of national symbols but mostly through daily habits of
which we are only dimly aware. Examples include the national
fag hanging in front of the post of ce and the national symbols
on the coins and banknotes we use each day. Billig introduced
the term banal nationalism to cover the ideological habits that
enable the established nations of the West to be reproduced. In
Billigs view, an identity is to be found in the embodied habits of
social life (1995, 8), including language. A. D. Smith (e.g., 1998,
Chapter 8) has emphasized how much of the efort of nation-
alism construction is aimed at reaching back to the past in the
interest of ethno-symbolism, and this can be seen particularly
in the strong investment made by modern cultures in maintain-
ing the standard language , by which is meant a form resistant to
change, hence, harking backward (see Hobsbawm 1990 ). Every
time we attend to the fact that someone has spoken or written
in a standard or nonstandard way, we take part, usually without
realizing it, in both the national construction of our language and
the linguistic construction of our nation .
John E. Joseph
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , B. [1983] 1991. Imagined Communities: Refections on the Origin
and Spread of Nationalism . 2d ed. London and New York : Verso .
Billig , M. 1995 . Banal Nationalism . London : Sage.
Blommaert , J. , ed. 1999. Language Ideological Debates . Berlin : Mouton de
Gruyter .
Bourdieu , P. [1982] 1991. Language and Symbolic Power: Te Economy
of Linguistic Exchanges . Trans. G. Raymond and M. Adamson, ed.
J. B. Tompson . Cambridge : Polity, in association with Basil Blackwell .
Cameron , D. , and D. Kulick . 2003 . Language and Sexuality .
Cambridge : Cambridge University Press .
Eckert , P. , and S. McConnell-Ginet . 1992 . Tink practically and look
locally: Language and gender as community-based practice . Annual
Review of Anthropology 21 : 461 90.
Identity, Language and Ideology and Language
372
these ideological assumptions are transmitted surreptitiously and
are mediated through forms of language that present as natural
or common sense certain beliefs and values that may prove to
be highly contestable or dubious in their own terms.
Take as an example the following discourse event that
unfolded over a few months in the British tabloid Te Sun . Tis
popular daily voiced vehement opposition to the British gov-
ernments plans to celebrate the advent of the millennium by
the construction, at taxpayers expense, of a festival dome in
Greenwich, London . Notice how in these extracts the paper
sometimes uses italicization to enforce the common-sense sta-
tus of its position on the Millennium Experience:
Te Sun Speaks Its Mind: DUMP THE DOME, TONY! (June 17,
1997; original emphasis)
MPS, businessmen and charities yesterday backed our see-sense
campaign to axe the 800 million Millennium Exhibition planned
for Greenwich. (June 18, 1997; italics in original)
Tat dammed Dome has disaster written all over it. Te creative
director accuses the Dome secretary of acting like a dictator
who is too easily swayed by public opinion. If only he was. Maybe
then this waste of public money would be axed. For thats what
public opinion wants. (Jan. 1, 1998; italics in original)
An appeal to commonsense values of the sort displayed here
allows the paper to present its objection to the dome as a posi-
tion with which any sensible member of society could concur.
Among other things, the papers tactic is a good example of natu-
ralization (Fairclough 1992 , 678), which is the process whereby
an ideological position is presented as if it were simply part of
the natural order of things. Naturalization encourages us to align
ourselves with mainstream or dominant thinking, even when
that thinking is itself partisan, self-serving, and driven by eco-
nomic and political interests. Indeed, to demur from Te Sun s
position would be to place oneself outside the community of
notional sensible subjects who share the same set of norma-
tive values as the paper. Yet if proof were needed of the partisan
and capricious nature of such naturalized ideological positions
in discourse, consider as a footnote the following breathtaking
U-turn that appeared in the same tabloid newspaper shortly
after the publication of the previous diatribes:
Te plans for the Millennium Experience are dazzling. If it all
comes of, the Prime Ministers prediction will be correct: the
Dome will be a great advert for Britain. (Feb. 24, 1998; italics in
original)
It may have been entirely coincidental that this sudden change
in direction occurred on the same day that the papers owner
pledged 12 million worth of sponsorship to the Millennium
Dome .
A range of linguistic models have been used over the last
quarter of a century to explore the interconnections between
language and ideology. Prominent among these has been
the concept of register, which is a variety of language that
is defned according to context and use. Linguists have
noticed that in times of war and confict, in particular, spuri-
ous specialist registers of discourse are quietly disseminated
through the public sphere by infuential social and political
held either individually or collectively by groups of people and
to the social conditions that frame these systems. In Marxs
original conception, ideology is seen as an important means by
which dominant forces in society, such as royalty, the aristoc-
racy, or the bourgeoisie, can exercise power over subordinated
or subjugated groups, such as the industrial and rural proletar-
iat. His famous axiom that the ideas of the ruling class are in
every epoch the ruling ideas ([1933] 1965 , 61), along with his
observation that the ruling material force is at the same time the
ruling intellectual force, has had a profound impact on the way
contemporary linguistic research has understood discourse
in the public sphere. Ideology, and its expression in the textual
practices that shape our everyday lives, is not something that
exists in isolation as a product of free will but is instead partial
and contingent. It is something whereby, as Louis Althusser
suggests, ideas are inserted into the hierarchical arrangement of
socially and politically determined practices and rituals, which
are themselves defned by material ideological state apparatuses
(Althusser 1971 , 158). In short, in the Marxist tradition, ideology
is, most importantly, a system of beliefs that fosters consent to
social hierarchy, particularly class hierarchy. Subsequent writ-
ers infuenced by this tradition have expanded the notion from
class hierarchy (as in class or, more specifcally, capitalist ideol-
ogy; see marxism and language ), to sex (patriarchal ideol-
ogy; see gender and language ), to colonialism (colonial
ideology), and so on .
Against this theoretical backdrop, scholars researching the
interconnections between language and ideology build from the
premise that patterns of discourse are framed in a web of beliefs
and interests. A texts linguistic makeup functions to privilege cer-
tain ideological positions while downplaying others such that
the linguistic choices encoded in this or that text can be shown
to correlate with the ideological orientation of the text. Even the
minutiae of a texts construction can reveal an ideological stand-
point, and productive comparisons can be drawn between the
ways in which a particular linguistic feature is deployed across
diferent texts. For instance, the following three simple examples
difer only in terms of the main verb used:
Te senator explained the cutbacks were necessary.
Te senator claimed the cutbacks were necessary.
Te senator imagined the cutbacks were necessary.
Whereas the frst example suggests that while the cutbacks were
unavoidable, the senators actions are helpfully explanatory,
the more tenuous claimed of the second example renders the
senators attempt to justify an unpopular measure less convinc-
ing. Te third example is arguably more negative again, with the
nonfactive verb imagined suggesting that the obverse condi-
tion applies in the embedded clause, namely, that the senator is
mistaken in his belief that the cutbacks were necessary.
Another important assumption in work on ideology and lan-
guage is that the linguistic structure of a text often works silently
or invisibly to reproduce relationships of power and dominance.
In consequence, the processor of a text such as the reader of a
tabloid newspaper, for example is encouraged to see the world in
particular ways and in ways that are often aligned with the domi-
nant or mainstream ideology espoused by the paper. Crucially,
Ideology and Language
373
of criticism is about the sorts of texts that analysts choose to
subject to ideological analysis. To be blunt, if we know a text to
be ideologically problematic at the outset, then any subsequent
linguistic analysis will only confrm what we already know, and
any linguistic feature uncovered through the analysis can by
imputation be passed of as ideologically insidious. Tis deter-
ministic approach connects with the third major area of con-
cern, which is simply that studies of ideology and language tend
to be elitist. If the main purpose of the analysis is to uncover and
challenge the repressive discourse practices of powerful, inter-
ested groups, then what needs to be considered before anything
else are the efects of these practices on ordinary (nonacademic)
people. Reactions of ordinary communities to what the analysts
deem ideologically insidious discourse are rarely considered;
instead, the academic analyst comfortably assumes the perspec-
tive of those for whom the text was intended, moving seamlessly
in and out of the multiple interpretative positions of specialist
and nonspecialist alike. It is still early to say how these serious
and far-reaching criticisms will afect the ways in which scholars
investigate the widespread interconnections between ideology
and language .
Paul Simpson
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Althusser , L. 1971 . Lenin and Philosophy and Other Essays .
London : NLB .
Blommaert , J. 2005 . Discourse: A Critical Introduction . Cambridge :
Cambridge University Press .
Charteris-Black , J. 2004 . Corpus Approaches to Critical Metaphor Analysis .
Basingstoke and New York : Palgrave MacMillan .
Chilton , P. 1985 . Language and the Nuclear Arms Debate: Nukespeak
Today . London : Pinter .
Fairclough , N. 1992 . Discourse and Social Change . Cambridge : Polity
Press .
. 2001 Language and Power . 2d ed. London : Longman .
Foucault , M. 1984 . Truth and power. In Te Foucault Reader , ed.
P. Rabinow , 5175. Harmondsworth, UK : Penguin Books .
Fowler , R. 1991 . Language in the News: Discourse and Ideology in the
Press . London : Routledge .
Fowler , R., R. Hodge, G. Kress , and T. Trew. 1979 . Language and Control .
London : Routledge .
Marx , K. [with Frederick Engels]. [1933] 1965 . Te German Ideology . Ed.
and trans . S. Ryazanskaya . London : Lawrence and Wishart .
Simpson , P. 1993 . Language, Ideology and Point of View .
London : Routledge .
. 2004 . Stylistics . London : Routledge .
Stubbs , M. 1997 . Whorfs children: Critical comments on critical dis-
course analysis (CDA). In Evolving Models of Language , ed. A. Ryan
and A. Wray , 10016. Swansea : BAAL .
Widdowson , H. 1995 Discourse analysis: A critical view . Language and
Literature 4.3 : 157 72.
. 1996 . Reply to Fairclough: Discourse and interpreta-
tion: Conjectures and refutations . Language and Literature
5.1 : 57 70.
IDIOMS
Idioms hold an important place in the class making up fxed, non-
literal expressions. Tese have a perplexing characteristic: Tey
groups. In the specifc context of the widespread proliferation
of nuclear arms in the 1970s, critical linguists adopted the term
Nukespeak , in an echo of George Orwells Newspeak, to refer
to a (mis)use of register in order to mask what for the general
public were the unpalatable horrors of nuclear confict (see
Chilton 1985 ). In fact, Nukespeak still reverberates in the con-
temporary discourses of war: Collateral damage refers to the
unintentional killing of civilians and noncombatants, incon-
tinent ordnance to poorly aimed missiles, and friendly fre to
the inficting of injury or death to ones allies. While the phrase
human remains transportation pods is a heavily sanitized label
for body bags, the expression advanced marine biological sys-
tems refers rather improbably to dolphins, which, when suitably
armed, apparently make excellent seaborne weapons systems .
In addition to the exploitation of register, the strategic use
of metaphorical language has also been identifed as a mecha-
nism for sustaining and disseminating ideological dogma (see
Charteris-Black 2004 ). Paul Simpson (2004, 42) ofers the follow-
ing examples from print and broadcast coverage of the confict
in Iraq in 2003:
Te third mechanized infantry are currently clearing up parts
of the Al Mansur Saddam village area.
Te regime is fnished, but there remains some tidying up
to do.
Of cial sources described it as a mopping up operation.
Tese examples rehearse the same basic conceptual meta-
phor through three diferent linguistic realizations. Te experi-
ence of war, the target domain of the metaphor, is relayed by the
idea of cleaning, the source domain (see source and target ),
such that the metaphorical formula might be rendered thus: WAR
IS CLEANING. Te ideological signifcance of this metaphor is
that it downplays the signifcance (and indeed risk) of the con-
fict, implying that it is nothing more than a simple exercise in
sanitation, a perspective, it has to be said, that is unlikely to be
shared by military personnel on the opposing side.
Ideological standpoint in language can also be productively
explored by means of comparisons between diferent texts, espe-
cially when the texts analyzed cover the same subject matter. Of
the range of linguistic models that have been thus employed,
those from functional linguistics have proved particularly
useful as an analytic tool for investigating ideological standpoint
across diferent portrayals of the same event or experience (see
Fowler 1991 ; Simpson 1993 ).
Te investigation of ideology in language is an undeniably
important focus for the language sciences. Tat said, there have
been a number of stinging attacks on this area of study from
respected authorities (e.g., Widdowson 1995 , 1996 ; Stubbs 1997 ;
Blommaert 2005 ), which have called into question the valid-
ity of key parts of the whole endeavor. Tese criticisms have
tended to cluster around three main issues. Te frst concerns
the term ideology itself, which, even since its inception in the
work of Marx and de Tracy, has proved too broad and too vague
a concept to slot comfortably into a formal analytic framework.
Indeed, Michel Foucault has argued that the notion of ideol-
ogy is dif cult to make use of because it always stands in vir-
tual opposition to something else ( 1984 , 60). Te second type
Ideology and Language Idioms
374
Wray , Alison . 2002 . Formulaic Language and the Lexicon .
Cambridge : Cambridge University Press .
IDLE TALK AND AUTHENTICITY
Martin Heideggers magnum opus , Being and Time , advances a
hermeneutical phenomenology that is, an interpretive descrip-
tion of what it means to be Dasein (a German word for exis-
tence, which Heidegger uses to mean human being). Dasein
is not the disembodied rational soul, mind, or subject described
by traditional metaphysics and epistemology but an embodied,
worldly person. Having a world is constitutive of human exis-
tence; as Heidegger says, Daseins being is essentially being-in-
the-world ( In-der-Welt-sein ) .
To be in the world is to be thrown into an environment and a
tradition beyond your choosing and to take future-defning pos-
sibilities from it as your own or to fail to do so. Te way in which
people typically fail to make their possibilities, hence their being,
their own is by simply doing what one does or, as Heidegger
says, being ruled by the one ( das Man ). Social conformity
cannot be avoided altogether, of course, but beyond a certain
point it amounts to what he calls inauthenticity, or disowned-
ness ( Uneigentlichkeit ). Authenticity, by contrast, he describes
as forerunning resoluteness ( vorlaufende Entschlossenheit ),
which is to say, facing up (resolutely) to the concrete situation
and embracing (running up into) your death. Forerunning res-
oluteness is nothing self-destructive or suicidal, though, since
by death, Heidegger means neither the biological end nor the
biographical conclusion of a life but, rather, the constant closing
down of possibilities.
A crucial contributing factor to Daseins characteristic lapse
into inauthenticity is its unavoidable involvement in a public
language governed by anonymous norms of correctness and
propriety. In Being and Time , Heidegger refers to language
not as a formal syntactic or semantic system, but as the
concrete manifestation of discourse ( Rede ), or expressive-
communicative behavior broadly construed. Most everyday
discourse, he says, is idle talk ( Gerede ) or chitchat, generic
conversation in which we merely pass the word along, as
opposed to speaking authentically. Although the public, generic
character of a shared language contributes to our lapse into idle
talk, not all speech is inauthentic. Dasein, that is, can speak in
conformity with publicly recognized norms of correctness and
yet still speak in its own voice .
Heideggers concept of idle talk owes a large but unacknowl-
edged debt to Sren Kierkegaards account of talkativeness, or
chat, in his critique of the present age in A Literary Review
([1846] 2001 ). Heidegger even follows Kierkegaard in describing
the banalizing, conformity-inducing efect of chatter as a kind of
leveling process. But whereas Kierkegaard saw such conform-
ism and superfciality as characteristic of modern European
culture, Heidegger regarded it is essential to our being-in-the-
world. For Heidegger, that is, there could be no shared public
world at all in the absence of a relatively bland background
of established ways of conducting oneself, linguistically and
nonlinguistically .
Taylor Carman
communicate something other than what the words usually
mean. Idioms are the best and most well known representative
of this class. Other examples are speech formulas (Nice to meet
you), proverbs (While the cats away, the mice will play),
clichs (Easy does it), and expletives (For heavens sake!). Tese
difer from metaphors, which, though also nonliteral, are made up
of novel word combinations. Terminology is inconsistent, but the
term formulaic language has become standard (Wray 2002 ).
Defnitions are elusive, and lines between the categories are
not always clear. In defning formulaic expressions, it is easier
to say what they are not: Tey are not newly created phrases or
sentences made up of lexical elements (words) according to
grammatical rules. Instead, they are learned, stored, and pro-
cessed as unitary confgurations. Te meaning of the idiom She
has him eating out of her hand does not convey information
about eating or hands but, instead, refers to a complex human
relationship whereby one person is submissive to the other.
Idioms beneft from this ability to pack an aura of connotations
in a complex meaning.
Idioms have two characteristic properties: stereotyped form
and conventional meaning. Stereotyped form means that cer-
tain words appear in a particular order with a particular speech
melody. Te idiom I wouldnt want to be in his shoes, to be
well formed, must have precisely those words in that order,
with the accent on his . Changes may be introduced because
idioms are decomposable. Linguists have attempted to charac-
terize syntactic operations that may be performed on idioms,
and many psychological studies have pursued this question.
It appears that much potential variation exists, depending on
communicative need . Te complex meanings associated with
idioms contain emotional and attitudinal nuances. Te idi-
omatic expression Its a small world signals recognition of a
chance meeting in an unexpected place by two acquaintances,
refecting surprise and serendipity . In contrast, the literal
statement Its a small tree merely conveys information about
relative tree size. Te stereotyped forms and conventional
meanings of idioms are known to the native speaker but dif-
fcult for the second language learner. Idioms are learned and
processed by cognitive and neurological mechanisms diferent
from those underlying novel expressions (Van Lancker Sidtis
2006 ). Teir number is often underestimated; in eforts at com-
piling lists, no upper limit has yet been determined. Corpus
studies often utilize computer search techniques to quantify
incidence, but a human interface is necessary for identifying
idiomatic forms .
Diana Van Lancker Sidtis
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Cacciari , C. , and P. Tabossi , eds. 1993 . Idioms: Processing, Structure, and
Interpretation . Hillsdale, NJ : Lawrence Erlbaum .
Nunberg , Geofrey, I. Sag , and T. Wasow . 1994 . Idioms . Language
70 : 491 538.
Titone , Deborah , and C. Connine . 1999 . On the compositional and non-
compositional nature of idiomatic expressions . Journal of Pragmatics
31 : 1655 74.
Van Lancker Sidtis , Diana . 2006 . Where in the brain is nonliteral lan-
guage ? Metaphor and Symbol 21. 4 : 213 44.
Idioms Idle Talk and Authenticity
375
is what writers in the European ethical tradition refer to as the
problem of relevant act descriptions [see Chapter 2 of Nell
1975 ].)
Tere are some general principles that may guide ijtihd. For
example, in cases that involve punishment of an ofender, one is
to incline toward mercy (see Waines 1996 , 81). More generally,
interpretation is to some degree constrained by two sorts of
coherence coherence with the body of the law and coherence
with the views of jurists. On the other hand, jurists often disagree.
Moreover, apparent contradictions with the body of law may be
resolved by reinterpreting other parts of the law. Tus, in debat-
ing a passage P, one jurist may cite passage N against the inter-
pretation put forth by a second jurist. But this second jurist may
interpret the passage N in such a way that it fts with his or her
initial interpretation of passage P. In this way, the criterion of
coherence is often not decisive .
In many ways, ijtihd is as much an attitude as a method. Te
point is suggested by a hadth that contrasts Quranic verses that
anyone can understand with those that someone with a pure
mind can understand (Gleave 1997 , 33). Ijtihd necessarily
draws on techniques of inference (such as analogy) and princi-
ples of selection across possible inferences (such as coherence).
But it fundamentally involves sincere and open-minded work to
understand the purport of a legal or other passage .
In this respect, ijtihd may be seen as an instance of a gen-
eral pattern in Islam. Te word Islam means submission.
Submission to Allah is paired with a struggle against anything
that would inhibit that submission. A possible inhibition may be
a ftnah , or trial. When faced with such a possible inhibition, the
believer engages in a struggle, or jihad , in order to bear witness to
truth. Tat inhibition may be ones own inclination to err. In that
case, ones jihad is against that inclination. It may also be against
some violence perpetrated upon oneself or others. In that case,
jihad may take the form of battle (hence, the popular conception
of jihad as holy war) .
One may think of ijtihd as analogous to jihad, thus as a strug-
gle to establish and bear witness for truth, in this case through
interpretation. (Te words are etymologically related.) In keep-
ing with this, the Quran opposes right interpretation to ftnah
connected with false interpretation (Ali 1995 , 3.6).
Tough linked in its origins to religious attitudes and faith in
revelation, the general concept of ijtihd may not be irrelevant
to secular forms of interpretation within or outside of Muslim
hermeneutic traditions (see philology and hermeneutics ).
Indeed, the concept seems germane to a range of legal, liter-
ary, conversational, and even scientifc discourses (on the
relevance of issues in interpretation, including scriptural inter-
pretation, to science, see Lecture 4 of van Fraassen). It suggests
something about the ways in which we come to understand
these discourses when our spontaneous or automatic response
has been interrupted and we must engage in self-conscious
refection on meaning (cf. passing theories ). Indeed, its
opposition to taqld suggests partially parallel oppositions
found in infuential Western theories, such as those of Bakhtin
(see dialogism ) and Martin Heidegger (see, for example,
Heidegger 1962 , 21114) .
Patrick Colm Hogan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Blattner , William . 2006 . Heideggers Being and Time . London :
Continuum .
Carman , Taylor . 2003 . Heideggers Analytic: Interpretation, Discourse, and
Authenticity in Being and Time. Cambridge : Cambridge University
Press .
Dreyfus , Hubert L. 1991 . Being-in-the-World: A Commentary on
Heideggers Being and Time, Division I. Cambridge, MA : MIT Press .
Heidegger , Martin. 1962 . Being and Time . Trans. J. Macquarrie and
E. Robinson. New York : Harper and Row .
Kierkegaard , Sren . [1846] 2001 . A Literary Review . Trans . A. Hannay .
New York : Penguin.
Wrathall , Mark . 2005 . How to Read Heidegger . New York : Norton .
IJTIHD (INTERPRETIVE EFFORT)
Ijtihd sometimes contrasted with taqld or uncritical accep-
tance of someone elses opinion (Chaumont 1997 , 11) is a con-
cept from Muslim legal theory (see legal interpretation ). It
refers to the interpretive efort involved in understanding a point
of law. Te various schools of Muslim legal thought difer on the
degree to which they accept or advocate ijtihd .
Tere are two places in which ijtihd is likely to arise as an
issue. One is in the extension of particular judgments to new
cases. When this occurs, the common method of inference is
qiys or analogy (see Jokisch 1997 ). Take a simple example
using one hadth . (A hadth is a brief narrative of a decision or
practice, most often from Muhammad, that has been passed
down by tradition and has roughly the status of a precedent in
law.) Muhammad says that there is no tax payment required
for fewer than fve camels (Ali n.d., 215). One might narrowly
constrain the application of the hadth, not extending it beyond
camels. Alternatively, one might calculate the worth of the
camels and use that as a basis for application. A third possi-
bility involves qiys. But this is not simple. For example, fve
cars would probably not parallel fve camels. Perhaps the anal-
ogy should be based on the importance of the objects for ordi-
nary functioning in society. Tis is where interpretive efort is
required .
Te second obvious place for ijtihd is in understanding the
general meaning of texts. Te Quran suggests this, explaining of
itself that some of its verses are decisive and others are alle-
gorical (Ali 1995 , 3.6). Te mere existence of the allegorical or,
in Dawoods translation, ambiguous verses indicates the
necessity of interpretive efort in these cases. Indeed, the dif-
ference in translations suggests the importance of interpretive
efort. Moreover, once meaning is determined, interpretative
efort may be required for understanding the purpose of a pas-
sage. For example, many stories in the Quran are presented as
literally true accounts of Allahs responses to human behavior.
Tese frequently serve as warnings. But understanding these
warnings may require interpretive efort to discern the exact
nature of the actions forbidden . When someone is punished,
we must infer what construal to give that persons actions so as
to understand the point of the punishment. For example, Lots
wife was punished because she remained behind when Lot
left (Dawood 1990 , 7:83). What is her sin disbelief in a prophet,
disobedience of her husband, association with the sinful? (Tis
Ijtihd (Interpretive Effort)
376
A last infuential tradition of E-language is platonism,
revived by J. Katz and others. According to this view, language
consists of abstract objects and properties that have an existence
independent of both our mind and cultural record .
As for I-language, Chomsky ( 1986 , 21) refers to O. Jespersen,
who claimed that there is some notion of structure in the mind
of the speaker. According to Chomsky, it is a distinct system of
the mind/brain that grows in the individual from an initial state
S
0
to a stable state S
s
. Tis growth, comparable to the growth of
an organ, involves only minimal external factors, such as those
that help set the parameters that distinguish the grammars of dif-
ferent languages. Seen this way, the study of language is part of
individual cognitive psychology and, ultimately, part of human
biology. In order to counter the obvious objection that language
also involves external elements, Chomsky makes a distinction
between the faculty of language in the narrow sense (FLN) and
the faculty of language in the broad sense (FLB). Te notion of
I-language particularly applies to FLN, which has recursion as its
core property (Hauser et al . 2002) .
It is a matter of controversy whether the opposition of
E-language to I-language makes sense, since FLN, no matter how
internal, involves words and therefore externally coded conven-
tions. Even in its narrowest sense, then, language seems to inte-
grate E- and I-elements .
Jan Koster
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , Noam . 1986 . Knowledge of Language . New York : Praeger .
Hauser , Marc D., Noam Chomsky , et al. 2002 . Te faculty of lan-
guage: What is it, who has it, and how did it evolve? Science 298
(November): 1569 79.
ILLOCUTIONARY FORCE AND SENTENCE TYPES
John L. Austin ( 1962 ) and John R. Searle ( 1969 , 1975 ) distinguish
the meaning (see sense and reference ) of a sentence from
its illocutionary force. Austin and Searle conceive of the illo-
cutionary force as the act performed by the speaker with his/
her utterance. Examples of illocutionary force are expressing a
belief, asking the addressee a question, and warning or advis-
ing someone. Te illocutionary force must be intended by the
speaker, and it must be possible for the addressee to recognize
it (see also speech-acts , performative and constative ,
perlocution ).
Illocutionary force interacts with sentence grammar. Tis
short entry focuses on illocutionary consequences of the syn-
tactic sentence types declarative and interrogative, with some
reference to imperatives. (See Knig and Siemund 2007 on
the cross-linguistic usefulness of this traditional three-way
distinction.)
THE MEANING OF DECLARATIVES. In philosophical logic and
linguistic semantics , propositions traditionally represent
the content of both that -clauses ( that it is raining ) and declara-
tives ( It is raining ). Austin and Searle draw on this notion in dis-
tinguishing the meaning of a clause from its illocutionary force
according to the schema Force(proposition) . For example, the
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ali , Maulana Muhammad , ed. and trans. 1995 . Te Holy Qurn .
Columbus, OH : Ahmadiyyah Anjuman Ishaat Islam .
. N.d. A Manual of Hadith . Lahore: Ahmadiyya Anjuman Ishaat
Islam .
Chaumont , ric. 1997 . Ijtihd et histoire en islam sunnite classique selon
quelques juristes et quelques thologiens. In Gleave and Kermeli
1997 , 723.
Dawood , N. J ., ed. and trans. 1990 . Te Koran . New York : Penguin.
Gleave , Robert . 1997 . Akhbr Sh usl al-fqh and the Juristic Teory of
Ysuf b. Ahmad al Bahrn. In Gleave and Kermeli 1997 , 2445.
Gleave , R., and E. Kermeli , eds. 1997 . Islamic Law: Teory and Practice .
London : Tauris.
Heidegger , Martin . 1962 . Being and Time . Trans. John Macquarrie and
Edward Robinson. New York : Harper and Row.
Jokisch , Benjamin . 1997 . Ijtihd in Ibn Taymiyyas fatw . In Gleave
and Kermeli 1997 , 19937.
Nell , Onora . 1975 . Acting on Principle: An Essay on Kantian Ethics .
New York : Columbia University Press .
van Fraassen , Bas C. 2002 . Te Empirical Stance . New Haven, CT : Yale
University Press .
Waines , David. 1996 . An Introduction to Islam . Cambridge : Cambridge
University Press .
I-LANGUAGE AND E-LANGUAGE
In order to counter earlier misunderstandings, Noam Chomsky
( 1986 , 20f.) made a distinction between E-language and
I-language. E-language stands for externalized language and
I-language for internalized language. E-language is defned
as language independent of the properties of the mind/brain .
I-language, in contrast, is language seen as a property of an
individuals mind/brain . Te neologism mind/brain refects
Chomskys belief that theories of mental faculties, particularly
generative grammars , are ultimately about the brain at
some level of abstraction.
Tere are various ways of seeing language as external to the
human mind. Te immediate target of Chomsky ( 1986 , 20) is
language as a collection (or system) of actions or behaviors of
some sort. Tis is the behavioristic view of language that Chomsky
criticized in L. Bloomfeld and B. F. Skinner. A related notion,
defended by W. V. Quine and R. Montague, sees human languages
as analogous to formal languages. According to this conception,
languages are extensionally defned as sets of sentences or well-
formed formulas, and there is no empirical dimension to the ques-
tion which grammar is the correct one: Grammars are equally valid
variants if they generate extensionally equivalent languages.
A very common tradition of E-language sees languages as
largely existing in some mind-external cultural record. Tis tra-
dition goes back to Aristotle, was selectively revived by J. G. von
Herder and the German Romanticists, and deeply infuenced
the European structuralism of Ferdinand de Saussure and the
American structuralism of F. Boas and E. Sapir. Ludwig
Wittgensteins idea that language consists of conventional rules
constitutive of language games also fts this tradition, as does
Karl Poppers proposal that language is part of his world 3 (some
recent ideas related to E-language may be found in meaning
externalism ). Relativistically interpreted, this tradition is at
variance with universal grammar .
I-Language and E-Language Illocutionary Force and Sentence Types
377
Tus, the semantic distinction between declaratives and
interrogatives is useful for distinguishing typical and untypical
uses of declaratives and for distinguishing typical and untypical
uses of interrogatives. Te fexibility for diferent uses for both of
these sentence types may suggest a purely pragmatic assignment
of illocutionary force to propositions and partitions .
GRAMMAR AFFECTS ILLOCUTIONARY FORCE. Yet, there is evi-
dence that grammar interferes with the assignment of illocution-
ary force:
First, a purely pragmatic account of declarative questions leads
to the expectation that a declarative questions like John is at
home? shares the illocutionary force of a yes/no question like
Is John at home? In the declarative question, the speaker would
indicate by the intonation that he/she does not commit to p , and
so the next likely interpretation would be that the speaker is won-
dering whether p is true and seeks the help of the addressee in
this regard. In fact, however, declarative questions strongly difer
from yes/no questions.
(3) [Telephone conversation] (4) [Steve calls Ann on the phone]
Ann: Hang on, Ill ask John Ann (picks up phone): Hello?
Steve: John is at home? a. Steve: # John is at home?
cf. b. Steve: Is John at home?
Te declarative question John is at home? can be asked only
where the speaker can assume that the addressee believes that
John is at home (Gunlogson 2001 ). Tis is the case in (3) but
not in (4a), where the declarative question is accordingly infe-
licitous (symbolized by #; see also felicity conditions ).
Interrogatives do not carry this strong requirement; compare
(4b). Tus, even though declaratives cannot have a statement
interpretation hardwired into them, they seem to have some-
thing more fexible hardwired into them that goes beyond the
propositional content: A declarative sentence seems to either
commit the speaker to p (statement) or assume a commitment of
the addressee to p (declarative question) (see Gunlogson 2001 ).
Second, root interrogatives also seem to have an illocutionary
element hardwired into them. If Steve and Ann share ignorance
of things mechanic, Steve may say (5a) to Ann, but not (5b).
(5) Steve to Ann (both ignorant about car problems):
a. Te carburetor of my car is broken. I wonder whether the
repairs will be expensive.
b. Te carburetor of my car is broken. # Will the repairs be
expensive?
Tis is unexpected on the view that a semantic partition is fexibly
augmented with illocutionary force by pragmatic inferences. In
(5a), the matrix predicate I wonder conveys that Steve is inter-
ested in the answer but does not have an expectation toward Ann
to provide it. Te felicity of (5a) shows that this is not excluded
for lack of conversational relevance or for other pragmatic rea-
sons. Why cannot the unembedded partition Will the repairs be
expensive? in (5b) have a pragmatically inferred similar illocu-
tionary interpretation? It seems that some hardwired element
of illocutionary interpretation here leads to an expectation of
an answer from the addressee and thus to the inappropriate-
ness in (5b). A promising hypothesis about such a hardwired
two aspects of the statement It is raining can be rendered as Te
speaker expresses that (s)he believes (that it is raining) .
THE MEANING OF INTERROGATIVES. Semantic theory (see, for
example, Karttunen 1977 ; Groenendijk and Stokhof 1982 ) has
since developed meanings for syntactic interrogatives that cover
unembedded interrogatives like (1a), (2a) and embedded inter-
rogatives like (1b), (2b). Tese meanings (difering in detail
between authors) may be paraphrased as in (1c), (2c). Te mean-
ings assigned by Jeroen Groenendijk and Martin Stokhof ( 1982 )
are called partitions .
(1) a. Is it raining? (2) a. Whom did he invite?
b. Mary wonders b. Mary wonders
[whether it is raining] [whom he invited]
c. the true one of c. the true one(s) of
{that it is raining, {that he invited Bill,
that it is not raining} that he invited Jane, }
Note that the partitions in (1c) and (2c) do not contain a compo-
nent of illocutionary force. Tey are also the meanings of embed-
ded interrogatives like (1b) and (2b), and embedded clauses are
standardly assumed not to have illocutionary force.
Tus, the meaning of a declarative is a proposition and the
meaning of an interrogative is a partition. Let us consider how
illocutionary force of two kinds may be added to these mean-
ings: i) illocutionary force as a statement: the speaker expresses
that (s)he believes something to be true, and ii) illocutionary force
as a question: the speaker wants to learn from the addressee the
truth in an as-yet open issue. Let us begin by assuming that the
choice between statement and question is pragmatically inferred
rather than grammatically triggered .
THE ILLOCUTIONARY FORCE OF DECLARATIVES. Te meaning of a
declarative ( It is raining) as a proposition ( that it is raining ) makes
it likely that the speaker intends a statement interpretation of his/
her utterance ( the speaker conveys that [s]he believes that it is rain-
ing), perhaps in connection with the maxim of quality in Grice
( 1975 ), which requires such truthfulness (see also cooperative
principle ). If we assume that this statement interpretation is not
hardwired into declaratives, we correctly allow declarative ques-
tions , as in It is raining? Here, a declarative sentence with ques-
tion intonation is used as a question. Unlike in statements, the
speaker does not convey that he/she believes that it is raining .
THE ILLOCUTIONARY FORCE OF INTERROGATIVES. No single state
of afairs (ordinary proposition) is ofered in the interrogative
meaning (the true one among , see [1c] and [2c]), and so
there is no specifc content for an interpretation as a statement.
Te interrogative meaning makes a question interpretation
particularly likely, in which the speaker wants the addressee
to pick the true one among the possibilities. If we assume that
this illocutionary force is not hardwired into root interrogatives,
we correctly allow the fexibility for untypical uses. In mono-
logical questions, such as in a lecture, the speaker later pro-
vides the answer him-/herself: Who was behind these reforms?
Everything points towards Bismarck (Brandt et al. 1992 ). In rhe-
torical questions like Am I your servant? the true one among
the given possibilities is to be inferred and need not be stated .
Illocutionary Force and Sentence Types
378
can be approximated in terms of a commitment by a salient indi-
vidual that is expressed or assumed. Sentence mood (such as
imperative vs. indicative) plays a further crucial role in condi-
tioning the illocutionary interpretation of a clause .
Hubert Truckenbrodt
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Austin , John L. 1962 . How To Do Tings with Words . Oxford : Clarendon .
Brandt , Margareta , Marga Reis , Inger Rosengren , and Ilse Zimmermann .
1992 . Satztyp, Satzmodus, und Illokution. In Satz und Illokution 1 ,
ed. Inger Rosengren , 190. Tbingen : Niemeyer .
Grice , H. P. 1975 . Logic and conversation. In Syntax and Semantics
3: Speech Acts , ed. P. Cole and J. Morgan , 4158. New York: Academic
Press .
Groenendijk , Jeroen , and Martin Stokhof . 1982 . Semantic analysis of
WH-complements . Linguistics and Philosophy 5 : 175233.
Gunlogson , Christine. 2001 . True to form: Rising and falling declaratives
as questions in English . Ph.d. diss., University of California at Santa
Cruz.
Heycock , Caroline 2006 . Embedded root phenomena. In Te Blackwell
Companion to Syntax. Vol. 2 . Ed. M. Everaert and H. van Riemsdijk ,
174209. Oxford : Blackwell .
Karttunen , Lauri 1977 . Syntax and semantics of questions . Linguistics
and Philosophy 1 : 3 44.
Knig , Ekkehard , and Peter Siemund . 2007 . Speech act distinctions in
grammar. In Language Typology and Syntactic Description. Vol. 1 . Ed.
T. Shopen , 276324. Cambridge : Cambridge University Press .
Portner , Paul . 1997 . Te semantics of mood, complementation, and
conversational force . Natural Language Semantics 5 : 167 212.
Schwager , Johanna Magdalena . 2005 . Interpreting imperatives . Ph.d.
diss., University of Frankfurt am Main.
Searle , John R. 1969 . Speech Acts. An Essay in the Philosophy of Language .
Cambridge : Cambridge University Press .
. 1975 . A taxonomy of illocutionary acts. In Language, Mind,
and Knowledge , ed . K. Gunderson, 34469. Minneapolis: University
of Minnesota Press. Repr. John R. Searle, Expression and Meaning
( Cambridge : Cambridge University Press , 1979), 129.
Truckenbrodt , Hubert . 2004 . Zur Strukturbedeutung von
Interrogativstzen . Linguistische Berichte 199 : 313 50.
. 2006 a. On the semantic motivation of syntactic verb movement
to C in German . Teoretical Linguistics 32: 257 306.
. 2006 b. Replies to the comments by Grtner, Plunze and
Zimmermann, Portner, Potts, Reis, and Zaeferer . Teoretical
Linguistics 32 : 387 410.
IMAGE SCHEMA
A foundational concept in cognitive linguistics , image
schemas are associated most closely with the work of Mark
Johnson ( 1987 ) and his collaborator George Lakof ( 1987 ). Image
schemas are thought to play a key role in the acquisition, struc-
ture, and use of language; a central tenet of cognitive linguistics
is that the perceptual interactions and motor engagements with
the physical world shape the way we think. An image schema
can be defned as a condensed representation of bodily motor
experience for purposes of mapping spatial structure onto con-
ceptual structure. According to Johnson, these patterns emerge
as meaningful structures based on our bodily movements
through space, our manipulations of objects, and our percep-
tual interactions (1987, 29). Jean Mandler aptly likens image
requirement is again that a commitment by a salient individual
is assumed or expressed. In the case of interrogatives, a commit-
ment to (1c)/(2c) is a commitment to the true element(s) of the
partition, that is, knowledge of the correct answer (whatever it
may be). Te assumption of such a commitment in root interrog-
atives is compatible with the examples discussed here. In rhetor-
ical questions, knowledge of the correct answer is assumed of the
speaker and perhaps of the addressee. In monological questions,
knowledge of the correct answer can be assumed of the speaker,
who will later provide this answer. In standard questions,
knowledge of the correct answer would then be assumed of the
addressee. Ten (5b) would be infelicitous because knowledge of
the correct answer could not be assumed of either the addressee
or the speaker (see Truckenbrodt 2004 , 2006a , 2006b ).
Tird, German declaratives (syntactically V2-clauses) can
replace embedded dass -(that)-clauses under a range of atti-
tude verbs, including sagen (say) and glauben (believe) and
excluding bezweifeln , (doubt) and leugnen (deny): Maria
glaubt/*bezweifelt, Hans ist zu Hause, Mary believes/*doubts,
Hans is at home. A semantic restriction seems to be in efect that
relates declaratives to a salient individuals beliefs (or a similar
domain) Marias beliefs in the preceding example, the beliefs
of the speaker or the addressee in unembedded use (statements
and declarative questions) (Truckenbrodt 2006a , 2006b ). More
generally, there is a class of syntactic phenomena ( root phe-
nomena) that occur in unembedded clauses, as well as embed-
ded under the verbs believe , say , and other predicates that have
been characterized as assertive (see discussion and references
in Heycock 2006 ). It is possible that root phenomena are phe-
nomena that trigger a semantic requirement of a commitment
by a salient individual. Such a requirement might lead to the
observed restrictions on embedded use. Te requirement would
crucially contribute to illocutionary force in unembedded use .
Finally, verbal mood (indicative, subjunctive, imperative) is a
grammatical category that interacts with illocutionary force. Te
imperative seems to be interpreted deontically (Schwager 2005 ),
that is, along the lines of You should with strength varying
between a demand or request ( Open the window! ) to invitation
( Have another piece of cake! ) or even a wish ( Have a good break! ).
A strong interaction of verbal mood with illocutionary force is
this: Have fnished eating by 12:30! cannot be a statement to the
efect that the addressee has fnished eating by 12:30 (rather than
by 12:45, as the addressee may mistakenly believe). Nor can it
be a question that wants the addressee to clarify whether in fact
the addressee has fnished eating by 12:30. More generally, the
deontic interpretation of the imperative cannot be replaced with
an epistemic interpretation (i.e., one that negotiates knowledge),
while the indicative in declaratives and interrogatives typically
leads to epistemic (statement and question) interpretations. (See
Portner 1997 on indicative and subjunctive verbal mood.)
In sum, the grammatical distinction between declaratives
and interrogatives leads to a semantic distinction between a
proposition and a partition. Tis distinction is useful for a frst
understanding of the typical and untypical illocutionary force of
declaratives and of interrogatives. Illocutionary force may in part
be assigned pragmatically. However, the assignment seems also
to be subject to a grammatically triggered interpretation in root
clauses. For declaratives and interrogatives, these interpretations
Illocutionary Force and Sentence Types Image Schema
379
motion; and caused motion emerge as preverbal, perceptually
based conceptual primitives underlying cognitive development
and language acquisition. Chris Sinha and K. Jensen de Lopez
take a slightly diferent tack in emphasizing the social-cultural
and artifact dimensions of the acquisition of locative preposi-
tions in and under in Danish-acquiring, English-acquiring, and
Zapotec-acquiring children. Tey conclude that the sociocul-
tural environment plays a greater role in cognitive development
than previously thought. Image schemas may in fact be distrib-
uted throughout the local environment, with considerable dif-
ferences appearing among cultures with very diferent material
makeups .
Finally, image schema theory is a central pillar of the neural
theory of language project initiated by Jerome Feldman ( 2006 ),
Lakof ( 1987 ), Srini Naryanan, and others. A crucial component
of this neural computational model is the execution or x-schema
protocol for representing human actions, which include models
for enacting drop -schema that simulate the neural computa-
tional activity involved. For instance, a computational represen-
tation for distinguishing between verbs such as push and shove
invokes the slide x-schema but difers in the microdetails of
body-part movement and acceleration .
Todd Oakley
WORKS CITED AND SUGGESTION FOR FURTHER READING
Cienki , Alan . 1998 . Straight: An image schema and its transformations .
Cognitive Linguistics 9.2 : 107 49.
Feldman , Jerome . 2006 . From Molecule to Metaphor . Cambridge,
MA : MIT Press .
Gibbs , Raymond W. 2005 . Te psychological status of image schemas.
In Hampe 2005 , 11335.
Hampe , Beate , ed. 2005 . From Meaning to Perception: Image Schemas
in Cognitive Linguistics . Berlin : Mouton De Gruyter . An edited collec-
tion of essays presenting a wide range of views on the nature of image
schemas.
Johnson , Mark . 1987 . Te Body in the Mind . Chicago: University of
Chicago Press. Tis book is considered the locus classicus of image
schema theory.
Lakof , George . 1987 . Women, Fire, and Dangerous Tings .
Chicago : University of Chicago Press .
Mandler , Jean . 2004 . Te Foundations of Mind . New York : Oxford
University Press .
Matsumoto , Yo . 1995 . From attribution/purpose to cause: Image
schema and grammaticalization of some cause markers in Japanese.
In Lexical and Syntactic Constructions and the Construction of
Meaning , ed. Marolijn Verspoor , K. Dong , and E. Sweetser , 287307.
Amsterdam : John Benjamins .
Sinha , Chris , and K. Jensen de Lopez . 2000 . Language, culture, and the
embodiment of spatial cognition . Cognitive Linguistics 11.1/2 : 17 41.
Turner , Mark . 1991 . Reading Minds . Princeton, NJ : Princeton University
Press .
Verspoor , Marolijn . 1995 . Predicate adjuncts and subjectifcation.
In Lexical and Syntactic Constructions and the Construction of
Meaning , ed. Marolijn Verspoor , K. Dong , and E. Sweetser , 43349.
Amsterdam : John Benjamins .
IMPLICATIONAL UNIVERSALS
typological universals seek to capture the limits of gram-
matical variation, that is, the extent to which particular traits
schemas to the representations one is left with when one has
forgotten most of the details (2004, 7981) .
Lack of specifcity and content makes image schemas highly
fexible preconceptual and primitive patterns used for reasoning
in an array of contexts (Johnson 1987 , 30). A partial inventory of
image schemas is as follows: container; balance; blockage;
counterforce; restraint removal; source-path-goal; link;
center-periphery; verticality; scale; part-whole; support .
Since the publication of the infuential works of Johnson and
Lakof, the notion of image schemas has been theorized, inves-
tigated, and applied in multiple domains of inquiry. Within
cognitive linguistics, the notion of an image schema has taken
center stage in research in semantic and grammatical analy-
sis, psycholinguistics , cognitive development, and neuro-
computational modeling, to name the most prominent areas of
activity in the language sciences. Controversy has ensued over
the defnition of and criteria for positing something as an image
schema, over its status as conscious or unconscious representa-
tions, and even over its status vis--vis individual and social cul-
tural cognition.
Image Schemas Among Cognitive Linguists
Image schemas play an important role in studies focusing on
polysemy of words and constructions, semantic change ,
and text analysis. Studies of words and constructions include
Alan Cienkis ( 1998 ) comparison of the metaphoric projec-
tions of straight and prjamo in English and Russian. Likewise,
image schemas fgure prominently in arguments about seman-
tic change. For instance, Marolijn Verspoor (1995) argues that
semantic change preserves image schematic structure, whereas
Yo Matsumoto ( 1995 ) challenges a strong version of that hypoth-
esis. Much text analysis and literary criticism inspired by cog-
nitive linguistics specify image schemas as that which make
linguistic innovation possible. Literary and textual criticism
within the image schema tradition is represented most promi-
nently by Mark Turner ( 1991 ) .
Raymond W. Gibbs ( 2005 ) has conducted extensive psycho-
linguistic experiments designed to demonstrate that image sche-
mas organize not only experience but also semantic structure and
usage. His early experiments support the claim that image sche-
mas are psychologically real and imply that they are enduring
mental representations , while his later experiments refne
this view, suggesting that they are real but not representational
structures per se. Rather, they are emergent structures con-
tinuously created on the fy as part of human beings dynamic
simulations of actions and situations. In short, Gibbs argues that
the psychological reality of image schemas emerges from con-
tinuous interaction in a three-dimensional world and not from
their being prestored representations in long-term memory
(2005, 132).
Image schemas fgure prominently in some developmen-
tal accounts of the acquisition of concepts and language. Two
notable lines of research in this area include Mandler ( 2004 )
and Sinha and Jensen de Lopez ( 2000 ). Mandler has argued that
image schemas arise from a process of perceptual meaning anal-
ysis. Human neonates and infants appear to be engaging in forms
of perceptual analysis of space and motion, such that primitive
schemas for goal-path; linked path; self-motion; animate
Image Schema Implicational Universals
380
also favor SOV order over non-SOV order. However, given the
data in (2b), the strength of the implicational universal with SOV
as the antecedent and postpositions as the consequent is greater
(95%) than the converse universal with postpositions as the ante-
cedent and SOV order as the consequent (91%) .
Te implicational universals in (1) and (2) are simple ones,
as they specify a dependency between only two traits. Relations
between several traits are captured by means of complex univer-
sals that may involve conjunctions or disjunctions of traits within
the antecedent and/or the consequent, as depicted in (3).
(3) a. X (Y & Z)
b. (X or Y) Z
Individual implicational universals may be combined into
chains or hierarchies, such that the implicatum (or conclusion)
of the frst universal is the implicans (or premise) of the second,
the implicatum of the second is the implicans of the third, and so
on. Since representing chains of implicational universals as such,
that is, as in (4a), is quite cumbersome, they tend to be depicted
in the form of a hierarchy, as in (4b).
(4) a. ((A B) & B C) & C D
b. D > C > B > A
It is important to note that the distributions captured in the form
of these typological hierarchies defne a frequency cline, such
that the phenomenon in D is more frequent than that in C, which
in turn is more frequent than in B, and so on. Tis follows from
the fact that if any term involved in the hierarchy is present, all
the terms to the left of it on the chain must also be present. And if
any term involved in the hierarchy is absent, all the terms to the
right of it must also be absent.
Typological hierarchies thus make very strong statements
about the possible distribution of properties across languages
and their overall frequency of occurrence. Consequently, much
typological research has been aimed at elaborating such hier-
archies, be it with respect to segmental inventories (see, e.g.,
the sonority hierarchy of Hopper 1976 , 196), morpho-syntactic
encoding (see, e.g., the complement deranking-argument
hierarchy of Cristofaro 2003 , 131), or behavioral properties
(see, e.g., Keenan and Comries 1977 noun phrase accessi-
bility hierarchy ) .
While statistical implicational universals are the dominant
means of expressing typological generalizations by typologists,
it must be pointed out that there is a long-standing controversy
over whether such universals do indeed capture signifcant
relationships between aspects of linguistic structure or merely
incidental relationships resulting from historical accident
(see, e.g., Maslova 2000 ; Bakker 2008). A technique of critically
evaluating such universals has been recently developed by
Elena Maslova ( 2003 ), and an excellent account of typological
universals in general is provided by William Croft ( 2003 , 4969,
1228) .
Anna Siewierska
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bakker , Dik. 2008 . LINFER: Inferring implications from the WALS data-
base. STUF 61 .3: 18698.
or features of languages may covary, both within languages
and across languages. Typological universals may be classifed
along several dimensions. Te most important distinction is
between substantive and implicational universals. Te for-
mer deal with a single property and refect what occurs in all
languages (e.g., all languages have vowels). Since such unre-
stricted universals tell us little about variation, they are con-
sidered to be of moderate interest . For the study of variation,
implicational universals are of considerably greater interest
since they refect not only the existence of variation but, cru-
cially, the constraints imposed on it.
Implicational universals relate two (or more) logically inde-
pendent features or properties and apply to some subset of
languages for which the given features or properties obtain.
Implicational universals may be absolute or statistical .
Absolute implicational universals hold for all languages; that is,
they specify that if a language has feature A, then it will have
feature B. By contrast, statistical implicational universals spec-
ify relationships that hold only at a certain level of probability.
Statistical implicational universals take the form, If a language
has feature A, then with greater than chance frequency it will
have feature B. Examples of the two types of implicational uni-
versals are given in (1a) and (2a), respectively, and the distribu-
tion of features that each universal refects is presented in (1b)
and (2b) (in the form of a tetrachoric table). (Te data depicted
in (2b), where the numerals refect the number of languages
with the specifed features, are taken from Greenberg ( 1963 ,
Appendix II.)
(1) a. If the demonstrative follows the head noun, then the rela-
tive clause also follows the head noun
b. DemN NDem
RelN +
NRel + +
(2) a. If a language has basic subject-object-verb (SOV) order,
than with greater than chance frequency it will have
postpositions
b. SOV non-SOV
prep 5 73
post 97 10
Note that the fve prepositional SOV languages are counterex-
amples to the universal. Although many absolute implicational
universals have been posited in the typological literature, most,
particularly of the simple kind (see the following), have turned
out to be in fact statistical.
.
Needless to say, these may be assessed
from the point of view of their relative strength (the number of
overall languages considered, the number of languages display-
ing the features in question, the number of exceptions to the
universals) and their validity in relation to criteria such as those
presented in Stassen ( 1985 , 201).
Implicational universals may be monodirectional or bidirec-
tional. Te absolute implicational universal in (1) is monodi-
rectional since NRel order is compatible with both DemN and
NDem. Consequently, while NDem order entails NRel, NRel
order does not entail NDem. Te statistical implicational univer-
sal in (2), on the other hand, is bidirectional; not only does SOV
order favor postpositions over prepositions, but postpositions
Implicational Universals
381
fnd a native speaker apparently referring to brass as vagavai. I
might conclude that vagavai does not mean gold. But I might also
conclude that the speaker does not know the diference between
brass and gold, that the speaker (e.g., a young child) is mistaken
about the meaning of vagavai, and so on .
Of course, Quines claims are more signifcant and less intui-
tive than this example suggests. To get a better idea of what is at
stake in the indeterminacy of translation, we might distinguish
two levels or perhaps two poles of uncertainty and indetermi-
nacy. Te frst is global. Tis is the level at which we may say that
any meaning may be mapped onto any translation, given enough
manipulation of the rest of the system. For example, if we are will-
ing to revise enough of our beliefs about the world, we can main-
tain a translation of the French word lapin (rabbit) as stone.
Of course, it wont be easy. But it is possible. We expect any word
to have a number of misuses even by fuent speakers. Typically,
however, the correct uses overwhelmingly outnumber the incor-
rect uses. We may have to change our assumed proportions in the
case of lapin . For example, we might infer that French speakers
are correct in their use of lapin only when they point to rabbit
sculptures or when, at a distance, they mistake a stone for a rab-
bit to put the matter in our commonsense, lapin = rabbit idiom.
Chomskys infuential criticisms of Quines views on inde-
terminacy most obviously concern this pole. Chomsky rightly
points out that all theories are underdetermined by evidence
(1980, 15). One consequence is that we rely on other criteria for
adjudication, such as simplicity. It is presumably simpler (or, in
another terminology, less ad hoc) to assume that fuent speakers
err at similar rates for all frequently used common nouns than to
assume that lapin is exceptional in this regard. Tus, one might
conclude that there is uncertainty here. However, the uncertainty
is adjudicable (at least within limits), and there is no reason to
conclude that this uncertainty implies indeterminacy.
Te other level or pole is highly local. It is limited to difer-
ences that we cannot formulate or, more generally, that we do
not encode (i.e., roughly, make into information that may be
processed cognitively) in particular cases. For instance, one of
Quines central examples is that of a feld linguist encounter-
ing gavagai. Te feld linguist takes the term to mean rab-
bit. However, the available evidence is equally consistent with
undetached rabbit part, rabbit stage, and so on. Here, one
might argue that this is really best understood as a case of global
uncertainty/indeterminacy, which we have already consid-
ered. Specifcally, given the structure of human perception and
human memory, it seems very unlikely that speakers of other
languages would not encode part and stage , thus distinguishing
rabbit, undetached rabbit part, and rabbit stage in some way. But
that is merely a problem with the particular example. Tere are
clear cases where one language involves distinctions that speak-
ers of another language do not encode. For instance, a particular
group may not distinguish between gold and brass or, better
still, fools gold. Examples such as this may suggest that there is a
sort of indeterminacy at the subencoding level across languages,
or even within languages. For example, when I use the word gold ,
I have no sense of diferent types of gold. In this sense, relative to
jewelers, my use of gold is indeterminate among those diferent
types. On the other hand, one might reasonably contend that this
is not really indeterminacy at all, since there is presumably some
Cristofaro , Sonia . 2003 . Subordination . Oxford Studies in Typology and
Linguistic Teory . Oxford : Oxford University Press .
Croft , William. 2003 . Typology and Universals . 2d ed.
Cambridge : Cambridge University Press .
Greenberg , Joseph. 1963 . Some universals of grammar with particu-
lar reference to the order of meaningful elements. In Universals of
Human Language , ed. Joseph Greenberg , 73113. Cambridge, MA:
MIT Press.
Hopper , Joan Bybee . 1976 . An Introduction to Natural-Generative
Phonology . New York : Academic Press .
Keenan , Edward , and Bernard Com rie. 1977 . Noun phrase accessibility
and universal grammar . Linguistic Inquiry 8 : 63 99.
Maslova , Elena . 2000 . A dynamic approach to the verifcation of distri-
butional universals . Linguistic Typology 3/4 : 307 33.
. 2003 . A case for implicational universals . Linguistic Typology
7.1 : 101 8.
Stassen , Leon 1985 . Comparison and Universal Grammar . Oxford : Basil
Blackwell .
INDETERMINACY OF TRANSLATION
In Word and Object ( 1960 ), one of the most infuential books in
the history of American philosophy, W. V. O. Quine presents an
argument along the following lines. One may always formulate
mutually exclusive hypotheses regarding the meaning of a lin-
guistic item such that there is no fact as to which hypothesis is
correct. Put diferently, there are always divergent ways of defn-
ing a given word such that no one of these ways is correct. Tis
highly infuential, and highly controversial, indeterminacy of
translation thesis extends Quines argument against analyt-
icity (the view that some statements are true simply due to the
meaning of the terms see Quine 1961 ; see also meaning and
belief ) and points toward his conception of ontological relativ-
ity (that reference is not absolute, but relative to a coordinate
system [Quine 1969 , 48], that Factuality is internal to our
theory of nature [Quine 1981 , 23]) .
To explore Quines idea, we might begin with a simple dis-
tinction between decidability criteria and demarcation crite-
ria. A decidability criterion allows us to chose among diferent
hypotheses regarding a words meaning (see word meaning ).
A demarcation criterion distinguishes in principle between the
validity of two such hypotheses. A demarcation criterion presum-
ably isolates a factual diference; a decidability criterion, frst of
all, isolates an evidential diference (or, more broadly, a distinc-
tion by the lights of scientifc method; thus, it may incorporate
simplicity or other adjudicative criteria). Quine argues that there
is no strict decidability criterion for meaning. We may refer to
this as uncertainty. But he also maintains that for at least some
set of mutually exclusive hypotheses, there is no demarcation
criterion either. Te latter is indeterminacy proper .
Quines arguments are bound up with his holism, the
view that hypotheses of any sort, including hypotheses regard-
ing meaning, are part of larger complexes of belief and are not
understandable outside those complexes. Crucially, this means
that apparent counterevidence regarding one part of the whole
need not have simple, local consequences. Such counterevi-
dence may be accommodated by alterations elsewhere in the
whole. Suppose I am a feld linguist investigating a new language.
I encounter the word vagavai , which I take to mean gold. I then
Indeterminacy of Translation
382
deictic use (see deixis ), but indexicals have many other uses.
I is used demonstratively when (1) is written near a picture of
Truman with an arrow pointing at the picture . It is used ana-
phorically in (2) as opposed to (3):
(2) Truman believes I am president.
(3) Truman believes that I am president.
It refers to Truman in (2), the speaker in (3). Indexicals are often
used pseudodeictically in novels. When Gore Vidal wrote (4) in
Live From Golgotha ,
(4) I am Timothy,
he was not referring to himself. Finally, indexicals are often used
non-referentially, as in
(5) Je means I in French.
(6) Many a car is such that it needs gas .
In a semantic use like (5), it is crucial that I be used with the frst
person meaning we have been focusing on; if it means iodine,
(5) is false. Sentence (6) illustrates the quantifcational use, with
pronouns bound by quantifers like variables in quantifica-
tion theory .
Indexicals present a problem for formal theorists because
assigning extensions and intensions (functions from possible
worlds to extensions) to indexicals cannot adequately represent
their role in determining the truth conditions of sentences (see
truth conditional semantics ). David Kaplans ([1977]
1989 ) seminal solution was to represent the meanings of indexi-
cals by assigning them characters (functions from contexts to
intensions). For I, Kaplan had in mind the function satisfy-
ing the following condition: i(c) is the intension whose value in
any world w is the speaker uttering I in c . Tus, the value of
i(c) in any context in which Teodore Roosevelt used I is the
constant function tr (w) whose value is Teodore Roosevelt for
every world. Te value of i(c) in any context in which Franklin
Roosevelt used I is the constant function fdr (w) whose value
is Franklin Roosevelt for every world.
Te character theory of meaning has many of the dif culties
of referential theories generally. For example, I and the per-
son identical to me have the same character, but not the same
meaning ( Freges problem ). Te character function for he would
seem to be undefned when it is used in After Santa came down
the chimney, he left the presents; Santa does not exist ( Russells
problem ). Quantifcational and semantic uses are nonreferential
for diferent reasons. Another dif culty is the Enterprise problem .
Suppose Mary points at the bow of a big ship and says Tat is
the Enterprise and then points at the stern and says Tat is not
the Enterprise . Te character function for that would assign it
the same intension in both these contexts the one whose value
at every world is the Enterprise . Yet Mary is not contradicting
herself the way she would if the same pointing gesture accompa-
nied both her utterances. Te essential indexical problem is that
sentences like (2) and (7) may difer in truth value, as in cases of
amnesia or delusion.
(7) Truman believes that Truman is president.
On the standard analysis, S believes p is true if and only if S
stands in the belief relation to the proposition expressed by
psychological fact about the degree of vagueness or ambigu-
ity of my usage at particular times and places.
Tese arguments, then, point toward global, though still adju-
dicable uncertainty along with local, subencoding indeterminacy
in the limited sense of vagueness or ambiguity. Both are signif-
cant and both are highlighted by Quines discussions. However,
neither is radical indeterminacy.
On the other hand, to make these arguments is to go against a
range of Quines other views for example, his behaviorism and
his commitment to ontological relativity. Like all theories, Quines
own theories operate holistically. Even if he were to accept some
version of the preceding arguments, he could accommodate
these arguments by alterations elsewhere in the system. Tus,
despite such arguments, radical indeterminacy remains at least
a continuing and important philosophical challenge .
Patrick Colm Hogan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , Noam . 1980 . Rules and Representations . New York : Columbia
University Press .
Fllesdal , Dagfnn , ed. 2001 . Philosophy of Quine. Vol. 3. Indeterminacy of
Translation . New York : Garland .
Glock , Hans-Johann . 2003 . Quine and Davidson on Language, Tought
and Reality . Cambridge : Cambridge University Press .
Quine , W. V . 1960 . Word and Object . Cambridge, MA : MIT Press .
. 1961 . From a Logical Point of View: Logico-Philosophical Essays .
2d ed. New York : Harper and Row .
. 1969 . Ontological Relativity and Other Essays . New York : Columbia
University Press .
. 1981 . Teories and Tings . Cambridge : Harvard University Press .
INDEXICALS
An indexical is a linguistic expression whose reference (exten-
sion) may vary from one context of use to another even when
used with the same meaning ( sense ) and with respect to the
same possible world or story. An example is the pronoun I ,
which I use to refer to myself and you to yourself. Sentence (1)
itself is indexical: true when used by Harry Truman in 1951 but
not when used by Churchill.
(1) I am president.
Tensed verbs ( am , was ) are indexical, as are many adjectives
( foreign , local ), common nouns ( enemy , neighbor ), and adverbs
( yesterday , recently ).
Indexicals are often defned simply as terms with diferent
extensions in diferent contexts. But plane has diferent exten-
sions (airplanes versus wood planes) because it is ambiguous
rather than indexical. I too is ambiguous, meaning one (1) or
iodine as well as me. But it is used with the very same mean-
ing when you and I use it to refer to ourselves. Furthermore, while
the 34th president refers to Dwight Eisenhower when describ-
ing the actual world, it refers to Adlai Stevenson when describ-
ing a hypothetical case in which he won in 1954. Te referent of
an indexical varies even when speakers are describing the same
possible world .
It is often said that the meaning of I is given by the rule that
its referent in any context of use is the speaker. Tis holds for the
Indexicals
383
varieties of language, varieties that did not ofer the same social,
cultural, and political rewards as others .
Te central insight here is indeed the parallelism between
linguistic variation and social diferentiation, between linguis-
tic diferences and social hierarchies and forms of stratifcation
often organized around the dynamics of prestige and stigma.
Tis insight has been developed by a number of scholars in soci-
olinguistics and related felds of study (Irvine 1989 ). I frst give a
brief survey of the development of this topic in sociolinguistics
and then turn to a discussion of two authors: Pierre Bourdieu
and Dell Hymes .
Inequality in Sociolinguistics
Bernsteins thesis about elaborated and restricted codes became
a deeply controversial one, though often for the wrong reasons. It
coincided with the emergence of modern sociolinguistics, a dis-
cipline that saw itself initially as devoted to the study of linguistic
variation and distribution as a horizontal exercise: an exercise
in mapping linguistic varieties over fragments of a population,
based on an assumption of the fundamental equivalence of every
linguistic variety. Tus, Bernsteins vertical image of stratifed
variation variation that comes with diferential value attribu-
tion was countered by William Labov ( 1972 ) and others, who
argued that a linguistic variety such as black American English
should not be seen as bad English (which Labov chose to equate
with Bernsteins restricted code) but as a complex, sophisticated
code used by virtuoso speakers. Labov, like Bernstein, started
from an awareness of real inequalities in language: Te black
American English variety was stigmatized in U.S. education and
its speakers were often negatively categorized. But Labovs eforts
were aimed at demonstrating the intrinsic linguistic equiva-
lence of these diferent varieties, whereas Bernstein addressed
the extrinsic attributional, ideological nonequivalence of
the varieties. Seen from a historical distance, both eforts were
connected by a joint concern for how particular language vari-
eties disenfranchized speakers, not because of their intrinsic
inferiority but because of social and political perceptions of the
varieties and their speakers. In other words, both were addressing
a language-ideological efect in which linguistically equivalent
varieties indexed socially unequal features and categories .
Similar ideas were to be found in the work of many of the early
sociolinguists. John Gumperz ( 1982 ) stressed the diferent ways
in which intercultural misunderstandings were not a result of
participants intentions but were an efect of small diferences
in linguistic and communicative structures in a particular con-
text of diferent contextualization cues. In his work as well, we
encounter an awareness of relativity : Diferent language varieties
can fulfll the same functions, but the perception of these func-
tions by people not sharing the contextual conventions of a par-
ticular variety can difer. Here again, we encounter the tension
between intrinsic equivalence and extrinsic nonequivalence.
From a diferent vantage point, sociolinguists concerned
with language policy stressed the fact that the political
institutionalization of language in multilingual societies (see
bilingualism and multilingualism ) usually involved a
stratifcation based not on intrinsic superiority of one language
but on prestige hierarchies and ideological images of society.
J. Fishman ( 1974 ) presented studies in which former colonial
p. But if propositions are intensions, then the complement in
(2) expresses the same proposition as the one in (7): the proposi-
tion true in any world if and only if Truman is president.
Fregean theories solve the essential indexical problem by tak-
ing indexicals to express a distinctive type of mode of presentation
( concept or mental representation) and taking proposi-
tions to be structured entities consisting of modes. Ten, (2)
entails that Truman believes the proposition whose subject is his
own self-concept, whereas (7) entails that Truman believes
the proposition whose subject is the concept of Truman. Marys
two utterances are not contradictory because they express prop-
ositions with diferent modes of presentation of the Enterprise . A
dif culty for Fregean theories is to explain why I is not ambigu-
ous if it is used by diferent speakers to express diferent con-
cepts. Another is to account for nonreferential uses, in which I is
not used to express the speakers self-concept .
Wayne A. Davis
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Braun , D. 1996 . Demonstratives and Teir linguistic meanings . Nos
30 : 145 73.
Burks , A. W. 1949 . Icon, index, and symbol . Philosophy and
Phenomenological Research 9 : 673 89.
Evans , G. 1980 . Pronouns . Linguistic Inquiry 11 : 337 62.
Forbes , G. 1987 . Indexicals and intensionality: A Fregean perspective .
Philosophical Review 96 : 3 31.
Frege , G. [1918] 1977 . Toughts. In Logical Investigations , ed.
P. T. Geach , 130. Oxford : Basil Blackwell .
Gale , R. M. 1967 . Indexical signs, egocentric particulars, and token-
refexive words. In Encyclopedia of Philosophy. Vol. 4. Ed. P. Edwards ,
1514. New York : Macmillan.
Kaplan , D. [1977] 1989 . Demonstratives. In Temes from Kaplan , ed.
J. Almog , J. Perry , and H. Wettstein , 481563. Oxford : Oxford University
Press .
Knne , W. 1997 . First person propositions: A Fregean account.
In Direct Reference, Indexicality, and Propositional Attitudes , ed.
W. Knne , A. Newen , and M. Anduschus , 4968. Stanford, CA : CSLI
Publications.
Nunberg , G. 1993 . Indexicality and deixis . Linguistics and Philosophy
16 : 1 43
Perry , J. 1979 . Te problem of the essential indexical . Nos 13 : 3 21.
INEQUALITY, LINGUISTIC AND COMMUNICATIVE
Te idea that linguistic variation goes hand in hand with linguis-
tic inequality was brought on stage by Basil Bernstein and started
infuencing the development of the new discipline of sociolin-
guistics , which became a discipline devoted to the study of the
unequal and nonrandom distribution of linguistic resources in
society. Bernstein ( 1971 ), in an attempt at understanding class
diferences in the education system, distinguished between two
codes, one elaborate and characteristic of middle-class chil-
dren, another restricted and characteristic of working-class
and minority children. Te elaborate code was the normative
one: It was privileged by the teachers and used as the yardstick,
not only for good language but also for wider behavioral and
cognitive assessments of pupils. Bernstein demonstrated that
speaking diferent varieties of language meant speaking unequal
Inequality, Linguistic and Communicative
384
a constant reenactment of communicative practices dependent
on these exclusive resources (Bourdieus analyses of academic
discourse are telling in this respect; see Bourdieu, Passeron, and
de Saint Martin 1994 ).
It is at this point that the contribution of Hymes ( 1980 , 1996 )
comes into view. He draws on a long anthropological tradition of
seeing language in context and use speech and when lan-
guage is seen from this practical angle, diversity and inequality
are the rule rather than the exception. Whereas Bernsteins and
Bourdieus refections on language remained largely abstract
and general, Hymess approach is soundly empirical. Hymes
starts from registers , repertoires, and genres rather than
languages, and such practical linguistic and communicative
instruments are performed in contexts where they take actual
shapes: conversations, stories, lectures, and so on. Hymes
himself focused on narrative , and he observed that one
form of inequality in our society has to do with rights to use
narrative, with whose narratives are admitted to have cog-
nitive function (1980, 126). Te reason is that conventions for
producing narratives are culturally and socially sensitive, and
particular contexts in society require particular types of narra-
tive. People can be very good storytellers in their neighborhood
but inefective ones in front of a judge in court, because the
stylistic and genre conventions for those particular events are
fundamentally diferent. Tus, the capacity to be recognized as
a competent speaker or an articulate storyteller, lecturer,
or conversationalist is a socially sensitive phenomenon on
which social forces of diferential value attribution operate. For
Hymes, this dialectic between real capacity and expected per-
formance defnes voice, and one can be voiceless for many
reasons (Blommaert 2005 ) .
Te approaches of Bourdieu and Hymes both present a fully
developed theory of inequality in the feld of language and com-
munication. From their work, we can see how the production of
meaning is a regulated, regimented process in which the speak-
ers choice is never unlimited and in which the efects of his/her
words are judged by others using social and ideological yard-
sticks. Te advantage of Hymess approach is its strongly devel-
oped empirical dimension, which ofers a range of opportunities
for applied research.
Evaluation
Despite its implicit focus on inequality, sociolinguistics privi-
leges diference and variation as its objects and avoids explicit
analyses of how such variation converts into social inequality.
Tis is an efect of the descriptive bias in sociolinguistics, as well
as of the hesitant and ambivalent relationship between socio-
linguistics and social theory (Williams 1992 ). Many studies of
variation, consequently, consider diferences merely in terms of
spread and distribution and see the power and inequality rela-
tionships among language varieties in simplifed terms. Te fact
that speaking with a stigmatized accent is not merely a matter
of distribution but a matter of social opportunities, determined
by the way in which others rank such accents in an ideologically
informed hierarchy, could be more central to sociolinguistic
refections.
Tis is a critical insight and eminently applicable to many
sociolinguistic and discursive phenomena in the contemporary
as well as local languages were lifted ideologically to the status
of prestige language by means of language policies favoring a
standardized , tightly controlled, and preferably ethnically
neutral language; C. Eastman ( 1983 ) described the diferent
stages of language planning, demonstrating how language vari-
eties could be turned into institutionalized, power-emanating
elements of social structure .
A central ingredient of every form of language planning is the
construction of a written, orthographically standardized variety
of language. Scholars in the feld of literacy also devoted atten-
tion to the ways in which literacy may introduce stratifcation in
languages and communities and often becomes an opportunity
as well as an obstacle defning peoples social mobility trajec-
tories (Street 1995 ; Kress 2000 ). Te creation of a literate norm
almost invariably means that access to the prestige variety of
the language comes to be controlled by the education system,
which functions as a very efective flter on social mobility .
State of Affairs
In all of the work discussed so far, there is an awareness of
inequality, notwithstanding that the emphasis would be on dif-
ference and variation, rather than on the power and inequality
dimensions of diference and variation. Other scholars developed
full-blown theories of linguistic and communicative inequality,
and two stand out: Bourdieu and Hymes.
Bourdieu ( 1991 ) emphasized the ways in which language is
part of the superstructural apparatus of society; as a nonmate-
rial resource, it can nevertheless be imbued with an economic
value: social or cultural symbolic capital. As in hard economic
sectors, the value of diferent resources is diferent and can fuc-
tuate, and not every resource can be traded for another one; the
terms of exchange are unpredictable. Bourdieus work on lan-
guage ftted into his larger program of analyzing the economies
of taste, culture, and ideas from within a generalized materialism
and with the aim of redefning (and empirically substantiating)
the notion of social class. In his view, the traditional distinc-
tion between material and immaterial resources in society was
unjustifed since the same forces of diferential value allocation
appeared to operate on both: Class is both a material and an
immaterial thing, and language diferences do play a role in this
value allocation. Consequently, the stability of class as a material
complex is also there in the immaterial aspects: A lack of real capi-
tal is often paralleled by a lack of symbolic capital, and social hier-
archies play as much into the symbolic diacritics as they do into
the material ones. (See also field and habitus, linguistic.)
It is not dif cult to see the similarities between the project
developed by Bourdieu and the (more limited) one presented
by Bernstein. Both treat language not just as an opportunity a
positive feature of humans but also as a constraint , as some-
thing inherently limited and limiting. And rather than enabling
people to perform particular acts, language also restricts them
and prevents them from performing other acts. Underlying this
idea of language as constraint is a model of society as nonegali-
tarian and stratifed, with relatively stable strata, some of which
are tightly controlled as to access while others are more demo-
cratically accessible. Particular language resources are required
to move into the more controlled strata, and membership of
these strata for example, elite, professional milieux requires
Inequality, Linguistic and Communicative
385
recently, brain measures. Te major behavioral measures rely on
infants ability to associate a behavioral response with stimuli.
Behavioral measures examine either their responses to new
stimuli after habituation or their attention (e.g., eye gaze, head
turn) to certain types speech or language stimuli. For example,
High-Amplitude Sucking and Conditioned Headturn are both
habituation paradigms. Researchers using these paradigms
habituate the infant to the stimulus, change the stimulus, and
then measure whether the infant responds to the change. Other
procedures, namely, the Headturn Preference Procedure and the
Intermodal Preferential Looking Paradigm, measure how long
infants attend to stimuli by measuring either headturns toward
attention-getting lights or gazes to pictures/video accompanying
the stimuli. Biological techniques measure reactions to speech,
most frequently by measuring infant heart rate. More recent
research has turned to brain measures, including electrophysi-
cal responses such as event related-potentials (ERP), electroen-
cephalogram (EEG), and neuroimaging techniques such as
fMRI (functional magnetic resonance imaging) to examine early
speech and language perception. For an overview of methods,
see Jusczyk ( 1997 ); Mills, Cofey-Corina, and Neville ( 1997 ) .
Using experimental methods, researchers have investigated
a wide range of issues concerning infants understanding of
speech and language, including responses to individual speech
contrasts, diferences in speakers, sensitivity to the prosody of
the language, lexical learning , and ability to detect gram-
matical patterns. (See also speech perception in infants. )
Tis research has shown that fetuses can hear during the third
trimester of development (although the sound is low-pass fl-
tered by the amniotic fuid and the uterine walls) and will recog-
nize recurrent patterns (e.g., rhymes ) from the mothers speech
(DeCasper et al. 1994 ). Neonates prefer their mothers voices
to others (DeCasper and Fifer 1980 ). In addition, neonates and
infants prefer to listen to infant-directed speech (speech with
fewer words per utterance, more repetition, slower articula-
tion, and greater prosodic swings) over adult-directed speech
(Fernald 1985 ; Cooper and Aslin 1990 )
Two-month-old infants can distinguish their native language
from other languages on the basis of prosodic characteristics; by
six months infants are sensitive to their native language stress
patterns and by nine months to native language phonotactics .
Infants use prosodic information to group words and are sen-
sitive to prosodic units that mark both clauses (by 46 months)
and phrases (by 9 months). As prosodic units often correspond to
grammatical units, this ability may give infants additional infor-
mation about grammatical structure. Tere is also evidence that
prosody can help infants remember information about speech
(Jusczyk 1997 ) .
Work on the perception of speech contrasts has shown that
from birth, infants can discriminate both consonants and vowels
that contrast in any of the worlds languages, even if the sounds
are not found in the infants native language (Werker and Curtin
2005 ; Kuhl 2000 ). Tis ability holds even if the sounds are pro-
duced by diferent speakers of diferent ages. Work has also
investigated the acoustic cues infants use to detect diferences,
as well as the nature of categorical perception. Results suggest
that infants can be sensitive to the same range of acoustic cues
as adults, including the efects of coarticulation (where sounds
world. Emergent applied studies have shown its relevance,
for example, in the feld of asylum seekers narratives (Maryns
2006 ), literacy in education (Collins and Blot 2003 ), and social
work (Hall, Slembrouck, and Sarangi 2006 ) .
Jan Blommaert
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bernstein , B. 1971 . Class, Codes and Control. Vol. 1. Teoretical Studies
Towards a Sociology of Language . London : Routledge and Kegan Paul .
Blommaert , J. 2005 . Discourse: A Critical Introduction . Cambridge :
Cambridge University Press.
Bourdieu , P. 1991 . Language and Symbolic Power . Cambridge,
UK : Polity.
Bourdieu , P. J. C. Passeron , and M. de Saint Martin . 1994 . Academic
Discourse: Linguistic Misunderstanding and Professorial Power .
Stanford, CA : Stanford University Press.
Collins , J. , and R. Blot . 2003 . Literacy and Literacies . Cambridge :
Cambridge University Press.
Eastman , C. 1983 . Language Planning: An Introduction . San Francisco :
Chandler and Sharp.
Fishman , J., ed. 1974 . Advances in Language Planning . Te Hague :
Mouton.
Gumperz , J. 1982 . Discourse Strategies . Cambridge : Cambridge University
Press.
Hall , C., S. Slembrouck , and S. Sarangi . 2006 . Language Practices in Social
Work . London : Routledge.
Hymes , D. 1980 . Language in Education: Ethnolinguistic Essays .
Washington, DC : Center for Applied Linguistics.
. 1996 . Ethnography, Linguistics, Narrative Inequality: Toward an
Understanding of Voice . London : Taylor and Francis.
Irvine , J. 1989 . When talk isnt cheap: Language and political economy .
American Ethnologist 12 : 738 48.
Kress , G. 2000 . Early Spelling: Between Convention and Creativity .
London : Routledge.
Labov , W. 1972 . Language in the Inner City . Philadelphia : University of
Pennsylvania Press.
Maryns , K. 2006 . Te Asylum Speaker: Language in the Belgian Asylum
Procedure . Manchester : St Jerome.
Street , B. 1995 . Social Literacies . London : Longman.
Williams , G. 1992 . Sociolinguistics: A Sociological Critique .
London : Routledge.
INFANTILE RESPONSES TO LANGUAGE
Infant responses to speech and language have been used to study
both early communicative development and early language per-
ception in infants. Observations of interactions between infants
and caregivers have shown that infants under three months of
age become quiet in response to speech. Infants as young as
six weeks engage in interactions or protoconversations , or turn-
taking in gaze and/or vocalizations with caregivers (Bruner 1975 ;
Trevarthen 1974 ). Between four and six months, infants respond
to speech with vocalizations (see babbling ), and by six to nine
months they begin initiating interactions, including games such
as peek-a-boo .
Experimental investigations into infant responses to speech
have yielded information about infants abilities to decode lan-
guage well before they begin to produce language themselves.
Numerous experimental methods have been employed to exam-
ine their perception, including behavioral, biological, and, more
Infantile Responses to Language
386
a compound noun (Oscar and Elmo are running) are matched
with intransitive actions. Tus, infants are able to both track
grammatical information and use word order to determine rela-
tionships in English (Hirsh-Pasek and Golinkof 1996 ) .
More work is needed to examine how infants from diferent
backgrounds, in particular infants from non-Western cultures,
non-Indo-European languages, and bilingual infants, perceive
speech and language. Existing work with bilingual infants sug-
gest that exposure to more than one language helps infants
maintain categorical contrasts found in both languages (Werker
and Tees 1984 ) and that lateralization can occur independently
for each language to which an infant is exposed (Conboy and
Mills 2006 ; see bilingualism, neurobiology of ). Research
is just beginning to explore the relationship between early speech
perception and later language development. Early results sug-
gest that some aspects of speech perception are related to later
vocabulary development (e.g., Newman et al. 2006 ).
One area of debate concerns protoconversations (turn-
taking in gaze and/or vocalizations), whether they are driven
primarily by the caregivers or the infant takes an active role in
shaping them. Research suggests an active role for infants in
co-constructing interactions (e.g., Trevarthen, Kokkinaki, and
Fiamenghi, Jr. 1999 ). Another long-standing debate resulting
from experimental work concerns whether infants responses
result from a specialized speech-processing system or from
general auditory processing plus categorization abilities. Most
researchers currently argue that the ability to discriminate
speech contrasts is domain-general with some innate perceptual
biases (Kuhl 2000 ). A further current debate parallels an ongo-
ing debate in cognitive science concerning whether infants use
statistical regularities alone or create abstract rules from statis-
tical information (Safran 2003 ). Evidence from other domains
(e.g., production) suggests that infants can abstract rules, and
most linguists would argue for rule creation rather than pure
statistical learning (Marcus et al. 1999 ), while many connec-
tionists would argue that rule-like efects are artifacts of stable
connections. Tey argue that learning is based on the specifc
stimuli and is possible without the abstraction of rules (Conway
and Christiansen 2006 ) .
Lynn Santelmann
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bruner , Jerome S. 1975 . From communication to language: A psycho-
logical perspective . Cognition 3 : 255 87.
Conboy , Barbara T. , and Debra L. Mills . 2006 . Two languages, one devel-
oping brain: Event-related potentials to words in bilingual toddlers .
Developmental Science 9.1 : F1 F12.
Conway , Christopher M. , and Morten H. Christiansen . 2006 . Statistical
learning within and between modalities: Pitting abstract against stim-
ulus-specifc representations .. Psychological Science 17 : 905 12.
Cooper , R. P. , and R. N. Aslin . 1990 . Preference for infant-directed
speech in the frst month after birth . Child Development 61 : 1584 95.
DeCasper , A. J. , and W. P. Fifer . 1980 . Of human bonding: Newborns
prefer their mothers voices . Science 208 : 1174 6.
DeCasper , Anthony J., Jean-Pierre Lecanuet, Marie-Claire Busnel,
Carolyn Granier-Deferre , and Roselyne Maugeais . 1994 . Fetal reac-
tions to recurrent maternal speech . Infant Behavior and Development
17 : 159 64.
in a word afect the articulation of each other) and the multiple
acoustic cues that can indicate the identity of a phonetic con-
trast. Te ability to discriminate phonetic contrasts from native
and unfamiliar languages is maintained until 6 months, but by
1012 months, infants only discriminate contrasts that are pho-
nologically contrastive in the target language. Tis shift from
language-general to language-specifc discrimination occurs
frst for vowels (by 68 months) and later for consonants (1012
months) (Werker and Curtin 2005 ) .
Infants perception of speech sounds shifts as they begin
to learn words. Infants younger than 8 months are sensitive to
phonetic diferences, while infants approaching one year ignore
allophonic variation and attend only to phonological diferences
found in their native language(s). At the same time, young word
learners (14-month-olds) do not attend to some phonetic detail
(e.g., dih vs. bih) when in a word-learning task (Stager and
Werker 1997 ), even though younger infants can discriminate
these contrasts in perception tasks and older infants attend to
these diferences in word-learning tasks. Tus, the experience
learning language infuences how and when contrasts are per-
ceived (Kuhl 2000 ).
Infants also attend to statistical regularities in the speech
stream. Tese abilities have been shown in the perception of
phonetic segments, in word segmentation, and grammatical
patterns (Safran 2003 ). Attention to statistical regularities can
help infants extract familiar patterns and track relationships
within a speech stream and, thus, may help infants uncover
basic structural relations in sentences as they move into the
acquisition of grammar. Infants sensitivity to grammar has
been demonstrated both behaviorally and with brain measures.
It begins with sensitivity to the phonetic form of function words
that mark grammatical relations. English-learning infants of
1011 months demonstrate sensitivity to the phonetic proper-
ties of functor words and can use them to help segment nouns.
French-learning and German-learning infants show even earlier
sensitivity to functor words, segmenting them by 78 months of
age. By 16 months, infants can distinguish passages in which
English functors are properly or improperly ordered (Shady,
Jusczyk, and Gerken 1998 ). By 18 months, infants are beginning
to be able to track grammatical relations, such as that which
occurs between is and - ing .
Infants acquiring both English and German are able to per-
form these tasks, though their ability to do so is infuenced by
language-specifc factors (Santelmann and Jusczyk 1998 ; Hhle
et al. 2006 ). English infants are sensitive to the distance between
the morphemes, whereas German infants are sensitive to the
type of constituent. In addition, by 17 to 18 months, infants are
able to use the presence of nouns in a sentence to help decode
verb meanings. Tis result is most likely due to the structure of
the two languages: English allows few elements between auxil-
iary and main verbs (mostly adverbs, or in questions, subjects).
Te structure of German (with nonfnite verbs occurring at the
end of the sentence) requires objects and adverbs to occur in
between the auxiliary and main verbs. It appears that infants
in the two languages are sensitive to these distributional prop-
erties in the syntax and are using these patterns to help orga-
nize the speech stream. Sentences with two nouns (Oscar chases
Elmo) are matched with transitive actions, while sentences with
Infantile Responses to Language
387
or salience, will have certain repercussions on its linguistic
realization; in particular, it will infuence grammatical choices
(most prominently word order patterns but also, for instance,
voice ), prosodic choices (choice of intonation contours and,
especially, placement of sentential stress ), and lexical choices
(such as defniteness, ellipsis , pronominalization, and use of
specifc particles).
Modern notions of information structure can be traced back
to the Prague School work on functional sentence perspective , as
summarized most accessibly in Firbas ( 1992 ). J. Firbas developed
the idea that sentence elements have varying degrees of commu-
nicative dynamism (CD), depending on the extent to which they
carry the message forward. In the unmarked case (the basic
distribution), degrees of CD will be refected in the linear order-
ing of elements, with sentences starting with the element carry-
ing the lowest CD (defned as the theme), followed by a gradual
rise in CD. It should be noted that there is a relationship between
the degree of CD of an element and its status in terms of given-
new (see the following) .
Given-New Structure versus Theme-Rheme Structure
Two dimensions, which may be to some extent correlated but
should be kept conceptually distinct, are important here. On
the one hand, the degree of givenness of a piece of information
refects the extent to which an element can be treated as in some
way recoverable from what precedes or can be assumed to be
present in the hearers consciousness; its thematicity, on the
other hand, refects the extent to which it represents what the
message is about. As an illustration of how these dimensions
infuence linguistic choices, consider the following example:
(1) A: What happened to Mary? Is she still single?
B: Well, Mary married John last year, but shes already
divorced him.
In Bs response, the noun phrase (NP) Mary in the frst sen-
tence (or, more accurately, the piece of referential informa-
tion represented by the NP Mary ) is given information, as it
was mentioned in the immediately preceding turn; John , on
the other hand, is new information, as it cannot be assumed to
be in the hearers consciousness. In Bs second sentence, both
Mary and John are now given information, which explains why
they can be pronominalized as she and him , respectively. At
the same time, it could be argued that both sentences are more
about Mary than about John and that Mary is, therefore, the
theme of both sentences, the remaining information being rhe-
matic (i.e., providing additional information about the theme).
Although both levels are functionally independent, they do
appear to correlate strongly, in that themes tend to consist of
given information.
Example (1) also illustrates some tendencies with regard to
word order and prosody. First of all, new information tends to
occur toward the right of the sentence, whereas given informa-
tion tends to be initial (in English, therefore, being an subject-
verb-object (SVO) language, there is a strong correlation between
givenness and grammatical subjecthood, as shown by Mary in
Bs contribution). Secondly, thematic information tends to be
sentence-initial. In fact, some linguists argue that the theme
is initial by defnition (e.g., Halliday 1994 ). Tirdly, the main
Fernald , A. 1985 . Four-month-old infants prefer to listen to motherese .
Infant Behavior and Development 8 : 181 95.
Hirsh-Pasek , Kathy , and Roberta M. Golinkof . 1996 . Te Origins of
Grammar: Evidence from Early Language Comprehension . Cambridge,
MA : MIT Press . Overview of studies examining infants sensitivity to
grammar, including description of methodologies .
Hhle , B., M. Schmitz, L. M. Santelmann , and J. Weissenborn . 2006 .
Te recognition of discontinuous verbal dependencies by German
19-month-olds: Evidence for lexical and structural infuences on
childrens early processing capacities . Language Learning and
Development 2.4 : 277 300.
Jusczyk , P. W . 1997 . Te Development of Speech Perception . Cambridge,
MA : Bradford Books . Overview of studies examining infants sensitivity
to grammar, including description of methodologies .
Kuhl , Patricia K. 2000 . A new view of language acquisition . Proceedings
of the National Academy of Science 97 : 11850 7. Overview of historical
positions and framework incorporating both innate biases and statisti-
cal learning.
Marcus , G. F., S. Vijayan, S. Bandi Rao , and P. M. Vishton . 1999 . Rule
learning by seven-month-old infants . Science 283 : 77 80.
Mills , Debra L., Sharon Cofey-Corina , and Helen J. Neville . 1997 .
Language comprehension and cerebral specialization from 13 to 20
months . Developmental Neuropsychology 13 : 397 445.
Newman , Rochelle S., Nan Bernstein Ratner, Ann Marie Jusczyk, Peter W.
Jusczyk , and Kathy A. Dow . 2006 . Infants early ability to segment the
conversational speech signal predicts later language development: A
retrospective analysis . Developmental Psychology 42 : 643 55.
Safran , Jenny R. 2003 . Statistical language learning: Mechanisms and
constraints . Current Directions in Psychological Science 12.4 : 1101 4.
Santelmann , Lynn M. , and Peter W. Jusczyk . 1998 . Sensitivity to discon-
tinuous dependencies in language learners: Evidence for limitations in
processing . Cognition 69 : 105 34.
Shady , Michele, Peter W. Jusczyk , and LouAnn Gerken . 1998 . Infants
sensitivity to function morphemes . Proceedings of the Annual Boston
University Conference on Language Development 19 : 553 63.
Stager , Christine , and Janet F. Werker . 1997 . Infants listen for more pho-
netic detail in speech perception than in word learning tasks . Nature
388 : 381 2.
Trevarthen , Colwyn . 1974 . Conversations with a two-month old . New
Scientist 62 : 230 5.
Trevarthen , Colwyn, Teano Kokkinaki , and Geraldo A. Fiamenghi , Jr.
1999 . What infants imitations communicate: With mothers, with
fathers and with peers. In Imitation in Infancy , ed . Jacqueline Nadel
and George Butterworth , 12785. New York : Cambridge University
Press.
Werker , Janet F. , and Suzanne Curtin . 2005 . PRIMIR: A developmen-
tal framework of infant speech processing . Language Learning and
Development 1.2 : 197 254. Develops a model of early speech percep-
tion/word learning. Concise overview of fndings of speech perception
research.
Werker , J. F. , and R. C. Tees . 1984 . Cross-language speech percep-
tion: Evidence for perceptual reorganization during the frst year of
life . Infant Behavior and Development 7 : 49 63.
INFORMATION STRUCTURE IN DISCOURSE
Te phrase information structure is used to indicate the organiza-
tion of elements within a sentence in terms of their pragmatic
contribution (in terms of givenness-newness, theme-rheme)
to a piece of discourse or text, as opposed to their syntactic
role (subject, object, etc.) or their semantic role (agent, goal,
benefciary, etc.). A sentence elements degree of importance,
Information Structure in Discourse
388
Previous Research on Given-New
In her infuential paper on the given-new distinction, E. F. Prince
( 1981 ) starts with an appealing classifcation of the literature on
given-new according to three approaches to givenness: given-
ness in terms of shared knowledge, in terms of cognitive salience,
and, fnally, in terms of recoverability or predictability. When
givenness in terms of shared knowledge is evaluated, the basic
criterion is the speakers assumption that the hearer knows,
assumes, or can infer a particular thing (but is not necessarily
thinking about it). Tis defnition is problematic in that it fails
to distinguish systematically between knowledge that the hearer
can derive through contextual clues (such as is needed for the
decoding of an anaphoric element) and what he/she knows
as part of his or her background knowledge of the world.
Givenness in terms of cognitive salience can be defned as
knowledge that the speaker assumes to be in the hearers con-
sciousness at the time of utterance. One problem with this view
is that while it might well represent an accurate picture of the
decisions that the speaker has to make on a cognitive level, this
dimension is inherently inaccessible to the outside observer. A
discourse analyst confronted with textual data can only assess
what the speaker might have assumed as given by examining the
linguistic choices the speaker has made (in terms of word order,
intonation, pronominalization, defniteness, and so forth) .
Te view of givenness in terms of recoverability , fnally, is most
often associated with Hallidays systemic-functional framework
(e.g., 1994 ), which was in turn infuenced by the Prague School.
Given information is defned as information that is predictable
or derivable from the preceding discourse context, new infor-
mation being defned as what the speaker presents as not being
recoverable. Important to note here is that, for Halliday, given-
new information is marked almost exclusively through prosody,
more particularly the placement of the tonic nucleus (or sentence
stress). Word order does play an important part in information
structure, but is argued by Halliday to be an indicator of thema-
ticity rather than givenness-newness. Having said that, he does
acknowledge that some noncanonical word order formats, such
as cleft sentences, can be used as markers of given-new fow .
Princes taxonomy of given-new information ( 1981 ) defnes
givenness (or assumed familiarity, as she calls it) in terms of
speaker assumptions regarding an elements cognitive salience
but ofers an attempt at a more sophisticated classifcation.
First of all, elements can be new, inferrable, or evoked (given).
New entities can be either brand new (i.e., not assumed to be in
any way known to the hearer) or unused (i.e., part of the hear-
ers background knowledge but not in his/her consciousness at
the time of utterance). Inferrables can be retrieved from other
entities via inferential processes (as in Te car was useless, as
the battery was fat , where the battery is inferrable from the car
through the inference that cars have batteries ). Evoked entities,
fnally, can be either textually evoked (i.e., from the surround-
ing text) or situationally evoked (i.e., through the physical con-
text). Te familiarity status of an element will have an infuence
on its potential linguistic realization: Brand-new entities, for
instance, will be typically realized as indefnite full NPs (e.g.,
a guy in I met a weird guy yesterday ), whereas unused entities
will be defnite NPs (e.g., the game in I went to the Knicks game
yesterday ) .
sentence stress, or tonic nucleus (i.e., the syllable carrying the
stongest degree of prosodic prominence), normally falls some-
where on the new information, namely, on John in sentence 1 of
Bs turn and perhaps on the second syllable of divorced in sen-
tence 2.
Te efects of givenness and/or thematicity on word order
are acknowledged by most authors. However, in a substantial
proportion of the worlds languages, word order is partially, or
even predominantly, determined by syntactic rather than infor-
mation structure considerations. English is a case in point since
it has a fairly strict SVO order from which it is hard to deviate.
Some languages, therefore, have developed other mechanisms
for marking information structure, such as particular nonpro-
totypical patternings resulting in diferent ordering of elements.
Examples that are prominent in English are the passive voice
(not, strictly speaking, a word order variation but nevertheless
a syntactic operation that has a big impact on linear order); cleft
and pseudocleft constructions (e.g., It is a beer that I would like
or What I would like is a beer ); the fronting of an element (some-
times referred to as topicalization , or Y-movement; e.g., A
beer I would like ); extraposition of an element (e.g., It is enjoy-
able to drink a beer on a warm evening ); and left and right dis-
location (e.g., [ As for ] John, he loves beer and He loves beer, John
[ does ], respectively). Some languages may employ other formal
resources for marking thematicity or givenness-newness, the
best known examples perhaps being particles (such as ga and
wa in Japanese).
Example (1) also reveals some potential problems for any
defnitions of givenness or thematicity. First of all, it raises the
question whether the given-new distinction is a simple binary
distinction. Te verb married in Bs frst sentence, for instance,
is new in the sense that it has not been mentioned, but one could
also argue that previous mention by A of the element single has
to some extent activated its antonym (see spreading activa-
tion ), so that married does not have quite as high a degree of
newness as, say, last year . Secondly, it is important here to distin-
guish between thematicity on a sentence level, which attempts to
provide an answer to the question what a sentence (or clause) is
about, and the notion of aboutness on the macro level of a longer
stretch of discourse; in the latter case, the term topicality or topic-
hood is often employed (see Brown and Yule 1983 ). In the previ-
ous example, one could argue that the entire exchange is more
about Mary than it is about John , making Mary the discourse
topic for the whole exchange (but not necessarily the theme for
all sentences comprising the exchange) .
It should be noted, incidentally, that conceptual vagueness
and terminological confusion abound in the literature: Te
terms topic and theme are often used interchangeably, as are
terms such as given-new and background-focus . As has already
been pointed out, Firbas defnes the theme as the element with
the lowest degree of CD. Others defne theme in terms of about-
ness, which is reminiscent of the concept of psychological sub-
ject, or what the speaker is talking about. M. A. K. Halliday
defnes the theme as the point of departure for the message,
which always correlates with clause-initial position; the rheme
then represents what is said about the theme. He thus assumes
a direct link between discourse function (theme) and linguistic
form (word order) .
Information Structure in Discourse
389
through the channel (for instance, the speaker produces Paul
but the hearer understands ball) .
While mainstream linguistics is focused on human language,
information theory has been applied to many other contexts,
such as the communication systems of other species (McCowan,
Hanser, and Doyle 1999 ; Suzuki, Buck, and Tyack 2006 ), genetic
information storage in the DNA (Li and Kaneko 1992 ; Naranan
and Balasubrahmanyan 2000 ), and artifcial systems such as
computers and other electronic devices (Cover and Tomas
1991 ).
Information theory has myriad applications even within the
domain of the language sciences. I give only a few examples. First,
it provides powerful metrics in psycholinguistics for mea-
suring the cognitive cost of a) processing a word (McDonald and
Shillcock 2001 ), b) an infectional paradigm (Moscoso del Prado
Martn, Kostic, and Baayen 2004 ), or c) the whole mental lexi-
con (Ferrer i Cancho 2006 ). Second, information theory allows
one to explain certain actual properties of human language . For
instance, the tendency of words to shorten as their frequency
increases (Zipf 1935) can be interpreted as increasing the speed
of the information transmitted (e.g., the number of messages
per second) by assigning shorter codes to more frequent codes .
Another well-known property of human language is G. Zipfs
law for word frequencies, one of the most famous laws of
language . It has been argued that this law could be an optimal
solution for for maximizing the information transmitted when the
mean length of words is constrained (Mandelbrot 1966 ) or maxi-
mizing the success of communication while the cognitive cost of
using words is minimized (Ferrer i Cancho 2006 ) . Tird, informa-
tion theory has shed light on the evolution of language . It
has been hypothesized that the presence of noise in the commu-
nication channel could have favored the emergence of syntax
in our ancestors (Nowak and Krakauer 1999 ), which turns out to
be a reformulation of fundamental results from standard infor-
mation theory (Plotkin and Nowak 2000 ). Finally, information
theory ofers an objective framework for studying the diferences
between animal communication and human language .
It is well known that the occurrence of a certain word depends on
distant words within the same sequence, for example, a text, in
human language (Montemurro and Pury 2002 ; Alvarez-Lacalle
et al. 2006 ), and information theory studies in other species
provide evidence that long-distance dependences are not
uniquely human (Suzuki, Buck, and Tyack 2006 ; Ferrer i Cancho
and Lusseau 2006 ). Furthermore, research on humpback whale
songs (Suzuki, Buck, and Tyack 2006 ) questions the conjecture
of Marc Hauser, Noam Chomsky, and W. Tecumseh Fitch ( 2002 )
that only humans employ recursion to structure sequences .
Ramon Ferrer i Cancho
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Alvarez-Lacalle , Enric, Beate Dorow, Jean Pierrer Eckmann , and Elisha
Moses . 2006 . Hierarchical structures induce long-range dynamical
correlations in written texts . Proceedings of the National Academy of
Sciences USA 103.21 : 7956 61.
Cover , Tomas M ., and Joy A. Tomas . 1991 . Elements of Information
Teory . New York : Wiley .
Ferrer i Cancho , Ramon . 2006 . On the universality of Zipfs law for word
frequencies. In Exact Methods in the Study of Language and Text: To
Challenges for Future Research
One important drawback of most research in this area of dis-
course analysis has been that quite often, a top-down analytical
apparatus is employed, whereby a classifcation of givenness
types is proposed on the basis of constructed examples and
is then applied to actual texts (often focusing on narra-
tives, mostly written narrative texts). On the whole, informa-
tion structure in conversation has not received the attention it
deserves; some so-called word order variations appear primar-
ily to be interactive phenomena, cases in point being left and
right dislocations (see Geluykens 1992 , 1994 ). In particular, the
dynamic, procedural nature of conversation as a collaborative
enterprise and the efect that the turn-taking system may have
on the givenness status of elements have not yet been exam-
ined systematically. In addition, there is a dearth of experimen-
tally based analyses trying to determine the exact efect of one
potential variable (such as the referential distance between an
element and its previous mention, for instance) on the given-
ness status of an item. If one assumes that givenness erodes
over time due to the limits of short-term memory (a reasonable
enough assumption), then that efect should be measurable
under controlled conditions. Experimental studies, however,
tend to be limited to the impact of givenness status on prosodic
realization .
Ronald Geluykens
WORK CITED AND SUGGESTIONS FOR FURTHER READING
Birner , B. J. , and G. Ward . 1998 . Information Status and Noncanonical
Word Order in English . Amsterdam and Philadelphia : John
Benjamins .
Brown , Gillian , and George Yule . 1983 . Discourse Analysis .
Cambridge : Cambridge University Press .
Firbas , J. 1992 . Functional Sentences Perspective in Written and Spoken
Communication . Cambridge : Cambridge University Press .
Geluykens , R. 1992 . From Discourse Process to Grammatical
Construction: On Left-Dislocation in English . Amsterdam : Benjamins .
. 1994 . Te Pragmatics of Discourse Anaphora in English: Evidence
from Conversational Repair . Berlin : Mouton de Gruyter .
Halliday , M. A. K. 1994 . An Introduction to Functional Grammar . 2d ed.
London : Edward Arnold .
Lambrecht , K. 1996 . Information Structure and Sentence Form .
Cambridge : Cambridge University Press .
Prince , E. F. 1981 . Toward a taxonomy of given-new information. In
Radical Pragmatics , ed. P. Cole , 22355. New York : Academic Press .
INFORMATION THEORY
Tis is a general purpose and abstract theory for the study of
communication. Standard information theory was founded
by Claude Shannon ( 1948 ) and is based on a communication
framework in which a) the sender must transform a message
into a code and send it through a channel to the receiver, and
b) the receiver must obtain a message from the received code.
Communication is successful if the message of the sender is the
same as the message obtained by the receiver. For instance, a
speaker utters a word for a certain meaning and then the hearer
must infer the meaning that the speaker had in mind. Noise can
alter the code when traveling from the sender to the receiver
Information Theory
390
mathematics. Te issue in linguistics is empirical, though. Te
question of whether some knowledge of language must be innate
has sparked a great amount of controversy over the last decades,
during which philosophical positions ( rationalism versus empir-
icism ) were not always kept separate from the empirical issue.
Prominent protagonists include Noam Chomsky, Jerry Fodor,
Ray Jackendof, Steven Pinker, Kenneth Wexler, Geofrey Pullum,
Terrence Deacon, Michael Tomasello, and earlier, from some-
what diferent perspectives, W. V. O. Quine and Eric Lenneberg.
Te debate arises in essentially two domains: concept learn-
ing and the acquisition of the structure of a language. Concept
learning is associated with Quines thesis of the indetermi-
nacy of translation ( 1960 ), illustrated by his gavagai
example: Suppose one is trying to learn a hitherto unknown lan-
guage. One is walking around accompanied by one of the speak-
ers of the language. Suddenly a rabbit runs by and the speaker
utters gavagai . As Quine shows, there are innumerable English
translations one could come up with for gavagai , among which
it would be impossible to decide on the basis of the experience
alone. To the extent to which in practice we succeed in doing
so, this is something to be explained, for instance, by ascribing
to us an innate conceptual structure, which is how Fodor ( 1983
and subsequent work) proposes that we should account for our
convergence.
Te task of acquiring a language can be characterized as fol-
lows (as in Chomsky 1986 and subsequent work). A human infant
who is exposed to a language will in the course of roughly four to
six years (seven years if one counts full mastery of irregular mor-
phology , or raising verbs; for the latter, see Hirsch and Wexler
2007) acquire a knowledge of that language identical to that of
a human adult who knows the same language (with the excep-
tion of the lexicon, which keeps on growing in adulthood). In this
respect, he or she is entirely unlike apes, dogs, bees, and so on,
that, regardless of how much they are exposed to a language, will
never reach anything like human adult competence. It is also
uncontroversial that a child is not more predisposed to learn-
ing one language than another. But given that a child exposed
to Dutch ends up learning Dutch and not, for instance, Chinese,
and that the converse holds for a child exposed to Chinese, the
input must be crucial in determining which language ends up
being learned.
Consequently, language acquisition can be schematically rep-
resented as in Figure 1, where the various stages that a child goes
through are represented by S
i
and the D
i
are the data to which
he or she is exposed, yielding a change in state (not prejudging
the question of whether these stages involve major qualitative
discontinuities). S
0
is termed the initial state, before the child is
exposed to language. (For the discussion, it is immaterial at what
age we put S
0
. If exposure to language can already take place in
some form in the womb, as there is evidence to believe, S
0
can be
put just before the frst exposure in the womb.) S
n
is the steady
adult state that does not change anymore if over time more data
are presented (with the exception of the lexicon) .
Tis sketch as such should be uncontroversial, and even by
a professed non-nativist such as Tomasello ( 2003 ), it can be
rejected only at the price of incoherence. Te real issue involves
the properties of S
0
(abstracting away from the possible role of
maturation of the brain). By defnition, S
0
is the state of being
Honor Gabriel Altmann , ed. Peter Grzybek and Reinhard Khler ,
13140. Berlin : Gruyter . It discusses the problems of classic models for
explaining Zipfs law for word frequencies.
Ferrer i Cancho , Ramon , and David Lusseau . 2006 . Long-term cor-
relations in the surface behavior of dolphins . Europhysics Letters
14 : 1095 1101.
Hauser , Marc, Noam Chomsky , and W. Tecumseh Fitch . 2002 . Te fac-
ulty of language: What is it, who has it and how did it evolve? Science
298 : 1569 79.
Li , Wentian, and Kunihiko Kaneko . 1992 . Long-range correlations and
partial 1/ f spectrum in a noncoding DNA sequence . Europhysics
Letters 17 : 655 60.
Mandelbrot , Benoit . 1966 . Information theory and psycholinguistics: A
theory of word frequencies. In Readings in Mathematical Social
Sciences , ed. P. F. Lazarsfeld and N. W. Henry , 15168. Cambridge,
MA : MIT Press.
McCowan , Benoit, Sean F. Hanser , and Laurance R. Doyle . 1999 .
Quantitative tools for comparing animal communication sys-
tems: Information theory applied to bottlenose dolphin whistle reper-
toires . Animal Behavior 57 : 409 19.
McDonald , Scott A ., and Richard Shillcock . 2001 . Rethinking the word
frequency efect: Te neglected role of distributional information in
lexical processing . Language and Speech 44 : 295 323.
Montemurro , Marcelo , and Pedro A. Pury . 2002 . Long-range fractal cor-
relations in literary corpora . Fractals 10 : 451 61.
Moscoso del Prado Martn , Fermn , Alexander Kostic , and Harald R.
Baayen . 2004 . Putting the bits together: An information theoretical
perspective on morphological processing . Cognition 94 : 118.
Naranan , Sundaresan , and Vriddhachalam K. Balasubrahmanyan . 1998 .
Models for power law relations in linguistics and information sci-
ence . Journal of Quantitative Linguistics 5.3 : 35 61. A summary of
many optimization models of Zipfs law for word frequencies that are
based on information theory.
. 2000 . Information theory and algorithmic complex-
ity: Applications to linguistic discourses and DNA sequences as com-
plex systems. Part I: Ef ciency of the genetic code of DNA . Journal of
Quantitative Linguistics 7.2 : 129 51.
Nowak , Martin A. , and David Krakauer . 1999 . Te evolution of language .
Proceedings of the National Academy of Sciences USA 96 : 8028 33.
Plotkin , Joshua , and Martin A. Nowak . 2000 . Language evolution and
information theory . Journal of Teoretical Biology 205.1 : 147 59.
Shannon , Claude E. 1948 . A mathematical theory of communication .
Bell Systems Technical Journal 27 : 379 423, 62356.
Suzuki , Ryuji, John R. Buck , and Peter L. Tyack . 2006 . Information
entropy of humpback whale songs . Journal of the Acoustical Society
of America 119 : 1849 66.
Zipf , George Kingsley . 1935 . Te Psycho-biology of Language .
Boston : Houghton Mif in .
INNATENESS AND INNATISM
What does our knowledge come from? Innatism is the position
that at least some of our knowledge is inborn rather than derived
from experience . If so, the question naturally comes up con-
cerning what types of knowledge should be taken to be innate,
giving rise to specifc hypotheses about innateness. In linguis-
tics, innateness became an important issue with the develop-
ment of generative grammar , giving rise to a debate that
is inextricably linked to language acquisition, namely, whether
or not language acquisition is based on an innate universal
grammar (UG). Innatism originated as a philosophical posi-
tion, often focusing on concepts varying from god to justice to
Innateness and Innatism
391
a reasonable length, the number of well-formed sentences in
English is astronomical. It has been estimated that the number
of grammatical English sentences of 20 words and less is 10
20
(Levelt 1967 ). (Note that this very normal sentence is exactly 21
words and costs nine seconds to pronounce). At an average of six
seconds per sentence, it will cost 19 trillion years to say (or hear)
them all. In the case of nonstop listening, the percentage of these
a child could have heard in six years time is 0.000000000031,
clearly still a gross overestimation as compared to what the child
can actually be expected to hear. So on the basis of at most such
an extremely small percentage of the potential input, the child
gets to know how language works.
Tere are many further practical complications we ignored,
such as lack of homogeneity and the presence of errors in the
data. If we were to take these into account, the task would only
become more formidable. Tis sets the stage for the logical
problem of language acquisition (for instance, Chomsky 1965 ,
1986 ), which can be formulated as the projection problem:
Consider a given fnite set taken from some (infnite) superset.
Determine (the characteristic function of) this superset on the basis
of this subset.
Like anyone can see, this task is in its generality impossible. For
any fnite given set, the projection problem has infnitely many
solutions.
For a concrete illustration, consider the following task involv-
ing the completion of a series of numbers:
(1) 1,2,3,4,5,6,7, .
One might say that fnding the next number is easy. It should
obviously be 8. But of course, this is not at all guaranteed. It is easy
to think of a perfectly well-behaved function that enumerates the
frst seven natural numbers, followed by their doubles, triples,
quadruples, and so on. Tis illustrates a very simple point: Tere
is no general procedure to establish the correct completion of
some initial part of a sequence, whether a sequence of numbers
as in (1) or a data sequence (D
1
, D
2
, D
i
, D
n
) as in Figure 1.
Tis fact refects a poverty of the stimulus in a fundamental
sense, as a trivial logical truth.
Te completion task may become possible, however, if it is
redefned as the task to fnd a solution within a restricted space of
possible solutions . In that case, certain instances of the projection
problem become even trivial. For instance, (1) can be trivially
completed if it is given that there is a constant diference between
each member of the series and its successor.
As E. Gold ( 1967 ) showed, even highly restricted hypothesis
spaces may not ensure a solution of the projection problem as
defned. Te task may become easier if the input does contain
systematic evidence as to what is not in the target language.
Since, as is generally acknowledged, the input to the child does
able to acquire language, distinguishing humans from apes,
dogs, and so on. Tat is, it refects mans innate capacity for
language. By the defnition of the term universal grammar as it
is used in generative grammar, S
0
coincides with UG. Whereas
a non- nativist could object to the term UG, nothing more is
involved than a terminological issue. What is really at stake are
questions such as the following: i) What properties must S
0
have
in order to be able to account for the fact that language can be
acquired, given what we know about time course and access
to data? ii) Which of these properties are specifc to man? And
iii) What aspects of S
0
are specifc to language, and how are the
other aspects related to other human cognitive capacities? Te
frst question can only be successfully approached by carefully
investigating necessary and sufficient conditions for
learning and for learnability of (classes of) languages of the
human type. Te second and third questions require an under-
standing not only of the human language capacity but also of
those other cognitive capacities among which it is embedded.
Logically, S
0
could be empty. However, this would entail no
diference between humans and animals, contrary to what we
know. So for this empirical reason alone, S
0
cannot be empty.
But as is discussed in the next section, it would also entail that
humans cannot acquire language contrary to what we know.
The Logical Problem of Language Acquisition
A useful strategy for demonstrating the dif culty of a problem is
to simplify it. If the simplifed problem is still hard, one knows
that the original problem is at least as hard. So we take a highly
simplifed question as a starting point (Wexler and Culicover
1980 ): What does a person who knows a language minimally
know? A reasonable answer is the following: A person who knows
a language knows at least which strings of words correspond
to well-formed sentences in that language and which strings
dont. (Tis simplifcation is valid irrespective of the changes in
the signifcance attached to this particular aspect of linguistic
knowledge, from Chomsky 1957 to Chomsky 1995 .)
In this simplifed picture, we view a language as a subset
of the set of all expressions one can form over a given vocabu-
lary. Tat is, assuming that the vocabulary of English contains
the elements the, dog, bites, man , the set of English sentences
will contain the dog bites , the dog bites the man , and so on, but
not the bites man , bites dog the , and so on. Te task of the child
acquiring English, therefore, minimally includes determining
what the full set of English sentences is like on the basis of the
sentences he/she is exposed to for some period of time, lets say
for six years. Te question is then to get an impression of how
hard this task is.
Note that there is no upper bound to the length of individual
sentences. Tis makes the set of sentences in a language efec-
tively infnite. However, even if restricted to sentences under
S
0
D
1
D
2
D
i
D
n
D
n+1
D
n+2
S
1
S
2
S
i
S
n
S
n
S
n
........
Figure 1.
Innateness and Innatism
392
require a theory of how to acquire the ability to use the context of
utterance, or a formal proof that this structuring of the presenta-
tion suf ciently aids the child in setting up correct hypotheses
and rejecting incorrect ones but, crucially for the non-nativists,
without attributing to the child innate knowledge as to what evi-
dence is to be absorbed and what evidence ignored. Tese are all
equally susceptible to the poverty of the stimulus argument in
the logical sense.
Clearly, it is important to have a conception of what it means
for something to be a property of UG. For instance, Chomsky
( 1980 ) suggests that the specifed subject condition (SSC) defn-
ing the domain in which anaphors must be bound (condi-
tion A) and pronominals must be free (condition B) is part of
our genetic endowment. Te question is, then, what it means
for UG. Must an SSC be hardwired as such? Or is it suf cient
if the restrictions on binding descriptively captured by the SSC
follow from basic properties of our computational system, mod-
ulo properties of the mental space in which these computations
take place? For instance, Chomsky ( 1995 ) and subsequent work
argue that grammar is based on a very simple set of combina-
tory principles (essentially merge and Agree) and conditions
that follow from general properties of computation. If so, that
is what UG essentially amounts to. Reuland ( 2005 ) shows
that the core of condition B the need to license refexivity of
a predicate follows from the fact that no computational sys-
tem can distinguish between indistinguishables , as in the case
of the arguments of a refexive predicate. Elements like self or
other morphological markers must be added for the system
to handle these arguments. To derive condition A, no more is
needed than a general principle of economy of encoding , the
general combinatorics of the language system, and the lexical
semantics of self as an identity predicate. Tus, prima facie
substantive properties of language and UG reduce to the inter-
action between general properties of mental computations and
lexical representations. If so, there is indeed no sense in which
conditions A and B are acquired. Tey refect basic properties
of the system embedded in our wetware but it takes extensive
linguistic research to show that this is so .
Language and Pattern Recognition
Non-nativists crucially invoke general learning strategies that
originate in our general cognition. However, a statement that
language is an emergent property of our general cognitive sys-
tem requires a substantive theory of its workings, specifying how
its operations account for language with the same amount of
precision as the rules of formal linguistics (none of the proper-
ties in the explanans should invoke the explanandum). So far, no
precise proposals have been made available.
Te essence of Tomasellos claim is that there is no interest-
ing problem in language acquisition, since whatever is needed is
provided by our abilities for pattern recognition . Tomasello (and
others) assume that there exists an ability for pattern recognition
that provides the tool for fnding the patterns in language. Of
course, we humans have the ability to fnd patterns. However, it
is a fallacy to think that what we do is fnding the patterns that
are out there. Te main message about concept learning to be
gleaned from Quines insights is that our mind must impose pat-
terns. Elementary considerations from particle physics show us
not contain systematic negative evidence, it becomes of prime
importance to identify the types of hypothesis spaces that do
allow learning of natural languages by presentation only. Te
absence of negative evidence, together with the fact that a sub-
stantial number of actual utterances a child may hear will be less
than entirely well formed, are often referred to as poverty of the
stimulus as well. But this is, in fact, not the same notion as the
fundamental, logical one employed in the discussion of Figure 1
and (1). For a rational debate it is crucial to keep the, frst, logical
sense and the, second, narrower, empirical sense apart.
Given the poverty of the stimulus in the logical sense, lan-
guage acquisition cannot be accounted for without the assump-
tion of innate genetically determined restrictions. Te poverty
of the stimulus in the empirical sense may help provide further
evidence on what these are. In the generative literature, it is
often claimed that these restrictions have the form of an inven-
tory of grammatical principles. It is presumably this reference
to grammatical principles that led to the poverty of the stimulus
debate that is, for or against the existence of innate principles
of grammar as it is usually conducted with its emphasis on the
poverty of the stimulus in the narrower sense. But for a fruitful
discussion, it is crucial to distinguish between the minimal prop-
erties that S
0
must have in order to explain language acquisition
and the further question of whether S
0
has properties that are
specifc to language .
Learnability and Complexity
Each restriction on the hypothesis space defnes a class of gram-
mars and languages. As pointed out by Gold ( 1967 ) and Wexler
and Peter Culicover ( 1980 ), learnability does not depend on the
complexity of the individual language/grammar but only on the
structure of the class in which the selection must be carried out.
Many contributions to the debate center on specifc exam-
ples, such as the question of whether the child uses and under-
stands utterances that are unexpected given the input up to a
certain stage, whether or not the input is restricted (as in the case
of motherese , a restricted register caretakers use in addressing
their children), or whether the input is richer and provides more
clues than one might have initially expected. As Pullum and
Barbara Scholz ( 2002 ) point out, one must distinguish between
the general issue of a specifc genetic endowment and debates
as to whether particular clues that have been argued to be non-
existing can or cannot be found in natural language corpora. But
important as issues of the latter type are, they are independent of
the general problem.
Deacon ( 1997 ) attributes to Chomsky the position that natural
language is too complex to be learned without rich innate mech-
anisms and then proceeds to argue that it is in fact not as com-
plex as is being claimed. Part of the argumentation in Tomasello
( 2003 ) is based on the same premises. As demonstrated in the
previous section, none of these issues bears on the poverty of the
stimulus in its logical sense.
Te same applies to arguments stating that the input is much
richer than nativists presuppose. As Wexler and Culicover
( 1980 ) show, proposals involving an enriched input (a more
structured presentation in which pragmatic clues are provided
by the context of the utterance, as in the case of motherese ) mag-
nify the logical problem instead of decreasing it. Such proposals
Innateness and Innatism
393
depth of misunderstanding involved: It is an empirical problem,
but analyzing the logical problem is essential for solving it.
All this does not demonstrate that the restrictions necessary
for language acquisition are specifc to language. It does show
that such restrictions are there and have to be studied if we are to
understand language acquisition. No insight can be gained unless
precise and substantive hypotheses are formulated and tested in
a way that refects what we already know about language .
Eric Reuland
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , Noam . 1957 . Syntactic Structures . Te Hague : Mouton.
. 1965 . Aspects of the Teory of Syntax . Cambridge, MA : MIT Press.
. 1980 . On cognitive structures and their development. In
Language and Learning, Te Debate between Jean Piaget and Noam
Chomsky , ed. M. Piattelli-Palmarini , 3552. Cambridge : Harvard
University Press.
. 1986 . Knowledge of Language: Its Nature, Origin and Use . New
York : Praeger.
. 1995 . Te Minimalist Program . Cambridge, MA : MIT Press .
Deacon , Terrence . 1997 . Te Symbolic Species . New York : Norton.
Fodor , Jerry . 1983 . Te Modularity of Mind . Cambridge, MA : MIT Press .
Gold , E . 1967 . Language identifcation in the limit . Information and
Control 16 : 447 74.
Hirsch , C. , and K. Wexler . 2007 . Te late development of raising: What
children seem to think about seem. In New Horizons in the Analysis
of Control and Raising , ed. W. Davies and S. Dubinsky , 3770.
Heidelberg : Springer.
Kandel , Eric, James Schwartz , and Tomas Jessell . 2000 . Principles of
Neural Science . 4th ed. New York : McGraw-Hill.
Lenneberg , Eric H. 1967 . Biological Foundations of Language . New
York : Wiley .
Levelt , Willem . 1967 . Over het Waarnemen van Zinnen . Groningen,
Germany : Wolters.
Pullum , Geofrey K ., and Barbara C. Scholz . 2002 . Empirical assessment
of stimulus poverty arguments . Linguistic Review : 19.1/ 2: 9 50.
Quine , Willard Van Orman . 1960 . Word and Object . Cambridge, MA: MIT
Press.
Reuland , Eric . 2005 . Binding conditions: How are they derived? In
Proceedings of the HPSG05 Conference Department of Informatics,
University of Lisbon , ed. Stefan Mller . Stanford, CA : CSLI Publications .
Available online at: http://csli-publications.stanford.edu/
Tomasello , Michael . 2003 . Constructing a Language . Cambridge : Harvard
University Press.
Wexler , Kenneth , and Peter Culicover . 1980 . Formal Principles of
Language Acquisition . Cambridge, MA : MIT Press.
INTEGRATIONAL LINGUISTICS
Tis is the application of integrational semiology to the study
of language. Integrational linguistics is based on the assump-
tion that human communication, whether verbal or nonverbal,
involves the creation of signs in particular contexts whereby two
or more individuals engage in joint integration of their respec-
tive activities. Tus, for example, speech communication would
be impossible without integration of the biomechanically sepa-
rate activities of vocalization and hearing. Written communica-
tion requires the production of marks on a surface that can be
integrated with programs of optical scanning. Te appropriate
integration of these and many other activities is one of the major
the same: Our common senses are blatantly incapable of seeing
reality as it is. As our extended senses (in the form of experi-
mentation and model building) teach us, what is actually out
there bears minimal resemblance to what we can observe with
our common senses. If even individual events that we observe,
imprint in our memory, and store can only have a remote resem-
blance to what is there, the more so for the patterns that we
fnd.
Any pattern involves extrapolation beyond what can be
observed; even the simplest of observations requires an active
mind shaping our internal representation of what is out there.
We can do no more than impose a pattern (one of the zillions of
possible patterns any piece of reality embodies), hopefully in
a way compatible with our survival. Te good thing is that evo-
lution resulted in our patterns being useful enough for every-
day purposes since we managed to survive (for the moment).
Te bad thing is that we can do so not because we are so smart
but because we are so limited. We can learn only if we ignore
the many logically possible alternatives. Evolution keyed us
to the universe. Te key to understanding how we learn is to
understand our limitations, the logically possible patterns we
ignore.
In this respect, learning a language is like fnding the pat-
terns in the surrounding universe. In another respect, there is
a diference, although it has been misconstrued. Deacon ( 1997 )
proposes that our capacity to learn language is not surprising
since language is a human product and, therefore, is made to
be learned. However, unless we are careful, this leads us back
to the whole range of issues discussed, framed slightly difer-
ently: How are the properties allowing language to be learned
refected in its structure, and what does the fact that we acquire
language tell about our cognitive abilities?
Nevertheless, it also contains a relevant insight. Unlike the
physical world, language is a product of the human mind. So we
know that the childs mind is keyed to getting to know language
in a way he or she will never know the physical world: Complete
knowledge is attainable. As in the case of the physical world, the
input for learning language is external. However, unlike in the
case of the physical world, we know the nature of the input to
an extent that is unique. So language acquisition refects labo-
ratory conditions for learning, facilitating an understanding of
learning per se.
It is surprising that so many researchers of human learning
have such a hard time accepting the implications of the projec-
tion problem, though the moral is so simple, like the frst law of
thermodynamics prohibiting the perpetuum mobile. At it reads
in the well-known words of C. P. Snow:
You cannot win (that is, you cannot get something for nothing)
Applied to learning:
Learning a recursive step by presentation only, without restrictions
on the hypothesis space, is as impossible as creating the perpetuum
mobile.
Te last sentence in Tomasello reads: How children become
competent users of a natural language is not a logical problem
but an empirical problem (2003, 328). Paradoxically, I agree
and, at the same time, would like to say that it illustrates the
Innateness and Innatism Integrational Linguistics
394
The Principle of Noncompartmentalization
It also follows from the semiological axioms that there is no strict
or objective dividing line between linguistic knowledge and non-
linguistic knowledge, or, as some theorists put it, between knowl-
edge of a language and knowledge of the world. Recognition of
this indivisibility is referred to by integrationists as the principle
of noncompartmentalization . In other words, human beings do
not live in a communicational environment where what pertains
to language belongs to one compartment and the rest belongs to
some other compartment (or compartments).
Non-compartmentalization is also heresy in orthodox lin-
guistics, since it implies that linguistics cannot be a science.
(A physicist who confessed inability to diferentiate between the
facts of the physical world and the nonphysical world would be
confessing to a similar heresy.) In an academic milieu where
every inquiry aspires to be scientifc, this doctrine is not
popular.
Te integrationist view can be illustrated by considering what
happens at a cocktail party. Physically, a certain level of audible
vibration is generated (often said to be deafening, though very
small as compared with the energy required to light one electric
lamp). Physiologically, there is much expenditure of efort in
terms of the muscular action of vocal apparatus (but again very
small by comparison with the efort required to walk across the
road). Mentally, there is doubtless engagement in interactions
with others, but it cannot be quantifed. So where in all of this is
the language component? It seems to be in there somewhere,
but exactly where defes exact location. To ask where is itself a
nonsense question. And to grasp why it is a nonsense question is
already halfway to subscribing to the integrationist principle of
noncompartmentalization.
Te orthodox linguistic answer is that the language compo-
nent resides somewhere in the heads of the talkers and listen-
ers. But so, presumably, does their knowledge of football, food,
local politics, and everything else; that is, all the things being
talked about at the cocktail party. So exactly the same compart-
mentalization perplexity arises at one remove. Human beings
cannot in their everyday lives distinguish between knowing
something about X and being able to talk about it .
The Principle of Cotemporality
Te integrationist principle of cotemporality complements the
principle of noncompartmentalization. Everyday experience
recognizes that an event occurring at time t may afect how we
interpret an event occurring earlier or later than t . Temporal
sequence is an intrinsic aspect of contextualization.
Compare the situation in which (1) landlord says Te water
is turned of today and tenant says I must have a shower with
the situation in which (2) tenant says I must have a shower and
landlords response is Te water is turned of today. Ostensibly,
the same information has been exchanged and the same words
used. But what emerges from the communicational episode is
quite diferent in the two cases.
For integrationists, the question of temporal sequence
involves both verbal and nonverbal behavior. In brief, there is no
way that what is said can be set apart from the train of events in
which it occurs, whether these are verbal or nonverbal. We all
know this. Tat is why in legal disputes courts treat diferently
functions of the cerebral cortex, and failure to achieve the inte-
grational profciency required for social fuency in communi-
cation is commonly perceived as a defect or handicap of some
kind (e.g. deafness, dyslexia , etc.) when due to physiological
factors. Integrationists diferentiate forms of communication
according to the range of activities typically integrated by the
participants, and the kinds of integrational profciency typically
required for participation. A blind person, for obvious reasons,
lacks the integrational profciency presupposed in various forms
of visual communication (see blindness and language ).
From this perspective, the term language does not correspond
to any one mode of communication but straddles or confates
several. In this respect, integrational linguistics difers radically
from mainstream schools of thought in linguistics and neigh-
bouring disciplines, which tend to assume that language is a sin-
gle human faculty, common to all humanity, and that languages
(English, French, Latin, etc.) are diferent social codes enabling
individual users to exercise this faculty. For integrationists, on
the other hand, individuals are not language users but language
makers . Tey make language by their creative integration of ver-
bal signs into a myriad diverse activities, in both expected and
unexpected ways, with due regard for the circumstances, just as
they make human relationships by the ways in which they inter-
act with others in particular cases .
Axioms of Integrational Semiology
Te axioms of integrational semiology are as follows:
1. What constitutes a sign is not given independently of the
situation in which it occurs or of its material manifestation in
that situation.
2. Te value of a sign (i.e., its signifcation) is a function of the
integrational profciency that its identifcation and interpre-
tation presuppose.
As applied to language studies, this means that verbal com-
munication of whatever kind cannot be decontextualized .
Episodes of communication are episodes in the lives of partic-
ular individuals at particular times and places. Tese episodes
have to be studied as such. We learn nothing from an analysis
telling us, for instance, that someone uttered the sentence John
loves Mary , that John is the subject of the sentence, Mary is the
direct object of the verb love , love is a transitive verb, and so on
nothing, that is, except information about the metalinguistic
assumptions of the analyst. (Tese assumptions may be worth
studying in their own right, but that is not the same as studying
the facts pertaining to the utterance in question. On the contrary,
the analysis is one that already embarks on a decontextualization
of the episode allegedly described.)
It is not simply a matter of knowing who said what to whom ,
where , and in what circumstances . Nor are the circumstances
what happened immediately before and after. Orthodox modern
linguistics, like traditional grammar, routinely assumes the legit-
imacy of abstracting from all these features of context. Its state-
ments are supposedly generalizations across indefnitely many
unidentifed episodes of language use. Integrational linguistics
rejects on principle the legitimacy of such generalizations: Again,
they tell us nothing about linguistic facts only about the intel-
lectual preferences or prejudices of the analyst .
Integrational Linguistics
395
to agree about the meanings of words than they are obliged to
agree about the value of goods. In both cases, it usually suits their
purposes to make some kind of compromise with those they are
dealing with. But the nature of this semantic compromise is
essentially ad hoc .
Parameters of Communication
For integrationists, there are certain capacities required of an
individual in order to participate in communication. Tese
capacities are of three kinds: biomechanical , macrosocial , and
circumstantial . Te frst relates to the human organism and the
ability to integrate activities requiring a wide variety of physi-
ological and mental processing (e.g., the very diferent biome-
chanical requirements involved in speech production and
hearing). Te second relates to the human ability to integrate
particular activities into sets of assumptions provided by social
conventions of various kinds. Te third relates to the ability to
integrate ones activities into whatever else is going on at the
time.
A simple illustration of all three is provided by what happens
when a motorist encounters a pedestrian about to cross the road.
At the macrosocial level, there are assumptions about the road,
the conventions of the highway code, and so on. At the biome-
chanical level, there are questions about what the motorist sees
within a certain range of visibility and the alertness necessary for
initiating the appropriate physiological actions for driving the
vehicle. At the circumstantial level, even when all of these aspects
have been taken into account, the individual motorist must con-
stantly be prepared to modify any action taken, depending on the
moment-to-moment behavior of the pedestrian and other road
users.
Tis does not mean that the levels of integration are inde-
pendent. Circumstantial factors intervene across the board. An
accident may be caused because the pedestrian is shortsighted
or the road badly lit. Te drivers exercise of biomechanical skills
may be afected by knowing or not knowing that the brakes of
this particular vehicle are not very reliable.
Much recent work in integrational linguistics has focused on
analyzing how, at the macrosocial level, societies succeed in con-
structing supercategories , which integrate what would otherwise
be quite separate activities in such a way as to set up a common
framework for the intellectual and practical pursuits of society as
a whole. Tese supercategories include science, art, history, and
religion, each with a dedicated discourse of its own .
Roy Harris
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Harris , R. 1998 . Introduction to Integrational Linguistics . Oxford :
Pergamon . A concise survey of the whole feld.
Harris , R. , and C. Hutton . 2007 . Defnition in Teory and Practice .
London : Continuum . An integrationist approach to problems of def-
nition, with particular reference to lexicography and the law.
Love , N., ed. 2006 . Language and History: Integrationist Perspectives .
London : Routledge . Collection of papers bearing on the construction
of the macrosocial supercategory history.
Toolan , M. 1996 . Total Speech: An Integrational Linguistic Approach to
Language . Durham, NC : Duke University Press . Includes interesting
discussions of literal meaning, metaphor , and related issues.
a case in which A insulted B , who then struck A , from a case in
which A struck B , who then insulted A . As individuals, we are
time-bound agents in all our activities. Our linguistic acts do
not have some special time track of their own. Tere is no such
thing as a contextless linguistic sign.
Tis does not mean that the context simply is a given
sequence of events. Integrationists take a very diferent view of
context from that usually found in orthodox linguistics. Context
is not to be equated with situation (which may be irrelevant)
and even less with preceding or following speech-acts . It is
not some kind of local backdrop against which communication
takes place. Context, for the integrationist, is always the product
of contextualization, and each of us contextualizes in our own
way. Te individual participants in any communication situation
will each contextualize what happens diferently, as a function
of the integrational profciency each exercises in that situation.
Tis does not mean that we can never reach communicational
agreement, but it explains why we often do not. It is not enough
to say that every act of communication is unique. Each such act
is in principle subject to multiple contextualizations and recon-
textualizations. Tat is what makes it essential in linguistic analy-
sis for the analyst to specify what forms of contextualization are
presupposed (a requirement that the great majority of orthodox
linguists ignore). For integrationists, a sign is not a sign until it
has been contextualized: Te act of contextualization and the
establishment of the sign are one and the same .
Meaning
It follows, then, that integrationists take a quite diferent view of
meaning from that which informs most work in orthodox linguis-
tics, where the meaning of a linguistic form is usually construed
as some kind of concept (as in the defnitions of conventional
dictionaries) or, even more vaguely, mental representation .
Tis is traditionally imagined to yield a more or less permanent
value attached to the form and known to all competent speak-
ers and writers of the language in question (even if the language
is no longer living). Tus, the meaning of classical Latin aqua is
treated as timeless: It still means for Latin specialists what it
meant in the days of Julius Caesar and will mean in a thousand
years from now.
For integrationists, this assumption conficts with the princi-
ple of cotemporality. Tere are no atemporal invariants in lan-
guage. Meaning is made by participants as part of the process of
communication. It is thus subject to the principle of cotemporal-
ity in just the same way as all other aspects of communication.
Tere are no fxed meanings. Tere is nothing in language to pro-
vide us with a miraculous guarantee of the stability of meaning(s)
over time or even from one moment to the next. To demand
such a guarantee for any mode of communication is as futile as
demanding that a currency remain stable in value from one day
to the next. (Te indeterminacy holds regardless of whether peo-
ple shopping in High Street are aware of it.)
We have no option but to interpret particular episodes of
communication by integrating them into the unique temporal
sequence of events that constitutes our previous experience.
Tus, where there are two or more participants, what is com-
municated must be open to two or more interpretations. Tese
cannot be guaranteed to coincide. People are no more obliged
Integrational Linguistics
396
difers in meaning from
(4) Lola believes that Mark Twain is Samuel Clemens.
For if (3) and (4) have the same meaning, then they are either
both true or both false. But (3) may be true while (4) is false for
example, if Lola does not realize that Mark Twain and Samuel
Clemens refer to the same man. Since (3) is just like (4) except for
an occurrence of Samuel Clemens instead of Mark Twain , Mark
Twain and Samuel Clemens difer in meaning.
Such are two arguments that Mark Twain and Samuel Clemens
difer in meaning. But if the extension of a proper noun is its
referent, then Mark Twain and Samuel Clemens have the same
extension. Tus, some (but not all) conclude, there is a distinc-
tion between two aspects of meaning: intension and extension.
Extensional and Nonextensional
Arguably, then, a true sentence (3) may be turned into a false sen-
tence (4) merely by substituting an occurrence of Mark Twain by
Samuel Clemens . Suppose that the extension of a sentence deter-
mines its truth value and that Mark Twain and Samuel Clemens
have the same extension. If so, then to change (3) into (4) is to
replace part of sentence (3) with a coextensional expression,
while failing to preserve the extension of the entire sentence.
Contexts like Lola believes that are typically thought to
be nonextensional. An extensional context is a context wherein
substitution by a coextensional expression always preserves the
extension of the larger expression. A nonextensional context is
thus a context wherein substitution by a coextensional expres-
sion sometimes fails to preserve the extension of the larger
expression.
Nonextensional contexts appear to be widespread in natural
language. In English, for instance, propositional attitudes
(like Lola believes that , John desires that ), as well as
many other constructions (involving, for example, seeks,
admires, avoids, resembles, necessary, possibly,
must, may, obviously, and because), have been held to
be nonextensional.
Nonextensional contexts are often called intensional. But
sometimes, more carefully, intensional is reserved for those
contexts wherein substitution of a cointensional expression
always preserves the intension of the larger expression. Contexts
wherein cointensional substitution may fail are then called non-
intensional or hyperintensional . Lola believes that is one
context thought by some to be non-intensional in this sense. For
example, according to some, eye doctor and ophthalmolo-
gist have the same intension, but Lola believes that Eve is an
eye doctor and Lola believes that Eve is an ophthalmologist
have diferent intensions: If Lola doesnt know the word ophthal-
mologist , the former may be true and the latter false .
Semantics Without Intensions
W. V. O. Quine argued that intensions and other elements of tra-
ditional theories of meaning have no place in a scientifc descrip-
tion of the world. He claimed that intensions are on a par with
the Homeric gods: Intensions play no useful explanatory role in
a scientifc description of the world. According to one of his most
infuential arguments, there is no noncircular way to make sense
of traditional notions like meaning, synonymy, analyticity ,
Wolf , G., and R. Harris , eds. 1998 . Integrational Linguistics: A First Reader .
Oxford : Pergamon . Collection of papers covering a wide range of topics
from an integrationist perspective.
INTENSION AND EXTENSION
semantic theories commonly distinguish two aspects of linguis-
tic meaning: intension and extension. Roughly, the intension of a
linguistic expression is what it means, and its extension is what it
refers to. For example, on some views, Neil Armstrong and the
frst human being to walk on the Moon have the same exten-
sion but diferent intensions. Often it is held that intension deter-
mines extension (expressions with the same intension have the
same extension) but is not determined by extension (expressions
with the same extension may have diferent intensions).
Tis entry describes various roles that intensions and exten-
sions play in contemporary semantic theories. While much
recent discussion uses formal methods, the presentation here is
nontechnical.
Some possible terminological confusions should be noted.
A long tradition of distinguishing diferent aspects of meaning
(including Porphyrys third-century commentaries on Aristotle,
the Port-Royal logic of 1662, Mill [ 1843 ] 1872, and Frege [ 1892 ]
1997) has left us with a hodgepodge of terms and distinctions.
One should be wary whether, for example, the Port-Royal log-
ics comprehension and extension, or Freges Sinn and
Bedeutung (often translated sense and reference; see sense
and reference ), or Mills connotation and denotation
mark the same distinction as intension and extension in
recent theories. Moreover, extension is sometimes, but not
always, taken to be synonymous with denotation, designa-
tion, or referent, while intension, intensional, and inten-
sionality are sometimes, but should not be, confused with
intention, intentional, and intentionality. Arguably, there
is a relation between a speakers intentions and the meanings of
the words he or she uses; however, the connection should not
be drawn via terminological confusion. intentionality, the
distinctive property of thoughts and other mental phenomena,
is yet something else entirely.
Why Accept a Distinction Between Intension and
Extension?
Knowing what sentence (1) means is enough to be able to know
that it is true.
(1) Mark Twain is Mark Twain.
However, (2) difers.
(2) Mark Twain is Samuel Clemens.
Although (2) is true, merely knowing what (2) means is not
enough to be able to know that it is true. So (1) and (2) difer
in meaning. But (2) is just like (1) except for an occurrence of
Samuel Clemens instead of Mark Twain . So, Mark Twain and
Samuel Clemens difer in meaning since the meanings of (1) and
(2) are determined only by the meanings of their parts and the
way they are put together. (See compositionality .)
Similarly,
(3) Lola believes that Mark Twain is Mark Twain.
Intension and Extension
397
Montague proposed complex intensions, mapping not possible
worlds but indices a combination of a possible world with
persons, places, times, and so on. David Kaplan ( 1989 ) instead
divided intensions into two pieces: character (the linguistic
meaning of an expression type) and content (the meaning of an
occurrence of an expression). Roughly, character plus context
determines content, and content plus context yields extension.
Kaplans arguments about indexicals and demonstratives, as
well as Saul Kripkes and Hilary Putnams on proper names and
natural kind terms , have led many to accept that these terms
are devices of direct reference; in efect, occurrences of these
terms have extensions but no intensions. Others have responded
by adopting the framework of two-dimensional semantics,
where expressions are assigned two diferent intensions (Garcia-
Carpintero and Maci 2006 ).
While semantic theories inspired by Carnap and Montague
do deal in intensions, they have much in common with the no-
intension theories discussed earlier. Not only are both sorts of
theories compositional and referential, but both sorts of theories
can also arguably fail in various ways to capture intuitive notions
of meaning. For example, if the intension of a sentence is a func-
tion from possible worlds to truth values, then if two sentences
have the same truth value at every possible world, then they have
the same intension. Tat means that all true mathematical sen-
tences have the same intension, for presumably a mathematical
truth is true at every possible world. But 2 < 3 and 2 + 2 = 4 do
not have the same meaning.
Within current frameworks, lively discussion about these and
other questions continues. (For a useful introduction, see von
Fintel and Heim 2007 .) More radically, others propose to rework
the foundations of semantics to produce more fne-grained
intensions (e.g., Fox and Lappin 2005 ). And then there is the
extreme view of J. J. Katz ( 1990 ): We should give up the claim that
intension determines extension and adopt an internalist notion
of intension better suited to the traditional duty of explaining
analyticity, synonymy, meaningfulness, and so forth.
A variety of views about the distinction between intension and
extension are alive. Widely, but not universally, it is thought that
both intensions and extensions are needed in semantics. Debate
continues about what intensions and extensions are, about what
the relation between intension and extension is, and about the
sorts of linguistic expressions that have them .
Patrick Hawley
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Carnap , Rudolf . [1947] 1956 . Meaning and Necessity: A Study in Semantics
and Modal Logic . Chicago : University of Chicago Press .
Chierchia , Gennaro , and Sally McConnell-Ginet . 2000 . Meaning and
Grammar . 2d ed. Cambridge, MA: MIT Press.
Church , Alonzo . 1951 . A formulation of the logic of sense and deno-
tation. In Structure, Method and Meaning: Essays in Honor of H. M.
Shefer , ed. P. Henle , H. Kallen , and S. Langer , 324. New York : Liberal
Arts Press .
Davidson , Donald . 1967 . Truth and meaning . Synthese 17 : 304 23.
Reprinted in Inquiries into Truth and Interpretation ( Oxford : Oxford
University Press , 1984 ).
Fox , Chris , and Shalom Lappin . 2005 . Foundations of Intensional
Semantics . Oxford : Blackwell.
and the like. Indeed, he argued, there is no distinction between
sentences true simply in virtue of meaning analytic truths and
other true sentences synthetic truths (Quine 1953 , 1960 ).
Greatly infuenced by Quine, Donald Davidson ( 1967 ) pro-
posed a semantic theory for natural language with no place for
intensions. In a Davidsonian theory, each linguistic expression
is paired with an extension (other terms like referent or semantic
value are often used instead of extension ), and rules of composi-
tion state how the extension of a larger expression is determined
by the extensions of its parts. For example, in one version of the
theory, the extension of John is John, the extension of runs
is the set of running things, and the extension of John runs is a
truth value (see truth conditional semantics ).
Challenging for this approach are the problems that drive
some to posit intensions in the frst place. If Mark Twain and
Samuel Clemens have the same extension, then how to explain
the diferences between (1) and (2) or between (3) and (4)? But
if Mark Twain and Samuel Clemens have diferent extensions,
then how to explain why Mark Twain is Samuel Clemens is
true? (For possible answers see, for example, Larson and Segal
1995 .)
Semantics with Intensions
More commonly, intensions are taken seriously. In Meaning and
Necessity , Rudolf Carnap ([1947] 1956) presented his method
of intension and extension as an improvement over Gottlob
Freges way of distinguishing sense and reference. In Carnaps
system, each meaningful expression is assigned both an exten-
sion and intension. For example, the extension of human is the
class of human beings, the intension of human is the property of
being human, the extension of Walter Scott is Walter Scott, the
intension of Walter Scott is an individual concept of Walter Scott,
the extension of a sentence is its truth value, and the intension
of a sentence is the proposition it expresses. Notably, unlike
Freges senses, Carnaps intensions do not vary with linguistic
context, thereby, according to Carnap, avoiding a serious objec-
tion both to Freges approach and the related proposal by the
infuential logician Alonzo Church ( 1951 ).
In applying a precise distinction between intension and exten-
sion to natural language, Richard Montague ( 1974 ; see montague
grammar ) was particularly infuential. Inspired by Carnap and
Church, Montague took an intension to be a mathematical func-
tion. Tis permitted a graceful way for meanings to combine, as
the application of function to argument. In one version of this
sort of theory ( possible world semantics ), the intension of
a sentence is a function from possible worlds to truth values, and
the extension of that sentence relative to a possible world is the
result of applying its intension function to that possible world. For
example, the intension of the sentence Hong Kong is in China
might be a function yielding the value true for all possible worlds
where Hong Kong is in China and false otherwise. Variations and
improvements to Montagues general approach have dominated
the feld of formal semantics (Gamut 1991 ; Heim and Kratzer
1998 ; Chierchia and McConnell-Ginet 2000 ) .
Complications and Problems
Complications ensue when accounting for context-sensi-
tive expressions like the indexicals I or here or now.
Intension and Extension
398
categories. On the one hand, we have reference , denotation ,
and extension ; on the other hand, we have content , meaning ,
sense , connotation , and intension . Te relationship between
these categories of terms is best illustrated via the intentionality
of linguistic expressions.
Just as names are about, in the relevant sense, the objects for
which they stand, so, one might think, predicates are about
the things of which they are true. Green is about the green
things, happy about the happy things, and so on. Te things
that words are about, in this sense, are their references (deno-
tations, extensions). But, plausibly, a theory of reference for a
language would not be a full account of the content (meaning,
sense) of expressions of the language. To adapt an example from
W. V. O. Quine ( 1953 ), the sentences Dolly is a renate and
Dolly is a cordate may be alike with respect to the reference
of the expressions that compose them (because the set of cor-
dates is identical to the set of renates) even though, intuitively,
the two sentences say diferent things about Dolly. So it seems
that two expressions can have the same reference while difer-
ing in content. But many have thought that, as Gottlob Frege
(1892) suggested, the converse does not hold: two expressions
cant have the same content without also having the same refer-
ence. Intuitively, two sentences cant say the same thing about
the world or express the same thought without being about the
same things. Tis combination of views that the content of an
expression is standardly something over and above its reference,
and that the content of an expression determines its reference
is very widely accepted. (Tough not universally; it is rejected by
defenders of a Chomskyan internalist view of meaning who take
meanings to be internal to the language-processing systems of
language users [Pietroski 2003 ] and by skeptics about content
[Quine 1953 , 1960 ; Kripke 1982 ].)
Tese views about the relationship between content and ref-
erence structure much contemporary work on intentionality, for
if content determines reference, it is natural to think that content
explains reference: Intentional phenomena come to be about
things by virtue of their possessing a content. Tis way of think-
ing about intentionality has several virtues. One is that it seems to
ofer an explanation of the aforementioned example of Zeus; if
aboutness is typically explained by possession of a content, then
perhaps the sense in which Zeus aims to be about something is
that it, like expressions that are genuinely about something, has
a content. Its just that in the case of Zeus, this content fails to
determine a reference.
Virtually nothing more can be said about content, reference,
and the relationship between the two without entering into mat-
ters about which there is not even rough agreement. Teorists
difer about what sorts of things contents are, about whether
there are any expressions for which content and reference
coincide, and about whether there are any kinds of expressions
that cannot possess a content without possessing a reference.
Canonical works on these topics include Frege ( 1892 ), Russell
( 1905 ), Frege (1918), Carnap ( 1947 ), Kripke ( 1972 ), and Kaplan
( 1989 ) .
Intentionality, Intensionality, and Intentions
It is worth mentioning at this point two persistent, though purely
terminological, sources of confusion about intentionality: the
Frege , Gottlob . [1892] 1997 . On sense and reference. In Te Frege
Reader , ed. M. Beaney , 15171. Oxford : Blackwell .
Gamut , L. T. F. 1991 . Logic, Language and Meaning . Chicago : University
of Chicago Press .
Garcia-Carpintero , Manuel , and Josep Maci , eds. 2006 . Two-Dimensional
Semantics . Oxford : Oxford University Press .
Heim , Irene , and Angelica Kratzer . 1998 . Semantics in Generative
Grammar . Oxford : Blackwell .
Kaplan , David . 1989 . Demonstratives. In Temes from Kaplan , ed. J.
Almog , J. Perry, and H. Wettstein , 481563. New York : Oxford University
Press .
Katz , J. J. 1990 . Te Metaphysics of Meaning . Cambridge, MA : MIT Press .
Kripke , Saul . Naming and Necessity . 1980 . Cambridge : Harvard University
Press .
Larson , Rich , and Gabriel Segal . 1995 . Knowledge of Meaning . Cambridge,
MA : MIT Press .
Mill , John Stuart . [1843] 1872 . A System of Logic . 8th ed.
London: Longmans .
Montague , Richard . 1974 . Formal Philosophy: Selected Papers of Richard
Montague , ed. Richmond Tomason . New Haven, CT : Yale University
Press .
Porphyry . 2003 . Porphyrys Introduction . Trans. and commentary by
Jonathan Barnes. Oxford : Oxford University Press .
Putnam , Hilary . 1975 . Te meaning of meaning. In Philosophical
Papers. Vol. 2. Mind, Language and Reality . Cambridge : Cambridge
University Press .
Quine , W. V. 1953 . Two dogmas of empiricism . Philosophical
Review 60 : 20 43. Reprinted in From a Logical Point of View, 2d ed.
( Cambridge : Harvard University Press , 1961 ).
. 1960 . Word and Object . Cambridge, MA : MIT Press .
von Fintel , Kai , and Irene Heim . 2007 . Intensional Semantics . M anuscript,
Massachusetts Institute of Technology .
INTENTIONALITY
Aboutness
Te closest thing to a synonym for intentionality is aboutness;
something exhibits intentionality if and only if it is about some-
thing. Te relevant sense of about is best elucidated by exam-
ple: Te sentence Saul Kripke is a philosopher is about Saul
Kripke; my belief that the weather in South Bend is dreary is about
the city of South Bend, Indiana; the black lines and curving blue
stripe on the map in my hand are about the streets of South Bend
and the St. Joseph River; the position of the needle on the gas
gauge in my car is about the amount of gasoline in its tank. While it
is dif cult to fnd an uncontroversial and illuminating paraphrase
of the relevant sense of about , its hard to deny that there is some
reasonably clear sense of aboutness common to these examples.
Tis characterization of intentionality as aboutness is only
true to a frst approximation because something can exhibit
intentionality without being about anything if it purports to be
about something. Zeus is not about, does not represent, any-
thing; this name, unlike Saul Kripke, does not have a worldly
correlate. Nonetheless, Zeus counts as an example of intention-
ality by virtue of the fact that it (in a dif cult to explain sense)
aims to be about something, even if it does not succeed .
Intentionality, Content, and Reference
Glossing over a wealth of distinctions, we see that the vocabulary
used in discussions of intentionality is divisible into two broad
Intentionality
399
presented, in judgement something is af rmed or denied, in love
loved, in hate hated, in desire desired, and so on.
Tis is characteristic exclusively of mental phenomena. No
physical phenomenon exhibits anything like it. We can, there-
fore, defne mental phenomena by saying that they are those
phenomena which contain an object intentionally within them-
selves. (Brentano [1874] 1997 , II.i.5)
We can think of Brentanos thesis as having two components:
Intentionality is necessary for mentality; all mental states
exhibit intentionality.
Intentionality is suf cient for mentality; everything that
exhibits intentionality is a mental state.
Te claim of necessity is uncontroversial when we are thinking of
propositional attitudes like believing, supposing, and judging. It
is more controversial but still plausible when we think of percep-
tual states; the sense in which my visual experience is currently
of or about a computer screen is recognizably the same as the
sense in which a name is a name of its bearer.
Bodily sensations like itches and pains, however, may seem
to be counterexamples to Brentanos claim that intentionality is
necessary for mentality. My sensation of throbbing pain is clearly
a mental state but can it be said to represent, or be about, any-
thing at all? Many have thought not and have seen the attempt
to fnd intentionality in sensations as an ad hoc attempt to fnd
something common to mental phenomena (Rorty 1979 ). But
this negative verdict can be challenged, and it has been in recent
philosophy of mind. For one thing, pains are felt as located, and
given this, it is not implausible to think of them as about the part
of the body where they are felt to be (Tye 1995 ; Byrne 2001 ).
On the face of it, the other half of Brentanos thesis that inten-
tionality defnes the mental seems to be less well-of. How can
one claim that intentionality is suf cient for mentality when
things that are clearly not mental states like words, parts of
maps, and gas gauges exhibit intentionality?
Original and Derived Intentionality
Te best answer to this question invokes a distinction between
original and derived intentionality. We began by noting the
diversity of things that exhibit intentionality: mental states, lin-
guistic expressions, maps, gas gauges. But it is plausible to think
that at least some of these intentional phenomena acquire this
status via a relation to some other more fundamental intentional
phenomenon. If this is correct, we can recast the second half of
Brentanos thesis as the claim that only mental phenomena have
original intentionality: intentionality not explicable in terms of
other intentional phenomena.
Tis sort of defense of Brentano carries with it a commit-
ment to the research program of explaining the intentionality
of language, maps, and gas gauges in terms of the intentionality
of the mental. Tis research program has considerable promise
and has received sophisticated development over the last few
decades, with most of the attention focused on explanations of
linguistic meaning in terms of mental content.
One well-developed attempt to provide such an explana-
tion begins with the thought that linguistic expressions mean
what they do because of what speakers intend to convey by
distinctions between intentionality and intensionality, on the
one hand, and intentionality and intentions, on the other.
Intensionality is a property of sentence contexts. Given any
context in a sentence, we can then ask: Can we, by replacing
one expression or phrase in that context with another that has
the same reference, change the truth value of the sentences as a
whole? If so, then the context is said to be intensional .
So far, the connection between intentionality and intensional-
ity may seem to be merely orthographic. But it has been claimed
that the latter is a criterion for the former: that descriptions of
intentional phenomena will always include an intensional con-
text (Chisholm 1957 ). For descriptions of propositional atti-
tudes like beliefs, this seems plausible. For example,
John believes that the worlds most famous sheep is famous.
may be true while
John believes that Dolly the sheep is famous.
is false, even if the worlds most famous sheep and Dolly the
sheep have the same reference. But the criterion seems to fare
less well in other cases. For example,
Te thick blue line on my map of South Bend represents the
St. Joseph River.
appears to ascribe the right sort of aboutness to qualify as a sen-
tence about intentionality, but the sentence does not contain
any intensional contexts. And many sentences that do contain
intensional contexts dont seem to be descriptions of intentional
phenomena. For example,
Mammals have a greater chance of heart failure than fatworms
because they are cordates.
and
Mammals have a greater chance of heart failure than fatworms
because they are renates.
difer in truth value, even though cordates and renates have the
same reference .
A second potential source of confusion is the similarity of
intention, and intentionality. Intention , like belief and
desire, is the name of a type of mental state. Like beliefs and
desires, intentions exhibit intentionality, but they are no more
essential to intentionality than other mental states .
Intentionality and Mentality
Tough intentionality is derived from intentio , a technical term
that had wide use in medieval philosophy, and intentio is itself
a translation of technical terms from premedieval Arabic phi-
losophy, modern usage of the term is usually traced to Franz
Brentanos 1874 Psychology from an Empirical Standpoint .
Brentano is standardly taken to have made two basic claims
about intentionality, the frst of which is that intentionality is
internally related to mentality:
Every mental phenomenon is characterized by what we might
call direction toward an object. Every mental phenomenon
includes something as an object within itself, although they
do not all do so in the same way. In presentation something is
Intentionality
400
Te internal states in question will presumably be complex
physical states of subjects. Given this, we can ask: Is the content
of such states derived from the contents of its parts so that, in
the case of a state that has the content that grass is green, the
state would have one part representing grass, and another rep-
resenting the color green or are the fundamental content-con-
ferring properties a matter of the propositional attitude state as
whole? To take the former option is to endorse the language of
thought hypothesis (Fodor 1975 ; Rey 1995 ) and to take the latter
is to reject it (Stalnaker 1990 ; Blackburn 1984 ).
Whether or not the language of thought hypothesis is true,
the principal challenge in constructing a theory of content is to
specify the properties that confer contents on those representa-
tions. Here, the proliferation of theories is such that it is hardly
possible to do better than the following list of candidate comple-
tions of an internal representation x has the content p if and
only if:
x is actually caused by p s being the case / p s being the case
would, under epistemically ideal conditions, cause that inter-
nal state (Stalnaker 1984 ) / x covaries with p s being the case
during the learning period when the state is acquiring a
content (Dretske 1981 ).
It is the biological function of x to be present when p is the
case (Millikan 1989 ).
x has nomological connections of specifed kinds with prop-
erty p (Fodor 1990 ).
Tere is an isomorphism between the system comprised of x
and the rest of the agents internal representations and a sys-
tem containing p which maps x onto p (Cummins 1989 ).
A (specifed) theory maps x s functional role its causal con-
nections to perceptual input, behavioral output, and other
internal representations onto p (Block 1986 ; Harman [ 1988 ]
1999).
Te discussion so far leaves open an important metaquestion
about intentionality: Supposing that there is no reduction of
original intentionality to nonintentional facts, what attitude
should we take toward the claims about the intentionality of
mental states to which we unhesitatingly subscribe in daily life?
Some who have rejected such analyses have put alleged inten-
tional facts into the same category as alleged facts about phlo-
giston, witches, and other posits of false theories (Quine 1960 ;
Churchland 1981 ); others have taken the failure of reductions of
original intentionality to show that intentionality is an unana-
lyzable feature of the world, and no less real for that (Chisholm
1957 ; Searle 1983 ) .
Jef Speaks
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Blackburn , S. 1984 . Spreading the Word . Oxford : Clarendon Press .
Block , N. 1986 . Advertisement for a semantics for psychology. In
Mental Representation: A Reader , ed. S. Stich and T. Warfeld , 81141.
Cambridge, MA : Basil Blackwell .
Brandom , R. 1994 . Making It Explicit . Cambridge : Harvard University
Press .
Brentano , F. [1874] 1997 . Psychology from an Empirical Standpoint .
London : Routledge .
using them (Grice 1957 , 1969 ; Schifer 1972 , 1982 ). On this
view, what a speaker means by uttering an expression on an
occasion (speaker-meaning) is a function of the beliefs that
that speaker intends to bring about in his or her audience via
their recognition of that communicative intention, and, fur-
ther, what an expression means in a community is a function
of what speakers mean or would mean by using the expression
on various occasions. By this two-part reduction (of expres-
sion-meaning to speaker-meaning, and speaker-meaning to
communicative intentions), the intentionality of language is
explained in terms of the intentionality of intentions. Critics
of this approach have focused on its inability to explain uses
of language in thought and apparently normal examples of
communication in which speakers lack the requisite commu-
nicative intentions (Chomsky 1975 ; Schifer 1987 ). But despite
the problems faced by specifc versions of this reductive pro-
gram, there is widespread agreement that there is some way of
explaining the intentionality of language via the intentional-
ity of the mental states of language users if not their inten-
tions, then perhaps their beliefs (Lewis 1975 ). (Opposed views
of the source of the intentionality of language are defended in
Laurence 1996 and Brandom 1994 .)
The Reduction of Original Intentionality
Supposing that there is a genuine distinction between original
and derived intentionality, there is a further question about
whether original intentionality can itself be explained. Te sec-
ond thesis about intentionality often associated with Brentano
is that it cant be: Original intentionality is not only defnitive of
mentality but also inexplicable in nonintentional terms. By con-
trast, the view dominant in recent years may be summed up as
follows:
I suppose that sooner or later the physicists will complete the
catalogue theyve been compiling of the ultimate and irreduc-
ible properties of things. When they do, the likes of spin, charm,
and charge will perhaps appear on their list. But aboutness surely
wont; intentionality simply doesnt go that deep. If aboutness
is real, it must really be something else. (Fodor 1987 , 97)
In part because most recent theorists have adopted the view,
sketched here, that original intentionality is found at the level of
thought, these theorists have approached the task of explaining
original intentionality by constructing theories of mental con-
tent. Te standard method of theory construction takes as given
the following broad thesis: Being in a certain mental state is a
matter of being in an internal state that has properties that make
it a mental state of the relevant type with the relevant content.
Tis view is sometimes called the representational theory of the
mind though this label is sometimes used for the conjunction
of the present view with the language of thought hypothe-
sis (about which more later) and other times is called function-
alism though this label is sometimes used for the conjunction
of the present view with the thesis that the content-determining
properties of internal states are their functional roles.
Te natural next questions are, therefore: What properties of
internal states make them mental states of a certain type with
a certain content? And which internal states are the bearers of
content?
Intentionality
401
. 1982 . Intention-based semantics . Notre Dame Journal of Formal
Logic 23 : 119 56.
. 1987 . Remnants of Meaning . Cambridge, MA : MIT Press .
Searle , J. 1983 . Intentionality . New York : Cambridge University Press .
Stalnaker , R. 1984 . Inquir y . Cambridge, MA : MIT Press .
1990 . Mental content and linguistic form. In Context and Content ,
22540. New York : Oxford University Press .
Tye , M. 1995 . Ten Problems of Consciousness: A Representational Teory
of the Phenomenal Mind . Cambridge, MA : MIT Press .
INTERNAL RECONSTRUCTION
Internal reconstruction (IR) is a method, or group of methods,
used to establish unattested earlier forms of languages. It difers
from the comparative method , which has similar aims, in
being based on the features of one language, without external
reference, and it is therefore appropriately applied to languages
without relatives, such as language isolates or reconstructed
protolanguages. Languages reconstructed by these means are
referred to as prelanguages , as opposed to the protolanguages
that result from the application of the comparative method .
IR arose in the later nineteenth century as an extension of
the comparative method. For example, Ferdinand de Saussures
reconstruction of sonant coef cients later termed laryngeals
for Indo-European was based largely on this method, though it
was not recognized as legitimate at the time, and his reconstruc-
tions were largely ignored. Te method was identifed in the early
twentieth century but was considered too speculative, method-
ologically unsound, and lacking controls on its results. It was not
until documentary evidence for laryngeals was found in Hittite
that its value was recognized, and then not universally .
IR is not a single method but a group of approaches having
in common reliance on internal features of the language. It is
possible to identify three such approaches, which are not nec-
essarily completely distinct nor mutually exclusive: historical
morphophonemics, regularization of systems, and universal and
typological reconstruction.
Historical Morphophonemics
Tis approach, applicable only to phonology , is fairly well
defned methodologically. It is based on the structuralist
principle of the morphophoneme , which also corresponds to
the underlying phoneme of classical generative phonology.
Morphophonemes comprise a set of morphologically related
phonemes , that is, occurring in diferent forms of the same
morpheme . Te principle is that the diferent phonemes result
from the application of diferent phonological changes in difer-
ent contexts; by reducing these alternating phonemes to a single
form, we efectively establish the original phoneme from which
they are derived. For example, in Latin /s/ became /r/ between
vowels, giving alternations such as fos / for -. Synchronically, this
provides us with a morphophoneme {S}; historically, we assume
that this corresponds to an original phoneme /s/. A weakness of
this method is that it assumes that there was no morphophone-
mic alternation in earlier stages of languages .
Regularization of Systems
While historical morphophonemics works by eliminating
alternations and apparent irregularities, the regularization of
Byrne , A. 2001 . Intentionalism defended. Philosophical Review
110 : 199 240.
Carnap , R. 1947 . Meaning and Necessity: A Study in Semantics and Modal
Logic . Chicago : University of Chicago Press .
Chisholm , R. 1957 . Perceiving: A Philosophical Study . Ithaca, NY : Cornell
University Press .
Chomsky , N. 1975 . Refections on Language . London : Temple-Smith .
Churchland , P. 1981 . Eliminative materialism and the propositional
attitudes. Journal of Philosophy 78.2 : 67 90.
Crane , T. 1998 . Intentionality as the mark of the mental. In Current Issues
in Philosophy of Mind , ed. A. OHear, 22951. Cambridge : Cambridge
University Press .
Cummins , R. 1989 . Meaning and Mental Representation . Cambridge,
MA : MIT Press .
Dretske , F. 1981 . Knowledge and the Flow of Information . Cambridge,
MA : MIT Press .
Fodor , J. 1975 . Te Language of Tought . Hassocks, UK : Harvester .
. 1987 . Psychosemantics: Te Problem of Meaning in the Philosophy
of Mind . Cambridge, MA : MIT Press .
. 1990 . A theory of content, II: Te theory. In A Teory of Content
and Other Essays , 89136. Cambridge, MA : MIT Press .
Frege , G. 1892 . On sense and reference. In Translations from the
Philosophical Writings of Gottlob Frege , ed. P. Geach and M. Black ,
5678. Oxford : Basil Blackwell .
. 1918 . Tought. In Te Frege Reader , ed. M. Beaney , 32545.
Oxford : Basil Blackwell .
Grice , P. 1957 . Meaning. Philosophical Review 66.3 : 177 88.
. 1969 . Utterers meaning and intentions. In Studies in the Way of
Words , 86116. Cambridge : Harvard University Press.
Harman , G. [1988] 1999 . Wide functionalism. In Reasoning, Meaning,
and Mind , 23543. Oxford : Clarendon .
Husserl , E. [1901] 2002 . Logical Investigations . 2 vols.
London : Routledge .
Kaplan , D. 1989 . Demonstratives. In Temes from Kaplan , ed .
J. Almog , J. Perry , and H. Wettstein , 481563. Oxford : Oxford
University Press .
Kripke , S. 1972 . Naming and Necessity . Cambridge : Harvard University
Press .
. 1982 . Wittgenstein on Rules and Private Language: An Elementary
Exposition . Cambridge : Harvard University Press.
Laurence , S. 1996 . A Chomskian alternative to convention-based seman-
tics . Mind 105 : 269 301.
Lewis , D. 1975 . Languages and language. In Language, Mind, and
Knowledge , ed. K. Gunderson , 335. Minneapolis : University of
Minnesota Press. Reprinted in Lewis 1983 , 16388.
. 1983 . Philosophical Papers . Vol. 1. Oxford : Oxford University
Press .
Millikan , R. 1989 . Biosemantics . Journal of Philosophy 86 . 6 : 28197.
Moran , D. 1996 . Brentanos thesis . Proceedings of the Aristotelian
Society Supplementary Volume 70 : 1 27.
Perler , D., ed. 2001 . Ancient and Medieval Teories of Intentionality .
Boston : Brill .
Pietroski , P. 2003 . Te character of natural language semantics. In
Epistemology of Language , 21756. Oxford : Oxford University Press .
Quine , W. V. 1953 . Two dogmas of empiricism. In From a Logical Point
of View , 2046. Cambridge : Harvard University Press .
. 1960 . Word and Object . Cambridge, MA: MIT Press .
Rey , G. 1995 . A not merely empirical argument for a language of
thought . Philosophical Perspectives 9 : 201 22.
Rorty , R. 1979 . Philosophy and the Mirror of Nature . Princeton,
NJ : Princeton University Press .
Russell , B. 1905 . On denoting . Mind 14 : 479 93.
Schifer , S. 1972 . Meaning . Oxford : Oxford University Press .
Intentionality Internal Reconstruction
402
where the comparative method cannot be applied. Te meth-
ods and especially the approach that relies on universals and
typology are extremely powerful, and for that very reason they
have often been regarded with some suspicion and must be used
with caution .
Anthony Fox
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Campbell , Lyle . 1998 . Historical Linguistics: An Introduction .
Edinburgh : Edinburgh University Press. See Chapters 7 and 8.
Fox , Anthony . 1995 . Linguistic Reconstruction: An Introduction to Teory
and Method . Oxford : Oxford University Press. See Chapters 7 and 8.
Lehmann , Winfred . 1992 . Historical Linguistics . London : Routledge. See
Chapter 8.
INTERPRETATION AND EXPLANATION
An interpretation is an account of what is said, done, and thought
by some person or people an account at the level of content. An
interpretation may take as its subject some particular episode
involving some agents action. Consider a bosss remarks to a
new employee about her skirt. (Is the boss generally a fashion
maven? Does the boss wear skirts or chase skirts?) What is the
agent saying, doing, and thinking? An interpretation may take as
its subject some kind of utterance or activity in a social group.
Consider some students at a game chanting Were gonna beat
the hell out of you. What were they really saying, doing, and
thinking? An understanding of such matters an interpretation
depends epistemically on an associated explanatory understand-
ing of why such things are said and done .
An interpreted episode or a phenomenon typically involves
some linguistic activity. Yet an interpretation is more than a
straightforward translation for those utterances. Te chanting
of the students can be homophonically translated. But is their
utterance primarily a part of a ritual of hope and group identi-
fcation or an expression of belief? Consider also one historical
episode: George W. Bushs insistance that it is morally wrong
to destroy life in order to save life (quoted in Stolberg 21 May
2005). Again, for speakers of English, homophonic translation is
uncontroversial. Yet just what was expressed? What would his
principle allow, and what would it rule out? What is Bush saying
and doing here?
Interpretation goes beyond translation in part because inter-
pretation comes to terms with pragmatic elements of an
utterance what speech-acts are performed and how con-
versational implicature may outrun what is explicit. An
acceptable treatment of these matters is bound up with rich
understandings of agents situated projects their beliefs and
desires.
An interpretation may constrain the translation of some sub-
jects language on which it depends. Tis is common in anthro-
pological studies of a peoples religion or magic as one
translation may strongly suggest an understanding that is at odds
with an anthropological interpretation, whereas an alternative
translation may support that interpretation. Te classic debates
between symbolist and intellectualist anthropologists over how
to interpret various folk religions had implications for whether
systems method involves, in efect, the generalization of this
process to whole systems. Tus, where there are inconsistencies
in paradigms for example, with diferent classes of nouns or
verbs the method seeks to eliminate these diferences in order
to produce a single paradigm, on the assumption that inconsis-
tencies arise through change. For example, the diferent declen-
sions or conjugations of Latin and Greek or the forms of Germanic
strong verbs can be reduced to single patterns. Again, a weakness
of this approach is that it assumes complete consistency in the
prelanguage .
Universal and Typological Reconstruction
Tis is the most speculative and perhaps methodologically the
least controllable approach, which to some extent goes beyond
purely internal reconstruction inasmuch as it invokes general
properties of languages. Tese properties may be either universal
or typological. In the former case the method relies on language
universals, which are held to determine either the synchronic
structures of languages or the diachronic processes of change
(see synchrony and diachrony ). Tese have the efect of
constraining possible reconstructions since we may not postu-
late earlier states nor processes of change that do not conform to
the universals.
Tus, in phonology we may assume universal phonetic
processes (for example, the palatalization of velar consonants
in the neighbourhood of front vowels); a reconstruction that
assumes the reverse (palatalization with back vowels) would be
ruled out. Similarly, the universal process described by the term
grammaticalization the development of lexical items
into grammatical items precludes reconstructing the reverse.
Tere may also be constraints on structures and systems; if all
languages must have, say, vowels, then we cannot reconstruct
an earlier stage without them. In all such cases, our reconstruc-
tions will seek to establish earlier forms in compliance with
these universal constraints .
Te typological approach (see typology ) is similar, but it
refects language types rather than universal properties. Te
principle here is that language typology involves not just iso-
lated features but, rather, sets of harmonizing or co-occurring
features. According to one widely used typology, languages fall
into two basic types, VO and OV, according to the order of verb
and object, but this ordering is also refected in the order of other
items, such as noun and adjective (VO languages generally also
have NA, and OV languages have AN), and the occurrence of
prepositions or postpositions (VO languages favor the former,
OV languages the latter). Given such principles of harmony , a
language that has one such order, say, NA, will be expected to
have the other harmonic orders; if it does not, we can reconstruct
this harmony for an earlier period of the language. In the case
of English, for example, which has VO but AN, we can assume
an earlier stage that had either NA or OV (the latter is assumed).
Other typological parameters can be used in an analogous fash-
ion. A weakness of this method, however, is that it assumes that
languages will originally have been typologically consistent,
though consistency is not necessarily an inherent attribute of
language.
In spite of potential weaknesses, the methods of IR have
proved useful in establishing earlier stages of languages in cases
Internal Reconstruction Interpretation and Explanation
403
more, bears on which of several possible interpretations is ulti-
mately most satisfactory. Bush might be understood as having
advanced a moral principle that is implicitly qualifed so that
there is no inconsistency with his military policies. Alternatively,
there might have been unnoticed inconsistency as all humans
exhibit some inconsistency. Finally, one might consider whether
Bush might have noticed an inconsistency that he conveniently
omits to mention .
Te central point concerns how diverse information allows
one to decide among such alternative interpretations: It bears
on the explanation of what is said, done, and thought in the epi-
sode interpreted. Te political concerns of Bush and his advis-
ers might help to explain the assertion as political boilerplate.
Certainly the staking out of political positions, the responsive-
ness to constituencies, and the like can lead politicians to cast
about for simple (simplistic) formulations of sweeping principles
in which to wrap the desired policy. Is this the explanation, and
interpretation, of Bushs assertion? Understood and explained
as boilerplate, these questions about the exact content of the
asserted principle (whether or how it is implicitly qualifed), the
way it then squares with various policies, and the obviousness of
inconsistency (if any) may become less pressing as they matter
less to the explanation of the episode. What is then central is the
sense of certain politicians for how taking the moral high road
will play in the relevant constituencies. But perhaps one senses
that Bush is a politician whose public face is here tied to his
own moral view of the world. Ten, the degree of inconsistency
and how he could be insensitive to it become more central as a
matter of explanatory concern. In either case, one is informed
both by a general sense for what makes humans tick (cogni-
tively and otherwise) and for an antecedently formed impres-
sion of this person in particular by the sedimented results of
past interpretations. Tis is to draw upon generic resources for
explaining human action and thought and on resources more
specifc to this individual. Again, each is the fruit of past prac-
tice that is both interpretive and explanatory. One has a sense for
how humans think and act and for variations. One has a sense
for Bushs character as one variation, one that is the product
of his biography, and one that is evinced in past interpreted and
explained practice.
Just how one ends up interpreting the assertion in question
and the associated political or moral act depends on the explana-
tion for this episode that one judges to be most likely, given all that
one knows about the agent or agents involved and about humans
as agents . Interpretation, it seems, is a matter of abduction or
inference to the best explanation of an ongoing, always revis-
able, sort. (Readers should fnd it easy to explore many further
alternatives and wrinkles in understanding the Bush case and
to assure themselves that their plausibility devolves onto the
question of greater or lesser explicability.)
Te interdependence of interpretation and explanation, their
being two faces of the same coin, is widely appreciated. What is
contested is how to understand the explanation of thought and
deed. Tere are two broad schools of thought here, although rep-
resentatives of each have been diverse.
Some think of explanation as subsumption as a matter of
deploying generalizations to show that the case or phenomenon
in question was dependent on certain antecedents. To conceive
the associated linguistic constructions should be translated into
terms that suggested theories about causes (as intellectu-
alists such as Horton 1970 , 1982 urge), or into more guarded
terms that suggested symbolic expression (as symbolists such as
Leach 1954 and Beattie 1970 urge) perhaps understandings
and infuences. Here, translation (treating the literal said) and
interpretation (of what is ultimately said, done, and thought) are
interdependent. (Turner 1980 ; Henderson 1993 ; Risjord 2000 ;
Stueber 2006 ). As Donald Davidson ( 1984 ) argued, in settling
upon an interpretation and translation, several matters must
be sorted out together interrelated matters having to do with
belief, desire, and meaning (see meaning and belief , radi-
cal interpretation, agreement maximization, and
charity, principle of ) .
What, then, is the mark of a good interpretation? Surprisingly,
many theorists, both historical and contemporary from various
traditions, can be seen to agree on an answer to this epistemo-
logical question: A successful or adequate interpretation afords
an explanatory understanding of what is said, done, and thought.
An interpretation of someone or some folk as doing or saying
some sort of thing is of a piece with an explanatory understand-
ing. Epistemologically, the two stand or fall together. Te agree-
ment here is pervasive but does not run deep. Writers quickly
come to difer over what makes for a successful explanation of
thought and deed and, thus, what marks good interpretation.
(For example, those urging a strong principle of charity such
as Davidson typically think of explanation here as a matter of
exhibiting rationality, while others may see more place for expla-
nations that do not rationalize. Various conceptions of what
make for explanatory understandings are discussed later.)
Focus for a time on the agreement. Consider again Bushs
assertion that it is wrong to destroy life in order to save life. Te
frst thing to notice is the wide range of information that we rec-
ognize as relevant to its interpretation. It was uttered in the con-
text of legislation regarding research using embryonic stem cells.
In keeping with the venerable discussions of hermeneutics (see
philology and hermeneutics ), one should consider what
was also said in the wider context in which the assertion was
advanced. Its interpretation is bound up with the interpretation
of the whole of Bushs remarks which, in turn, is dependent
on successful treatment of the various component utterances.
Information having to do with much beyond this set of remarks is
also relevant. For example, how did Bush and his political advis-
ers understand their political situation? Tat Bush was politically
beholden to right-wing conservative Christians even was one
is relevant. Tat they hold that human life begins at conception is
relevant. So also is information regarding the wider political con-
text. Was the conservative Christian component of his political
base at that time disenchanted and thought to need fring up?
Sweeping moral statements are commonly best interpreted as
containing an implicit ceteris paribus clause. One might wonder
if Bushs statement should be understood as likewise qualifed.
What is known about whether Bush would authorize the military
to destroy lives to save lives? How does Bush understand col-
lateral damage in warfare? Is the military context one in which
ceteris is not paribus ? Is the principle unqualifed except that
it expresses a prima facie duty that can be overridden by other
duties? But what other duties? All this information, and much
Interpretation and Explanation
404
To illustrate, economic theory might inform our explanation
and interpretation of an individuals verbal exchange with his or
her broker as the reception of and risky use of insider information
to avert a loss, interpreting and explaining this as an expectable
form of proft maximization in light of the agents understand-
ing of situated risks of detection. (Perhaps the agent said, Tat
is unacceptable exposure to risk. Te explanation envisioned
supports interpreting this as an instruction to sell, rather than as
a merely general point about levels of risk in certain portfolios.)
Without drawing on theory, one might imaginatively put oneself
in the agents shoes (under tentative interpretation) and see that
one would do the same as the simulation theorist envisions. In
both cases, one appreciates that with diferent information and
antecedent beliefs and values, the agent would have done system-
atically otherwise. Tus, generalizations about economic agents
and imaginative simulation might support a single understand-
ing and explanation of the agent answering the why-question by
attention to the same dependencies (under interpretation).
If explanation is understood as a matter of answers to how-
and why-questions (and associated what-if-things-had-been-
diferent questions), then what have been taken to be competing
accounts of explanation can be understood in a fashion that
renders them compatible, and afords us a multifaceted under-
standing of the explanatory practice associated with interpreta-
tion. Te pivotal move is to abandon each tradition insofar as
it attempts to delimit what counts as an explanation treating
the erotetic understanding as the more generic and acceptable
account of explanation . Ten, approaches such as Woodwards
(appealing to subsumption under generalizations with signif-
cant invariance) and the simulation theorists (appealing to the
imaginative use of our of-line cognitive capacity) can each be
recognized as part of a full epistemic story. Subsumption and
simulation each have their role in coming to appreciate why the
agent or agents acted thus and what they would have done if
things had been relevantly diferent. Tus, my suggestion is that
explanation is the mark of a good interpretation, that explana-
tion typically itself supposes a tentative interpretation, that the
generic understanding of explanation is that provided by erotetic
logic, and that it commonly is attained by some amalgamation of
the epistemic resources that have been of concern in traditions
that have wrongly claimed to provide the account of explana-
tion in the human sciences .
David Henderson
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Beattie , J. H. M. 1970 . On understanding ritual. In Rationality , ed. Bryan
Wilson , 24068. Worchester, UK : Basil Blackwell .
Collingwood , R. G. 1946 . Te Idea of History . Oxford: Clarendon.
Davidson , Donald . 1984 . Belief and the basis of meaning. In Inquiries
into Truth and Interpretation , 14154. Oxford : Oxford University
Press .
Hempel , C. 1965 . Aspects of Scientifc Explanation and Other Essays . New
York : Te Free Press .
Henderson , David . 1993 . Interpretation and Explanation in the Human
Sciences . Albany : State University of New York Press .
Horton , Robin. 1970 . African traditional thought and Western sci-
ence. In Rationality , ed. Bryan Wilson , 13171 . Worchester, UK : Basil
Blackwell .
of the explanation of thought and action along these lines is to
generalize a common understanding of scientifc explanation. C.
Hempel (1965) provides a particularly clear variant in his vener-
able hypothetical-deductive and statistical-probabilistic models
of explanation. (However, there are reasons to think that the
approach to scientifc explanation is itself fawed; see, for exam-
ple, Salmon 1989 .) A very diferent subsumptive model is pro-
vided by James Woodwards ( 2000 ) discussions of explanation in
the special sciences. Woodward thinks that the generalizations
produced and deployed are themselves not exceptionless nomic
generalizations, but relatively invariant generalizations holding
within imperfectly specifed limits .
Te second school of thought views explanation of thought
and action as revealing intelligibility explanation is understood
as a matter of understanding what is said and done as having or
expressing a signifcance or meaning so that the whole thereby
becomes intelligible. Just what being intelligible comes to is
itself understood variously. Certainly a kind and degree of inter-
nal contentful coherence plays a role as does the interpreters
ability to then see how and why one would do as indicated on the
basis of such reasons. Tose who draw on the hermeneutic tradi-
tion are representatives. So also is R. G. Collingwood ( 1946 ) with
his conception of explanation as reenactment. Something like
Collingwoods approach has enjoyed a contemporary revival of
sorts within cognitive psychology where there has been much
work on explanation of thought and deed as a kind of simula-
tion in which ones own cognitive processes are taken of line
and put to work on imagined input that refects ones provisional
interpretation of a subject. One imaginatively puts oneself in
the others shoes and deliberates or reasons; if ones simula-
tion then accords with observed actions or expressed beliefs, one
has a prima facie successful explanation (see Stueber [ 2006 ] for a
recent overview and discussion) .
As noted, the representatives of each approach have been
diverse. Still, we can appreciate that while proponents of these
approaches difer over what makes for explanation, they com-
monly understand that interpretation epistemically depends
on explanation. (It is worth adding that most would also rec-
ognize that the information supposed in an explanation itself
depends on interpretations so that there is ultimately a holistic
epistemic interdependence here.) A unifed understanding can
then be had by way of drawing on a very general understanding
of explanation. When we do this, elements of the diverse inter-
pretive/explanatory approaches can be understood as comple-
mentary epistemologies of explanation, rather than as competing
accounts of what it is to explain thought and action. Let me
explain by drawing on what has come to be called the erotetic
account of explanation (see Salmon 1989 ; van Fraassen 1980 ).
Tis label refers to the logic of questions and answers erotetic
logic. On this approach, an explanation is an answer to a ques-
tion typically a why-question or a how-question. In either case,
the resources for answering the question allow one to under-
stand a pattern of dependencies. Te resources for answering a
why-question (let us focus on these) should allow us to appreci-
ate that what is done or said was dependent on certain standing
states and events. If we can do this, we can answer a range of
associated what-if-things-had-been-diferent questions explor-
ing these dependencies (see possible worlds semantics ) .
Interpretation and Explanation
405
same person writes two diferent texts for himself when reading
from diferent interpretive communities, for he understands the
Hebrew texts as prophesies of Jesus Christ only after his conver-
sion at Damascus .
Fishs theory has been criticized for making words have no
meaning. He responds with just the opposite: Words always have
meaning, in fact many meanings, all of which are constructed by
situated readers in various communities. Fish adds that his the-
ory is sociological, not normative, that is, it describes only what
people say (or think) a text means; it does not prescribe how
we ought to interpret texts. Finally, to the objection that some
authors use certain techniques to ensure that their texts convey
certain meanings, he responds that those meanings come to frui-
tion only if the reader belongs to the same interpretive commu-
nity as that author .
Jefrey R. Wilson
WORK CITED
Fish , Stanley Eugene . 1980 . Is Tere a Text in Tis Class? Te Authority of
Interpretive Communities. Cambridge : Harvard University Press.
INTERTEXTUALITY
Building on Mikhail Bakhtins ( 1981 ) discussion of the dialogic
nature of language, Julia Kristeva ( 1986 ) coined the term inter-
textuality for the multiple ways in which texts refer to and draw
on other texts. Tis notion highlights the interconnectedness of
texts and challenges deep-rooted literary values, such as auton-
omy, uniqueness, and originality (Allen 2000 , 56). An inter-
textual perspective views text production as a social practice in
which diferent texts, genres, and discourses are drawn upon and
text consumption as a process in which readers may bring addi-
tional texts not only those that have shaped production into
the interpretation process (Fairclough 1992 , 845). Te study
of intertextuality does not focus solely (or even primarily) on
the specifc prior texts that are brought into play in a given text;
rather, it also examines the implicit texts underlying production
and interpretation (e.g., presuppositions, genre conventions)
(Culler 1976 , 1388). Tus, a newspaper crime report has inter-
textual links not only to eyewitnesses accounts and previous
reports on the same and/or similar events but also to newswrit-
ing conventions, propositions that the journalist takes as given,
and even the journalists/readers understanding of crimes in
general .
Reported speech, a prime example of intertextuality, has been
extensively studied in sociolinguistics . Reporting speech is
always a reformulation of the original act. Even if prior speech is
reported verbatim, the reporting speaker may use prosodic fea-
tures like stress and intonation to indicate his/her interpre-
tation of the utterance, or he/she may frame the reported speech
in such a way as to manipulate the addressees perception of the
reported speaker. In some cases, material represented as reported
speech is not spoken by anyone at all. Tese observations have
led Deborah Tannen ( 1989 ) to conclude that reported speech is
primarily the creation of the reporting speaker and serves to cre-
ate a sense of interpersonal involvement between the reporting
speaker and the addressee in the reporting context .
. 1982 . Tradition and modernity revisited. In Rationality and
Relativism , ed. Martin Hollis and Steven Lukes , 20160. Cambridge,
MA : MIT Press .
Leach , E. R. 1954 . Te Political Systems of Highland Burma . London : Bell.
Risjord , Mark. 2000 . Woodcutters and Witchcraft . Albany : State University
of New York Press .
Salmon , Wesley. 1989 . Four decades of scientifc explanation. In
Scientifc Explanation, Minnesota Studies in the Philosophy of Science .
Vol. 13. Ed. P. Kitcher and W. Salmon , 3219. Minneapolis : University
of Minnesota Press .
Stolberg , Sherly Gay . In Rare Treat, Bush Vows Veto of Stem Cell Bill.
New York Times , 21 May 2005 . Stueber , Karsten . 2006 . Rediscovering
Empathy . Cambridge, MA : MIT Press .
Turner , Stephen . 1980 . Sociological Explanation as Translation .
Cambridge : Cambridge University Press .
van Fraassen , Bas . 1980 . Te Scientifc Image . Oxford : Oxford University
Press .
Woodward , James . 2000 . Explanations and invariance in the special
sciences . British Journal for the Philosophy of Science 51 : 197 254.
INTERPRETIVE COMMUNITY
In 1976, literary critic Stanley Fish used this term to describe
the unspoken (often unknown) alliances among readers who
share similar strategies for determining what a text means. Tis
theory of pragmatics, he says, is the explanation for both the
stability of interpretations among diferent readers (they belong
to the same community) and for the regularity with which a sin-
gle reader will employ diferent interpretive strategies and thus
make diferent texts (he belongs to diferent communities) (Fish
1980 , 171) .
Te notion of interpretive community insists upon the pri-
macy of situated readers, and it can be thought of as a theory
of creative reading. Fish says that a set of general assumptions
on how one ought to interpret a text precedes every act of inter-
pretation; thus, a reader always perceives a given text within an
already in-place hermeneutical framework. One does not read
the words on a page and then decide what those words mean
because no temporal separation exists between acts of percep-
tion and interpretation. Instead, ones community conditions
how its members read those words in the frst place. As such,
readers actually write a text for themselves as they read, for they
have a tool kit of interpretive strategies always at work determin-
ing what certain words will mean should they arise in a given
context. Readers using the same tool kit belong to the same
community.
One can see interpretive communities at work in Christian
typology , a mode of biblical exegesis that aims to square Old
Testament texts with the events recounted in the New Testament.
For the typologist, the belief that Jesus was God combines with
other assumptions in order to form the exegetes set of interpre-
tive strategies. Other readers who share these strategies make up
this exegetes community even if they do not know one another,
which explains how two Christians might independently inter-
pret some events in the Old Testament as prophesies of Jesus
Christ. Of course, a Jew, Gnostic, or pagan produces a much dif-
ferent meaning of those same Hebrew texts because he or she
works from a community that reads/writes those texts difer-
ently. And fnally, a look at Paul of Tarsus demonstrates how the
Interpretive Community Intertextuality
406
only because of the stereotype of Spanish speakers as lazy
and procrastinating .
Several issues continue to dominate sociolinguistic research
on intertextuality. One, to which I have alluded, focuses on
the relation of intertextuality to power. Others are concerned
with the conditions that make decontextualization and recon-
textualization possible, as well as the semantic and functional
changes that texts undergo as a result of recontextualization.
Intertextuality also raises interesting issues about authorship.
If all texts are created out of prior texts and conventions, what
is an author, and who is responsible for what is said/written?
Tese issues are likely to be worked out diferently in diferent
cultures .
Andrew Wong
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Allen , Graham . 2000 . Intertextuality . London : Routledge .
Bakhtin , Mikhail . 1981 . Te Dialogic Imagination . Ed. Michael Holquist,
trans. Caryl Emerson and Michael Holquist . Austin : University of Texas
Press .
Bauman , Richard . 2004 . A World of Others Words: Cross-Cultural
Perspectives on Intertextuality . Malden, MA : Blackwell .
Briggs , Charles , and Richard Bauman . 1992 . Genre, intertextuality, and
social power . Journal of Linguistic Anthropology 2 : 131 72.
Culler , Jonathan . 1976 . Presupposition and intertextuality .
MLN: Modern Language Notes 91 : 1380 96.
Fairclough , Norman . 1992 . Discourse and Social Change . Cambridge,
UK : Polity .
Hill , Jane . 1993 . Hasta la vista, baby: Anglo Spanish in the American
Southwest . Critique of Anthropology 13 : 145 76.
Kristeva , Julia . 1986 . Word, dialogue, and the novel. In Te Kristeva
Reader , ed. Toril Moi, 3461. New York : Columbia University Press .
Talbot , Mary . 1995 . A synthetic sisterhood: False friends in a teenage
magazine. In Gender Articulated , ed. Kira Hall and Mary Bucholtz ,
14365. Routledge : New York .
Tannen , Deborah . 1989 . Talking Voices . New York : Oxford University
Press .
INTONATION
this term refers to the fundamental frequency (or its percep-
tual correlate, pitch ) contour associated with phrases and other
large prosodic units. Language communities use intonation to
serve a wide range of functions, both grammatical and discourse
based. For example, intonation is used to signal prosodic bound-
aries: Te ends of utterances are characteristically associated
with terminal pitch excursions, either a rise or a fall depending
on semantic factors. Another function of intonation is to cue
many types of semantic distinctions, such as the diference
between yes/no questions and neutral declarative statements
in many languages. Intonation is also used to convey emotional
(see emotion and language ) and expressive states, as well as
pragmatic information .
Intonation is a universal property in that speakers of all lan-
guages manipulate pitch to communicate linguistic and para-
linguistic functions (see paralanguage and phonology,
universals of ). Even languages that use pitch to diferentiate
individual lexical items, for example, tone languages such as
Mandarin Chinese and pitch accent languages such as Swedish,
Reported speech is an example of what Norman Fairclough
( 1992 ) calls manifest intertextuality. Manifest intertextuality
refers to the way in which specifc other texts are overtly drawn
upon within a text. In addition to reported speech, it covers such
phenomena as irony, negation , and presupposition . In
contrast, constitutive intertextuality also known as interdiscur-
sivity refers to the way in which texts draw on abstract sets of
conventions like genres and styles. In their research on inter-
discursivity, several critical discourse analysts have
noticed a widespread appropriation of conversational styles in
public discourse. Focusing on a consumer feature about lipstick
from a British teenage magazine, Mary Talbot ( 1995 ) examines
how the text producer exploits features of conversational speech
(e.g., the pronoun you , expressive punctuation) to establish an
informal, friendly relationship with the reader. Tis practice,
however, is far from benign. Under the guise of ofering sis-
terly advice, the consumer feature serves as covert advertising
and trains teenage readers to become consumers of cosmetic
products .
Generic intertextuality, a notion developed by anthropologists
Richard Bauman and Charles Briggs , can be viewed as a particu-
lar kind of interdiscursivity. Tey defne generic intertextuality as
the construction of the relationship between a text and a genre,
and they are interested in how and for what purposes this rela-
tionship is established in communicative practice (Briggs and
Bauman 1992). Tey see genre as a speech style that serves as an
orienting schema for the production and reception of discourse.
Genre interacts with such factors as the interactional context and
the speakers/writers communicative goals in shaping a given
text. In turn, these factors may lead to the selective adoption of
the constituent features of the generic framework and create an
intertextual gap between the text and its generic model. Briggs
and Bauman argue that research on strategies for manipulating
intertextual gaps between texts and their generic schemas can
shed light on issues of power, ideology , and political economy.
In cultures with traditional genres that are invested with great
power, speakers/writers often minimize the distance between
their texts and these genres. Tis serves as a powerful strategy
for creating textual authority. At times, however, speakers/writ-
ers may maximize the intertextual gaps between texts and their
generic models. Tey may do so to resist the hegemony of estab-
lished genres or to claim authority in cases where creativity is
highly valued .
Appropriation, a specifc case of intertextuality, refers to the
practice of adopting words, expressions, or ways of speaking
that are generally thought to belong to someone else. Many
white American teenagers use elements of African-American
vernacular English (AAVE) in their speech so as to align them-
selves with hip-hop culture and/or to project an urban youth
identity by exploiting certain connotations of AAVE (e.g., tough-
ness). Appropriation, however, may also serve disaf liating or
even denigrating purposes . In the United States, some monolin-
gual Anglos use what anthropologist Jane Hill ( 1993 ) call Mock
Spanish that is, a sub register of colloquial English made
up of (pseudo-)Spanish expressions (e.g., maana ) to proj-
ect a congenial persona. Yet to make sense of Mock Spanish,
one also requires access to certain racist beliefs about Spanish
speakers. Maana works as a humorous substitute for later
Intertextuality Intonation
407
(Dutch, English, and German), tone languages (Cantonese and
Mandarin Chinese), pitch accent languages (Japanese, Swedish,
and Serbo-Croatian), languages lacking word-level stress, lex-
ical tones, or pitch accents (Korean and French), and indige-
nous languages of North America and Australia (Chickasaw and
Bininj Gun-wok, respectively) .
An issue common to the intonation systems of all languages
is the mapping between intonational tunes and their meanings.
Because intonation is used to convey many subtle diferences in
meaning, often in gradient fashion, it is a challenge to determine
which diferences in intonation merit diferent phonological anal-
yses. Phonological distinctions between intonational tunes must
be captured as diferences either in the sequence of tones com-
prising the tunes or in the alignment of those tones with words .
Yet another topic inextricably linked to intonation is pro-
sodic constituency. Since pitch excursions are often observed
at prosodic boundaries, a comprehensive analysis of intona-
tion hinges on the characterization of the types of constituents
constituting utterances. Research has shown that prosodic con-
stituency and the mapping between constituency and intonation
vary from language to language. For example, some languages
divide utterances into groupings of words that are characterized
by a tonal template. Tus, phrases in Chickasaw (Gordon 2005 )
are associated with a LHHL sequence, whereby the frst and the
last low tonal targets associate with the beginning and the end of
phrases, respectively, and the two high tones associate with the
second syllable (or the frst syllable if it contains a long vowel or
ends in a sonorant consonant) and the beginning of the fnal syl-
lable, respectively .
A number of researchers have published books providing
overviews of these and other issues in the study of intonation,
including Ladd ( 1996 ), Cruttenden ( 1997 ), and Gussenhoven
( 2004 ) .
Matthew Gordon
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Cruttenden , Alan . 1997 . Intonation . Cambridge : Cambridge University
Press .
Gordon , Matthew. 2005 . Intonational phonology of Chickasaw. In Jun
2005b, 30130.
Gussenhoven , Carlos . 2004 . Te Phonology of Tone and Intonation .
Cambridge : Cambridge University Press .
Hirst , Daniel , and Albert Di Cristo , eds. 1998 . Intonation Systems: A Survey
of Twenty Languages . Cambridge : Cambridge University Press .
Jun , Sun-Ah . 2005 a. Korean intonational phonology and prosodic tran-
scription. In Jun 2005b, 20129.
Jun , Sun-Ah , ed. 2005 b. Prosodic Typology the Phonology of Intonation
and Phrasing . New York : Oxford University Press .
Ladd , D. R. 1996 . Intonational Phonology . Cambridge : Cambridge
University Press .
Pierrehumbert , Janet . 1980 . Te phonology and phonetics of English
intonation. Ph.D diss., Massachusetts Institute of Technology.
Reproduced by the Indiana University Linguistics Club, Bloomington,
1987 .
IRONY
Tere are several types of irony described in the literature, all of
which rely on an incongruity or discrepancy between appearance
also have intonation systems that are evident when words are
grouped into larger prosodic constituents or uttered in isolation.
Intonation systems may vary substantially, however, from lan-
guage to language and also potentially from speaker to speaker.
Tus, while a yes/no question is associated with a terminal rise in
pitch in many languages (e.g., German, Japanese, and Korean),
there are other languages (e.g., Finnish and Chickasaw) that
mark yes/no questions with a fnal pitch fall. One nearly univer-
sal property, however, is the lowered pitch characterizing the
end of semantically neutral declarative utterances .
Te study of intonation has witnessed many important theo-
retical advances over the last 30 years. Whereas certain schools
of intonation (e.g., the British school) described pitch contours
in terms of their overall shape or gestalt, many current models
of intonation capture changes in pitch in terms of discrete tonal
sequences, thereby bringing the study of intonation into line with
the analysis of segmental and word-level phonological phenom-
ena. Tis type of approach assumes that peaks and troughs in a
fundamental frequency contour are attributed to phonological
high and low tones aligned with various phonological elements.
Actual surface phonetic intonation patterns result from inter-
polation between these high and low tonal targets.
As one of the pioneering works in this school of intonation,
Janet Pierrehumbert ( 1980 ) developed an analysis of English
intonation in which a wide range of pitch contours conveying
several diferent semantic and pragmatic functions are captured
as a sequence of phonological high and low tones associated
with hierarchically arranged prosodic constituents . A funda-
mental insight of Pierrehumberts analysis is that tones may be
classifed into two groups: those that are associated with promi-
nent, that is, stressed , syllables and those that are associ-
ated with the periphery, especially the end, of prosodic domains.
Tones that are associated with stressed syllables are termed pitch
accents. Pitch accents difer according to whether they consist of
a single tone, high or low, or a sequence of tones, such as H + L
or L + H, which phonetically yield a tone fall or rise, respectively.
In addition to pitch accents on certain stressed syllables, tonal
excursions are often observed at the end (and potentially the
beginning) of phrases. Tese phrase-level tonal movements are
attributed to boundary tones, which may be associated with rela-
tively small phrases, termed intermediate phrases , or with larger
phrases, termed intonation phrases or intonation units . Like
pitch accents, boundaries may be characterized by a single tone
or a sequence of tones. For example, intonation phrase boundar-
ies in Korean consist of as many as fve tonal targets, for example,
LHLHL, which conveys a sense of annoyance on the part of the
speaker (Jun 2005a) .
One of the challenges facing linguists interested in the
typological investigation of intonation is the relative dearth
of reliable descriptions of intonation on a broad cross sec-
tion of languages. Fortunately, recent years have witnessed
a dramatic expansion of cross-linguistic studies intonation.
Daniel Hirst and Albert Di Cristo ( 1998 ) studied intonation
in 20 languages, including several non-Indo-European lan-
guages. Jun (2005a) compiled investigations of 13 languages
all analyzed within a Pierrehumbert-type framework. A typo-
logically and geographically diverse array of languages is dis-
cussed in this work, including languages with word-level stress
Intonation Irony
408
(1995, 241). One cannot claim that the opposite counterpart to
the literal expression is that the kings were not celebrating with
Te Deums or that the irony can be substituted by a literal expres-
sion of the opposite, namely, that the kings were bewailing their
defeat with lamentations.
Empirically, it has been shown that the processing of state-
ments in a discourse context that emphasizes the irony is not
slower than that observed for the same statement in a discourse
that is consistent with its literal sense, a fnding incompatible
with predictions arising from the standard pragmatic approach.
Moreover, there is empirical evidence incompatible with the
notion of rejection and substitution of the literal, demonstrat-
ing instead that the diference between the literal and nonliteral
sense is important in determining the magnitude of the per-
ceived irony (the tinge hypothesis of Ellen Winner) and that rec-
ognition of irony requires the processing of both of the opposing
meanings in order to determine that the two messages are in an
ironic relation (the indirect negation hypothesis of Rachel Giora
[ 2003 ]) .
Other theories have de-emphasized the importance of the
literal expression as well and accord greater importance to psy-
chological factors. With pretense theory, there is the recognition
that the ironist is taking on the role of a person who holds the
opinion expressed in the irony, thus mocking both the opin-
ion and the people who would hold it. Two competing theories
( echoic mention and echoic reminder ) share the assumption that
a verbal utterance can be seen as a mention or an allusion about
the expression, about expectations, or about cultural norms. In
this way, the ironist communicates his or her attitude about the
actual and expected state of afairs (see Gibbs 1994 , Chapter 8,
for a review) .
Extant theories have been criticized rightfully for their
inability to encompass all types of irony and for a theoretical
vagueness that make them of questionable scientifc utility.
Te importance accorded background context or identifcation
of an ironic environment is especially problematic for process
and computational models of irony comprehension, given the
failure to identify any signal of irony that is both necessary and
suf cient, though recent models based on constraint satisfac-
tion or graded saliency principles are encouraging (see Colston
and Katz 2004 )
Albert N. Katz
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Colston , H. , and A. Katz . 2004 . Figurative Language Processing: Social and
Cultural Infuences . Mahwah, NJ : Erlbaum .
Gibbs , R. 1994 . Te Poetics of Mind . Cambridge : Cambridge University
Press .
Giora R. 2003 . On Our Mind: Salience, Context and Figurative Language .
Oxford : Oxford University Press .
Grice , H. P. 1975 . Logic and conversation. In Syntax and
Semantics: Speech Acts , ed. P. Cole and J. Morgan , 4158. New
York : Academic Press .
Hutcheon , L. 1994 . Ironys Edge . London : Routledge .
Katz , A., ed. 2000 . Uses and processing of irony and sarcasm . Metaphor
and Symbol 15 (Special Issue) : 1 116.
Sperber , D. , and D. Wilson . 1995 . Relevance: Communication and
Cognition . 2d ed. Oxford: Blackwell .
and reality. In dramatic irony, for instance, the incongruity is cre-
ated by having the audience aware of information about which a
character in a play is ignorant (such as in Sophocles Oedipus Rex ) .
In situational irony and irony of fate, the disconnect is between
ideal expectations of justice and actual (or idealized) outcomes,
such as would occur if Bill Gates, one of the worlds most wealthy
individuals, won a lottery or as exemplifed by Beethovens inabil-
ity, on going deaf, to hear his own musical masterpiece .
Te form of irony most studied in the language sciences
is verbal irony, traditionally conceptualized as a trope in
which the meaning one intends to communicate is opposite
of that expressed literally by the words that are used. Tus,
in Shakespeares Julius Caesar , when in his famous solilo-
quy (Act 3, Scene 2) Anthony states: For Brutus is an honor-
able man / So are they all, honorable men, it is understood
that Anthony is emphasizing the opposite of being honorable,
namely, the dishonorable action of Brutus and the other con-
spirators. In principle, with irony one should be able to convey
negative attitude by expressing something positive (as in the
Shakespearean example above) or positive attitude by stating
something negative.
Tere is also a general recognition that because the expressed
utterances are literally plausible, the recovery of ironic intent is
consequently highly context dependent and facilitated by sig-
nals fashioned (or unintentionally employed) by the ironist.
Tone of voice is one such hint in spoken language, but because
irony can be detected even when ironic intonation is not
employed (such as when reading text), this cue is not necessary.
Other cues include hyperbole, understatement, and excessive
politeness , but it is generally agreed that there is no signal that
points exclusively to irony.
Te context necessary for the recovery of ironic intent tradi-
tionally has been limited to the discourse in which the irony is
embedded, but in more recent years, the concept of context has
been widened, even for verbal irony, to encompass an ironic
environment that includes social-cultural factors, such as those
dependent on discursive communities that share knowledge,
beliefs, values, and communicative strategies (e.g., Hutcheon
1994 ) .
Over the past three decades, the standard pragmatic
approach to nonliteral language processing (see metaphor )
has framed much of the discussion on irony. Based on the prag-
matics of conversation and on speech-act theory, irony is
described accordingly as the outcome of a conversational
implicature initiated by a violation, or based on exploita-
tion, of H. P. Grices ( 1975 ) Maxim of Quality: Te recognition
that the literal expression does not make sense in the context
in which it is produced leads one to initiate a search for a con-
text-appropriate interpretation in which the literal sense of the
expression is denied, suppressed, and replaced by the logical
opposite .
Tere have been challenges over the years regarding the tra-
ditional emphasis on the initial processing of the literal expres-
sion and then substituting it with its opposite meaning, on both
logical and empirical grounds. Consider, for instance, the analy-
sis of a passage from Voltaires Candide given by Dan Sperber
and Deirdre Wilson : When all was over and the rival kings were
celebrating their victory with Te Deums in the respective camps
Irony
409
language is not. No one without the idea of reading and writing
could learn the language of mathematics. Symbolic language is
essentially written .
A second, related diference between the two sorts of language
is that natural language is enormously versatile. One can do all
sorts of things in and with natural language. Symbolic languages,
by contrast, are specifcally designed for particular purposes and
are useless for others. One cannot tell a joke or a story, describe
a room (or a language), or even greet a friend in the formula lan-
guage of arithmetic.
Natural language is also constantly evolving through its use;
it is inherently social and deeply historical. Symbolic language is
instead self-consciously created, often by a single individual, and
has no inherent impulse to change with use. Although French
has changed considerably over the past four centuries, the lan-
guage of algebra that Descartes introduced in 1637, and which
every schoolchild learns today, has changed not at all .
A subtler diference concerns the characteristic sorts of con-
cepts of the two languages. Te concepts of natural language
are paradigmatically object-involving and, for that reason, sen-
sory: we talk, frst and foremost, of the objects met with in every-
day experience, for example, of the things we eat, navigate by,
enjoy, and fear. Although our concepts of such things do involve
much more than the way they appear sensorily to us in appropri-
ate circumstances, they could not be understood in abstraction
from those appearances, that is, from the subjective character of
our experience of them. Te concepts of natural language (those
owing nothing to the development of symbolic language) are con-
cepts of sensory entities, of things that look, feel, taste, smell, and
sound in characteristic ways (compare embodiment ). Insofar
as they are, they resist expression in a symbolic language.
Consider, for example, the notion of a sphere. According to
Aristotle, a sphere is a common sensible; it is an object that
contrasts with a proper sensible, such as the color red or a
certain bitter taste, in being perceptible not merely through one
sense organ (say, by sight as color is, or by taste as bitterness is)
but through more than one. Red things have a characteristic
look; spheres have a characteristic look and a characteristic feel.
Aristotles concept of a sphere is not the modern, mathemati-
cal, and nonsensory concept of a two-dimensional surface all
points of which are equidistant from a center; it is the concept
of a three-dimensional, essentially sensory object. Te mod-
ern mathematical concept of a sphere can be expressed in the
symbolic language of mathematics (namely, as x
2
+ y
2
+ z
2
= r
2
);
Aristotles essentially sensory concept cannot .
We have seen that a symbolic language, unlike a natural lan-
guage, is self-consciously created and expresses concepts that
can be fully understood in abstraction from our sensory experi-
ence. It is unsurprising, then, that the rules governing the use of
signs in a symbolic language can be explicitly formulated and,
hence, that it is relatively easy to build a machine that correctly
manipulates those signs. Because natural language is instead
social, historical, essentially sensory and object-involving, mak-
ing the rules of its use explicit (in order, perhaps, to build a
machine capable of correctly manipulating its signs) is an alto-
gether diferent, and much harder, possibly intractable, prob-
lem. It may be that natural language users can only be grown or
raised, that they cannot be built .
L
LANGUAGE, NATURAL AND SYMBOLIC
Te paradigm of language is natural language, a naturally evolved
system of human communication using spoken or signed words
according to the various ways they can be combined. But in an
extended sense, many species of nonhuman animals also have
language, in this case, a means of communication through inar-
ticulate sounds; and humans similarly have various nonver-
bal means of expression and communication through facial
expressions, gestures , and body language more generally, and
through music, dance, and the imitative arts (see animal com-
munication and human language; art, languages of;
communication, prelinguistic ). For more than a thou-
sand years, we have also had the positional Arabic numeration
system, together with its algorithms for the basic arithmetical
operations, a symbolic language that was extended in the sev-
enteenth century to include the literal notation of algebra as
well. Te twentieth century, fnally, saw the development of very
powerful formal languages, for example, the language of math-
ematical logic and programming languages, both of which are
constituted by a collection of signs together with rules governing
their use, that is, by a syntax but not, at least not essentially, a
semantics (see artificial languages ) .
Although natural language is the paradigm of language, the
foregoing suggests a continuum of sorts: frst with systems of
animal and human nonverbal communication, then natural
languages, followed by the symbolic language of arithmetic and
algebra, and fnally programming and other formal languages.
Given their centrality, both along this continuum and in our
lives, the focus here is on natural language and the symbolic lan-
guage of arithmetic and algebra.
Natural language and the symbolic language of arithmetic
and algebra share two fundamental, and related, features. Both
are vehicles of inquiry and knowing, and both are a medium for
the expression of what we know. Not only do we learn from expe-
rience as animals can but we also, in virtue of our immersion in
natural language, can refect on what we learn, wonder at how
we learn, and question whether we really know what we seem to
know. Because we reason not only implicitly and involuntarily as
other animals do but also explicitly, in words, we can question
our reasons and strive to discover better ones. Just the same is
true of the language of arithmetic and algebra by means of which
we discover, for example, negative, rational, real, and complex
numbers, explore their manifold natures, and display in familiar
symbols that which we know. Tough for very diferent reasons,
neither systems of nonverbal communication nor merely formal
languages can serve in this way as a vehicle of critically refective
inquiry.
Natural language and the symbolic language of arithmetic
and algebra are also very diferent, however. Natural language
is, for example, primarily spoken (or signed) and a vehicle of
communication between a speaker and a hearer; the symbolic
language of mathematics is instead essentially written and serves
primarily as a vehicle of reasoning. Spoken natural language is
intelligible independent of written natural language; symbolic
Language, Natural and Symbolic
410
with no inherent impulse to change. Te two sorts of language
are, then, quite diferent. And yet most work in the language sci-
ences is pursued on the assumption that symbolic language dif-
fers from natural language only in its degree of clarity, rigor, and
perspicuity. We may, as a result, fail adequately to understand
either. By importing considerations relevant only to the work-
ings of a symbolic language, we are liable to falsify the work-
ings of natural language; and by taking natural language as our
paradigm, we risk radically misconceiving the nature of symbolic
language. (It has, for example, long been assumed that the con-
ception of generality, or quantification , that is needed in
mathematics is identical to that employed in natural language,
but perhaps this is just not so.) If natural and symbolic languages
are essentially diferent, then the language sciences need to take
those diferences into account, showing how they do, or do not,
matter to research programs in those sciences .
Tere are further ramifcations for psychology and for peda-
gogy. For example, given the diferences between natural and
symbolic language, it is reasonable to ask whether one reads
symbolic languages diferently than one reads written natural
language, whether one looks at the page of marks diferently in
the two cases. If one does, then perhaps what prevents some
students from thriving in mathematics is that they try to read its
symbolic language as if it were a written natural language. Maybe
students who are adept at modern, symbolic mathematics are, in
fact, primarily adept at catching on, without explicit instruction,
to the way of writing and reading it involves. Tese are testable
hypotheses, but they are hypotheses we will think to test only if
we comprehend the diferences between natural and symbolic
languages .
Danielle Macbeth
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Gupta , Anil . 1980 . Te Logic of Common Nouns . New Haven, CT : Yale
University Press . Argues for the two sorts of predication needed in
speaking about change.
Macbeth , Danielle . 2004 . Vite, Descartes, and the emergence of mod-
ern mathematics . Graduate Faculty Philosophy Journal 25 : 87 117.
Explores diferences between premodern, nonsymbolic and modern,
symbolic mathematical understanding.
. 2005 . Freges Logic . Cambridge : Harvard University Press . Argues
that Gottlob Freges language, unlike the language of mathematical
logic, which is merely formal, without content, is a fully contentful
symbolic language, a vehicle of inquiry.
Wittgenstein , Ludwig . 1953 . Philosophical Investigations .
Oxford : Blackwell . Deeply insightful investigation into the nature of
natural language.
LANGUAGE ACQUISITION DEVICE (LAD)
Tis term refers to Noam Chomskys early proposal of what
would be necessary for construction of a language acquisition
model (1965, 303). More precisely, this early proposal for a
language acquisition device (LAD) provides a logical analysis
of the components that would be minimally necessary for any
such model to account for language acquisition. Its components
are summarized in Figure 1 . Tese can be viewed as specifying
logically and abstractly what would be minimally necessary to
Te fact that natural language embodies a particular sensory
view of the world, one that is inextricably tied to human biol-
ogy, has other implications as well. Our common experience,
grounded in our common (biological) form of life, explains, for
example, the intertranslatability of all human natural languages
and predicts the untranslatability of the natural languages of
creatures evolved to have radically diferent (biological) forms
of life. Given the role of acculturation in the acquisition of natu-
ral language, any being capable of learning a particular natural
language must share at least some sense modalities with other
speakers of the language; but nothing in the very idea of accul-
turation into a natural language requires that there be some
favored sense modalities. Symbolic language is not similarly
rooted in our bodily being. Not only are symbolic languages
universal to all people whatever their natural language; they
are in principle, at least along this dimension, accessible to any
rational being.
Why, then, is translation from one natural language to
another so dif cult? Ludwig Wittgenstein suggests an answer
in his Philosophical Investigations . As he points out, although
some of our everyday concepts can be adequately understood by
reference to a standard or paradigm case, most (including lan-
guage ) are correctly applied to a range of things that exhibit only
a family resemblance . Whereas in the former instance all
correct applications refer back to the one standard or paradigm
case, in the case of concepts that exhibit only a family resem-
blance, similarities between correct applications need only over-
lap, like fbers twisted one on another over the length of a thread.
Such similarities are essentially historically conditioned: Some
applications will, at a particular moment in the evolution of the
language, seem appropriate, natural; others will not work so
well, or at all. And because what works crucially depends on pre-
vious successful applications, diferent natural languages come
to employ common words in quite diferent ways. Tere is noth-
ing like this in the case of symbolic language. Although the words
may be borrowed from natural language (e.g., group, ring,
and feld in abstract algebra), the concepts of a symbolic lan-
guage have a fxed content and univocal application. It is only
our understanding of those concepts that develops and deepens
in the course of inquiry. Te nature and structure of meaning is
quite diferent in the two sorts of language .
Te two sorts of language tend to difer, fnally, in their
predicative structure: a natural language is constitutively
narrative , a language within which to tell what happens ; a
symbolic language is not. And natural language is narrative
because the everyday world is a world of becoming, of change.
To speak of it, then, requires two modes of predication, that
marking what something is and that marking properties a thing
can acquire and lose and also (as a matter of its form, not merely
in its content) tense . Symbolic languages generally involve nei-
ther diferent modes of predication nor tense. Te language of
mathematics, for example, speaks timelessly of what is: 7 + 5 = 12,
( a + b )
2
= a
2
+ 2 ab + b
2
.
Natural language is primarily spoken and communicative,
narrative, essentially sensory and object-involving, and histori-
cal, that is, constantly evolving with use. A symbolic language,
such as the language of arithmetic and algebra, is essentially
written, non-narrative, nonsensory, and self-consciously created
Language, Natural and Symbolic Language Acquisition Device (LAD)
411
measure for (v) required specifcation. Te absence of a temporal
or developmental component was especially challenging, given
that an instantaneous view of language acquisition is obvi-
ously false (Chomsky 1975, 119, 121) .
Subsequent formulations of linguistic theory that pursue
a generative approach now ofer a number of developments
of the original LAD proposal. In general, they seek to defne
a theory of universal grammar (UG) as a theory of the ini-
tial state. Defned as an abstract partial specifcation of the
genetic program that enables the child to interpret certain
events as linguistic experience and to construct a system of
rules and principles on the basis of this experience (Chomsky
1980, 187) and as of course, not a grammar, but rather a sys-
tem of conditions on the range of possible grammars for pos-
sible human languages (Chomsky 1980, 189), UG opens the
possibility for developing future integration of logical and real-
istic approaches to language acquisition and language devel-
opment. a principles and parameters version of UG, for
example, formulates parameters as providing a specifc mech-
anism for (i)(v). (See Table 4.4 in Lust 2006 for examples of
current approaches to modeling language acquisition within
this framework.)
Combining these and subsequent theoretical advances with
signifcant recent empirical advances concerning infant map-
ping to language data (e.g., Jusczyk 1997 ; Morgan and Demuth
1996), which informs us on actual mechanisms in (ii) and (iii),
may now allow the feld to approach the more comprehensive
model of language acquisition that LAD was frst intended to
underlie (see discussion in Chomsky 2000). See Chapter 4 in Lust
2006 for further discussion of this LAD-based model and a review
of several proposed alternatives to the LAD, derived from oppos-
ing theoretical paradigms .
Barbara Lust
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baker , C. L. , and J. McCarthy , eds. 1981. Te Logical Problem of Language
Acquisition . Cambridge, MA : MIT Press .
Brown , R. 1973 . A First Language. Cambridge : Harvard University Press .
Chomsky , N. 1965 . Aspects of the Teory of Syntax . Cambridge, MA : MIT
Press .
. 1975 . Refections on Language . New York : Pantheon .
. 1980 . Rules and Representations . New York : Columbia University
Press .
the child as language learner in the initial state , that is, before
language experience.
Tese components, which were proposed as necessary for
language learning, were formulated as accounting for the map-
ping from primary language data (data which is necessarily
fnite and inherently variable) presented at the initial state to
complex, infnite, and systematic language knowledge in the end
state. Tey characterized the foundations of the specifc innate
abilities that make this achievement possible for the language
learner (Chomsky 1965, 27).
At the same time, the components in Figure 1 were to indicate
what a linguistic theory would have to treat in order to support
an acquisition model and thus attain explanatory adequacy
(see descriptive, observational, and explanatory
adequacy ). For example, such a theory would need to pro-
vide a representation of possible sentence in order to support
(i), a defnition of structural description for (ii), generative
grammar for (iii), a method for determining structural descrip-
tions for (iv), and an evaluation metric for (v).
Te original Chomsky proposal is often interpreted as though
it described a realistic device. However, it is best viewed as pro-
viding an explication of the logical foundations that any compre-
hensive model for language acquisition would have to presuppose
and account for, with the precise nature of each of the compo-
nents to be subsequently determined as an empirical matter
(Chomsky 1965 , 37). Perhaps at least partially because of this
divergence in interpretation, subsequent language acquisition
research over the last decades has frequently pursued divergent
paths, one pursuing the logical problem of language acquisi-
tion consisting of linguistic analyses of potential data mapping
(e.g., Baker and McCarthy 1981 ), another pursuing a realistic
approach consisting of empirical studies of language develop-
ment (initiated largely by Roger Brown and his students [1973]).
In fact, certain components in Figure 1 proved particularly
challenging to a realistic model. For example, if there were a real-
istic device, it would have to include a mechanism for (i) and (ii).
If (iii) were to be implemented in terms of an enumeration of
the class G1, G2 of possible generative grammars (Chomsky
1965 , 31), then there is the de facto impossibility of predefning
given grammatical hypotheses across 6,000 to 7,000 actually
potentially innumerable human languages, as well as the risk
of begging the question of language acquisition, that is, how the
individual grammars arise. Similarly, the nature of an evaluation
(i) Technique for representing input signals
(ii) Way of representing structural information about these signals
(iii) Some initial delimitation of a class of possible hypotheses about language structure
(iv) Method for determining what each such hypothesis implies with respect to each sentence
(v) Method for selecting one of the (presumably infinitely many) hypotheses that are allowed by
(iii) and are compatible with the given primary linguistic data.
Figure 1. LAD components (Chomsky 1965, 30).
Language Acquisition Device (LAD)
412
Sound Change
Te symmetry of the terms used to describe sound changes
assimilation/dissimilation, weakening/strengthening, deletion/
insertion makes it seem as though all directions of change were
equally probable. Te facts now known from a wide array of lan-
guages show instead that assimilation, weakening, and deletion
are vastly more common than dissimilation, strengthening, and
insertion. If sound change is rigidly defned as language internal,
phonetically motivated, and lexically general, then the less com-
mon types of change show up even more rarely.
Common assimilation changes involve consonants taking
on the properties of adjacent vowels, for instance, palataliza-
tion before front vowels, labialization or rounding before round
vowels. Or vowels can take on properties of consonants, as
when vowels become nasalized in the same syllable with nasal
consonants. For some of these changes, more detailed hierar-
chies of contexts for change can be established. Palatalization
of coronal consonants (such as [t], [d], [s]) occurs frst and most
commonly before a high front glide ([j]), occurs next before a
high front vowel ([i]), and then progresses to the lower front
vowels .
Paths of reductive change can also be established. For
instance, a common consonant reduction involves the loss of [p]
via the pathway p > f > h > . Parts of this path are documented
in diferent languages: Japanese has undergone a change that
reduced all prevocalic instances of /p/ to a fricative that assimi-
lates to the place of articulation of the following vowel. Spanish
and other Romance languages have undergone a change that
reduced word-initial [f] to [h] and further to . Te positions in
which such reductions occur also show regularity: Syllable-fnal
position favors the reduction of consonants, as does intervocalic
position. Word-initial position is least likely to condition reduc-
tion of consonants .
It is commonly assumed that such changes are caused by
phonetic tendencies. In the articulatory domain, some general
principles of change have been proposed that deal with the way
articulatory gestures change during production. Te primary
directions in phonetic change are the reduction of the magnitude
of the gestures, leading to reduction or loss, and their increased
overlap, leading to assimilation (Browman and Goldstein 1992 ;
Mowrey and Pagliuca 1995 ).
As mentioned earlier, the importance of universals of language
change is that they can predict and thus explain synchronic pat-
terns across languages. For instance, the presence of nasal vowels
in a language is almost always due to assimilation to a nasal conso-
nant. Sometimes this consonant is lost, leading to phonemic nasal
vowels. Te diachronic source explains why nasal vowels always
have a more restricted distribution and frequency compared to
oral vowels (Greenberg 1966 ). Similarly, the fact that some lan-
guages lack the phoneme [p] can be explained by its tendency to
undergo reduction. Te restrictions against certain consonants
in syllable-fnal position can be attributed to their propensity for
loss in that position (Vennemann 1988 , Blevins 2004 ).
After Sound Change: Morphologization
Another set of unidirectional changes involves the results of
sound change. Phonetically conditioned sound change creates
alternations that gradually acquire morphological or lexical
. 2000 . Te Architecture of Language , ed. J. Mukherji , B. Narayan
Patnaik , and R. Agnihotri . Oxford : Oxford University Press .
Jusczyk , P. 1997 . Te Discovery of Spoken Language . Cambridge, MA : MIT
Press .
Lust , B. 2006 . Child Language: Acquisition and Growth .
Cambridge : Cambridge University Press .
Morgan , J. , and K. Demuth , eds. 1996. Signal to Syntax: Bootstrapping
from Speech to Syntax in Early Acquisition . Mahwah, NJ : Lawrence
Erlbaum .
LANGUAGE CHANGE, UNIVERSALS OF
All aspects of language can change, and with every aspect we see
broad tendencies that have inspired the search for universals of
change. In some areas, powerful universals have emerged, sug-
gesting that many of the synchronic similarities we see across
languages may have their source in shared cognitive and process-
ing mechanisms. Researchers have identifed cross-linguistically
similar paths of change in sound change in morphological
change, and in grammaticalization .
Given any two related items, A and B, one could in principle
expect a change from A > B or from B > A to be equally prob-
able. However, the facts of change show that this is not the case.
Across the languages of the world, we fnd that one direction
predominates and the other is rare. Tus, we speak of unidirec-
tionality in language change. Claims about unidirectionality are
sometimes controversial because apparent counterexamples
do emerge. For this reason, it is important to defne types of
change clearly before making proposals about the directionality
of change.
Te importance of such unidirectional trends cannot be over-
stated. Universals attested in the diachronic plane appear to be
more powerful than those that can readily be stated for the syn-
chronic plane. Tus, diachronic typology is a theory of universals
that proposes that the synchronic patterns are not in themselves
the universals but, rather, the result of strong diachronic trends
(Greenberg 1969, Greenberg, Ferguson, and Moravcsik 1978 ).
Te structure of language is created by change that is ongoing in
language use (see usage-based theory ); because languages
are used by human beings in very similar ways across cultures,
languages tend to be similar to one another .
An important question to ask concerns the sense in which
there can be universals of change. No change has to occur; there
are many changes that could occur. Te universals, then, specify
only similar paths of change that can be found in diferent lan-
guages. If these languages are not closely related genetically,
then the similar paths of change cannot be attributed to a shared
history but must be viewed as independent developments. Tus,
even though we cannot say that a change has to occur, nor can
we predict when it will occur, there are still many substantive
predictions that we can make about change.
In examining trends in language change, it is important to
consider how well documented a particular change is. Some
changes are reconstructed on the basis of a comparison of
related languages (see comparative method ). Since known
trends in change are often used in this reconstruction, such
changes cannot be taken as evidence for trends in change. Only
well-attested changes are valid inputs to a theory of universals
of change.
Language Change, Universals of
413
habeo meant I have to say, In medieval Spanish, the verb had
reduced to he and consistently followed the infnitive. Eventually
the former verb (now a grammatical word) fused with the infni-
tive, giving decir . In further changes, the stem of some verbs lost
a syllable, in our example creating dir I will say. Te last stage
(loss) is occurring in this case as a new future from ir a + verb go
to + verb replaces the old future.
Parallel to this path of change are the many paths of semantic/
pragmatic change that have been identifed as creating the gram-
matical morphemes of the languages of the world. Te numeral
one with a noun tends to develop into an indefnite article; demon-
stratives develop into defnite articles and complementizers; verb
constructions involving a verb meaning want or go to plus
another verb develop into future markers; resultative construc-
tions with have or be and a past or passive participle, such as I
have gone, develop into perfect, pasts, and perfectives; construc-
tions with locative verbs or movement verbs develop into progres-
sives, which may go on to become imperfectives. Verbs meaning
know or be able develop into auxiliaries indicating possibility.
A preposition meaning to or in order to develops into an infn-
itive marker. Body-part terms such as head and back become spa-
tial and later temporal adpositions. Passive constructions develop
into ergative constructions. All of these paths of change (and many
others) are documented as independent developments in unre-
lated languages (Bybee, Perkins, and Pagliuca 1994 ).
In comparing these specifc paths, certain general patterns
are discernible: As a construction grammaticalizes, its mean-
ing becomes more general and abstract; its form becomes more
reduced and dependent upon surrounding material; it under-
goes an extreme frequency increase; and its category member-
ship can change, say, from verb to auxiliary, from noun or verb
to adposition. Te lexical material upon which grammaticaliza-
tion works is similar across languages, as are the mechanisms of
change. Te fact, for instance, that one can infer intention from
a construction of be going to VERB and then later infer future
from that intention seems not to be culture specifc, inasmuch
as the development of such a construction into future occurs in
languages all over the world. Tus, the commonalities found in
grammaticalization point to interactive, cognitive, and process-
ing mechanisms that are shared across cultures .
Joan Bybee
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Blevins , Juliette . 2004 . Evolutionary Phonology . Cambridge : Cambridge
University Press .
Browman , Catherine P. , and Louis M. Goldstein . 1992 . Articulatory pho-
nology: An overview . Phonetica 49 : 155 80.
Bybee , Joan , Revere Perkins , and William Pagliuca . 1994 . Te Evolution of
Grammar: Tense, Aspect and Modality in the Languages of the World.
Chicago : University of Chicago Press.
Greenberg , Joseph . 1966 . Language Universals: With Special Reference to
Feature Hierarchies . Te Hague : Mouton .
. 1969 . Some methods of dynamic comparison in linguistics.
In Substance and Structure of Language ed. Jan Puhvel , 147203.
Berkeley : University of California Press .
Greenberg , Joseph H. , Charles Ferguson , and Edith. Moravcsik , eds.
1978 . Universals of Human Language. Vols. 14. Stanford, CA : Stanford
University Press .
conditioning, as, for example, when vowel shortening before a
consonant cluster created an alternation in English keep/ kept,
sleep/slept and weep/wept.
Another example is the alternation in nouns such as wife, wives.
At frst, in Old English there were no voiced fricatives: /v/, /z/, and
// did not occur. Later, by a regular sound change /f/, /s/, and
// became voiced when they occurred between two vowels, as
in the plurals of house, wife and wreath (at that time the plural
suf x had a vowel in it, putting the fricative between two vow-
els). Nowadays, the alternation is not phonetically productive. In
words such as classes, efort, and ether , voiceless fricatives occur
between vowels. Also /v/, /z/, and // have become phonemes
and they occur in all positions. So now the alternation still found
in wife, wives , and so on is associated with certain nouns and the
morphological category of plural. In this way, morphologically or
lexically conditioned alternations are created; such alternations
tend to be unproductive phonologically and to involve segments
that are contrastive elsewhere .
Morphological Change: Analogy
Morphological and lexical alternations that are created in the way
just described tend to undergo further change on the basis of the
patterns found in the paradigms of the language. Certain gen-
eral trends in analogical change have been identifed (Mazcak
1980). In analogical leveling, when an alternation is lost, it is
the high frequency form of the paradigm, such as the singular
in nouns or the present indicative in verbs, that serves as the
basis for new formations. Tus, the alternation weep/wept is lev-
eled by the creation of a new past tense, weeped. Low-frequency
words are more likely to undergo analogical leveling than high-
frequency words. Tus, wept is more likely to change than kept .
Te productive pattern that prevails in leveling is the one that has
the largest number of members. One synchronic consequence
of these trends is that the irregular paradigms in a language are
usually among the most frequent.
Extension of an alternation to a paradigm that did not have it
before is less common than is leveling. It occurs where there is
strong phonological similarity to a productive pattern, as when
the regular verb string gets a new past tense strung due to its simi-
larity to swing, swung .
Morpho-Syntactic Change: Grammaticalization
Most change in the morpho-syntax is a product of the general pro-
cess of grammaticalization. Grammaticalization itself, however,
involves phonetic, semantic, and pragmatic change, in addition to
morpho-syntactic change. Various cross-linguistic paths of change
have been identifed in recent research into grammaticalization.
First, there is a general path of change for the form of the gram-
maticalizing element as it progresses from lexical to grammatical:
content item > grammatical word > clitic > infectional af x >
stem change > loss
Tis progression involves a loss of the properties originally
associated with the content word (its ability to occur as a noun
or verb) and a growing fusion of the element with lexical items
nearby. A good example is the development of the Spanish future
suf xes. In Latin, there was a construction that consisted of an
infnitive with the conjugated verb habere . For example, dicere +
Language Change, Universals of
414
Te common ancestor of an entire language family is assigned
a name by prefxing Proto- to the name of the family, as in Proto-
Indo-European and Proto-Afro-Asiatic. Te place where the pro-
tolanguage was spoken is called the homeland of the language
family .
Te study of language families is part of historical lin-
guistics and is contextualized within a particular model of
language change : divergence. When innovations in one part
of a language community fail to spread to other parts, diferences
accumulate until the community can be said to speak diferent
languages. It is this historical process that language-family the-
ory is meant to model. But perhaps because language families
are commonly illustrated by showing similarities between lan-
guages (e.g., English mouse is cognate with Latin mus ), the idea
arises that relatedness is about similarity between languages. In
fact, there is no requirement that cognates be similar at all (e.g.,
English two is related to Armenian yerku ), and many sources of
similarity are disavowed as being irrelevant to the model. Tese
include borrowing (see contact, language ), onomatopoeia,
universals ( absolute and statistical universals ), and
chance similarities .
Te study of language families typically involves one or more
of the following enterprises:
Demonstrating that languages are related
Reconstructing the common protolanguage
Subgrouping the languages by hypothesizing intermediate
ancestors
Associating linguistic data with historical and archaeological
data
Te following sections frst describe the traditional and still-
dominant methods for pursuing these tasks and then sketch and
evaluate some new methodologies.
Traditional Methods
Te traditional technique is the comparative method . Te
linguist studies characteristics that rarely recur across languages,
such as grammatical paradigms and the associations between
sound and meaning in morphemes . Eforts are made to discard
loans and onomatopoeia, although the former is a dif cult and
often intractable problem. Matching morphemes across lan-
guages by meaning, one looks for recurrent sound correspon-
dences. For example, English f corresponds to Latin p in father
= pater , feels = palpat , few = pauca , and many other words. If a
large number of recurrent correspondences are found, the lan-
guages are related. Te recurrences are also used to reconstruct
the protolanguage (see historical reconstruction ).
After a language family is identifed, the next step is sub-
grouping, identifying the branches or groups within the fam-
ily. Subgrouping seeks to uncover the history of the divergence
( cladogenesis ) of a language family. If the family contains three
or more languages, the linguist looks for evidence that some
proper subset of those languages may have descended from
an intermediate common ancestor. Tis is done by looking for
shared innovations ( synapomorphies ) sound changes or new
words or grammatical constructions that were not in the ances-
tor language but are found in two or more of the descendant
languages. For example, the fact that English, German, Swedish,
Maczak , Witold . 1980 . Laws of analogy. In Historical Morphology ed.
J. Fisiak , 2838. Berlin : Mouton de Gruyter .
Mowrey , Richard , and William Pagliuca . 1995 . Te reductive character
of articulatory evolution . Rivista di Linguistica 7 .1: 37 124.
Vennemann , Teo . 1988 . Preference Laws for Syllable Structure and the
Explanation of Sound Change . Berlin : Mouton de Gruyter .
LANGUAGE FAMILIES
A language family is a set of languages that developed from the
same ancestral language. Te best-known example is the Indo-
European family, which comprises more than a hundred lan-
guages that, even in premodern times, extended from the Indian
subcontinent to northwestern Europe. Tis family is well known
not only because it contains many of the worlds most widely
spoken languages, such as Bengali, English, French, German,
Hindi, Portuguese, and Russian, but also because it was the main
focus of research in the nineteenth century, when linguistics
was established as a modern science . However, many other lan-
guage families are subjects of intense research today, such as the
following:
Afro-Asiatic, which includes Arabic, Hebrew, and several
languages of northern Africa, including Ancient Egyptian,
Hausa, and Somali
Algic, which includes several native languages of North
America, such as Wiyot, Yurok, Cheyenne, Ojibwa, and
Shawnee
Austronesian, which includes more than a thousand lan-
guages spoken from Madagascar to Polynesia, such as Bahasa
Indonesia, Fijian, Tagalog, and Tahitian
Dravidian, which includes most of the non-Indo-European
languages of India, such as Malayalam, Tamil, and Telugu
Niger-Congo, which includes most of the languages of sub-
Saharan Africa, including Igbo, Swahili, and Zulu
Sino-Tibetan, which includes Burmese, Chinese, and Tibetan
Tupi, comprising several languages of South America, includ-
ing Guarani
Uralic, which includes most of the non-Indo-European lan-
guages of Europe, such as Estonian, Finnish, Hungarian,
Nenets, and Sami (Lapp)
Tis list is only a sample of the hundreds of known language
families and of the languages included in each one. For a com-
prehensive listing, see Gordon ( 2005 ).
Familial metaphors are the standard terms of art. Languages in
the same family are said to be genetically related to one another.
A language from which other languages developed is called an
ancestor, or parent, of those languages. Words that descend from
the same form in an ancestral language are related, or cognate.
Tis homey terminology is undergoing some competition with
that of modern cladistics as used in biology, but certain linguistic
concepts do not translate well. In biology, it is understood that
all biological taxa are related to one another, and family is but
a midlevel taxon. Linguists assume that relationships between
languages must be proved, and a language family is a maximal
taxon. In principle, even isolates languages that do not group
with other languages can trivially be considered families by
reinterpreting their dialect s as separate languages .
Language Families
415
is often left with unlikely cladograms like the 15-way branching
tree of Indo-European.
Another disappointment is that little progress has been made
in the past century in pinning down the Indo-European home-
land or proving that additional languages are related to English
topics of recurrent interest among linguists, archaeologists, and
enthusiasts alike. More disappointing is that when linguists have
claimed that language families such as Uralic are related to Indo-
European such groupings often being given the Eurocentric
name Nostratic (our family) the methodology has given no
frm guidance as to how signifcant the evidence is, with the
result that many linguists fnd themselves uncomfortably agnos-
tic on whether Nostratic has been proved or not. Unlike in mod-
ern experimental sciences, there are no statistical techniques for
estimating the probability that the number of correspondences
found is due to a real relationship between languages rather
than to chance. Rules of thumb were developed to provide some
guidance; a typical piece of advice is to treat words as potentially
cognate only if at least three of their consonants are found in
recurrent correspondence sets. But such rules are very approx-
imate, not tailored to the specifc structures of the languages at
hand, and they have discouraged linguists from applying the
method to languages with short morphemes.
Joseph H. Greenberg addressed several of these concerns
with a technique called multilateral comparison . Tables are con-
structed listing the translation equivalents for many concepts in
many diferent languages. It is claimed that the tabular layout
itself makes the relationships among the languages, even their
correct subgrouping, patent. Using this technique, Greenberg
presented an analysis of the languages of Africa (1963), which is
now considered standard, then went on to hypothesize language
families that lumped established families together into much
larger families what became known as deep linguistic relation-
ships . Te dozens of families and isolates of the Americas were
reduced to three families (Greenberg 1987); Indo-European,
Uralic, Japanese, and several other families were lumped into a
family called Eurasiatic (Greenberg 2002). Multilateral compari-
son has proven popular among enthusiasts, in part because it
requires no special language expertise, in part because it appears
to reveal many new, deep relationships. Unfortunately, there is
no way to evaluate a methodology that simply calls for contem-
plating raw data until patterns emerge .
Several researchers have shown, however, that some of
Greenbergs key ideas can be transformed into algorithmic
(reproducible) methodologies that introduce to language fam-
ily research the beneft of statistical signifcance testing. Robert
L. Oswalts procedure (1998) minimized experimenter bias by
requiring that a specifc concept list be used and that one specify
in advance specifc criteria for measuring degree of similarity
between two languages. William H. Baxter and Alexis Manaster
Ramer ( 2000 ) added reliable signifcance-testing procedures
based on randomization tests. Brett Kessler and Annukka
Lehtonen ( 2006 ) adapted the technique to handle multiple lan-
guages in a single test, informally confrming Greenbergs claim
that such large-scale comparisons are inherently more power-
ful than two-language comparisons. Don A. Ringe ( 1992 ; see
Kessler 2001 for extensive discussion and methodological refne-
ments) measured not similarity but the number of recurrent
and several other languages have f where Proto-Indo-European
had p is a shared innovation that indicates that those languages
may have a shared intermediate ancestor that underwent this
change; otherwise, we would have to assume that each of those
languages separately innovated the change of p to f or borrowed
the innovation from another language. In fact, the preponder-
ance of evidence supports such an intermediate language and
a branch ( clade ) of languages descending from it: the Germanic
languages. Other branches of Indo-European include the Balto-
Slavic (including Bosnian, Lithuanian, Polish, and Russian),
Celtic (including Breton, Irish, and Welsh), Italic (including Latin
and the Romance languages), Indo-Iranian (including Bengali,
Farsi, Pashto, and Urdu), and the extinct Anatolian branch,
which included Hittite, Luvian, and Lycian. Several other lan-
guages, including Greek, Albanian, Armenian, and half a dozen
extinct languages (see extinction of languages ), do not
share an agreed-upon intermediate branch at all.
Associating language history with external facts entails pinning
the protolanguage to a particular time and place its homeland
and demonstrating how it spread from there. Te time depths
under consideration mean that written records are rarely, if ever,
available. Te primary linguistic tool is to look for words found
in multiple branches of a language family and exhibiting all of
the regular sound correspondences; they are assumed to date
back to the protolanguage and, therefore, to name objects found
in its environment. For example, a pan-Indo-European word for
wheel suggests that the protolanguage split up no earlier than
the invention of the wheel, some six thousand years ago (Mallory
1989 ). Another technique is to look for areas of greatest linguis-
tic diversity. Te fact that the Austronesian languages are much
more diverse in Taiwan than anywhere else supports the theory
that they developed there longest, that is, that Taiwan was the
homeland for the Austronesian family (Blust 1999 ). A third tech-
nique is to seek archaeological evidence of population move-
ments that may have disseminated a language family. In the
case of Austronesian, knowledge of how people spread through
the Pacifc and Indian Oceans is consistent with the theory of
a Taiwan homeland. In the case of Indo-European, it has often
been noted that early adopters of horse-drawn wheeled chariots
would be in an ideal position to spread their languages through-
out much of Europe and Asia. A well-received theory points to
the chariot users who lived in the PonticCaspian region about
six thousand years ago (Mallory 1989 ) .
Challenges to the Traditional Method
Te traditional comparative method is still the basic framework
within which language families are researched, but it is not per-
fect. It is a complicated process that demands a great deal of
knowledge about all of the relevant languages. It can be misled
by loanwords, and it ofers little guidance in distinguishing true
shared innovations ( synapomorphies ) from independent iden-
tical innovations ( homoplasies ). Te linguist must constantly
decide whether multiple languages could have undergone a
particular change independently and how likely they would be
to have borrowed it. In reality, of course, anything that happens
once can happen twice, and there is nothing that is not subject to
borrowing (Tomason and Kaufman 1988 ). Te true solution is
probabilistic, but hard numbers are lacking, and the investigator
Language Families
416
there must come a point at which any remaining commonalities
between languages are indistinguishable from chance levels.
But even if the more pessimistic predictions are true and new
methods are unlikely to greatly expand intensively studied fami-
lies like Indo-European, they may greatly ease new analyses of
lesser-known languages.
New computerized cladistic methods are, likewise, already
aiding the analysis of complex language families and are provid-
ing Indo-Europeanists food for thought. However, the develop-
ment and application of such algorithms could beneft from the
compilation and deployment of data about the probability of
various types of linguistic innovations and borrowings.
To date, the new methodologies have not been adopted
by most practitioners. While it is easy to fault established
researchers for conservatism, it is also true that quantita-
tive methods typically cannot take into account the diverse
types of information that linguists are accustomed to reason-
ing with. Fortunately, the emerging partnerships between lin-
guists and cladists should help bridge the gap between old and
new approaches and lead to the widespread adoption of hybrid
methodologies .
Brett Kessler
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baxter , William H. , and Alexis Manaster Ramer . 2000 . Beyond lump-
ing and splitting: Probabilistic issues in historical linguistics. In Time
Depth in Historical Linguistics , ed. Colin Renfrew , April McMahon,
and Larry Trask, 16788. Cambridge, UK : McDonald Institute for
Archaeological Research .
Blust , Robert. 1999 . Subgrouping, circularity and extinction: Some
issues in Austronesian comparative linguistics. In Selected Papers
from the Eighth International Conference on Austronesian Linguistics ,
ed. E. Zeitoun and P. J. K. Li , 3194. Taipei : Academia Sinica .
Bryant , David , Flavia Filimon , and Russell D. Gray . 2005 . Untangling
our past: Languages, trees, splits and networks. In Te Evolution of
Cultural Diversity: A Phylogenetic Approach , ed. Ruth Mace , Clare J.
Holden , and Stephen Shennan , 6985. London : UCL Press .
Cavalli-Sforza , Luigi Luca, Paolo Menozzi , and Alberto Piazza , 1994.
Te History and Geography of Human Genes . Princeton, NJ : Princeton
University Press
Gordon , Raymond G. , Jr., ed. 2005. Ethnologue: Languages of the World .
15th ed. Dallas : SIL International . Content also available online
at: http://www.ethnologue.com/
Greenberg , Joseph H. 1963 . Te languages of Africa . International
Journal of American Linguistics 29 .1 (Supplement): Part 2.
. 1987 . Language in the Americas . Stanford, CA : Stanford University
Press .
. 2002 . Indo-European and Its Closest Relatives: Te Eurasiatic
Language Family: Lexicon. Stanford, CA : Stanford University Press .
Kessler , Brett , 2001. Te Signifcance of Word Lists. Stanford, CA : Center
for the Study of Language and Information .
Kessler , Brett , and Annukka Lehtonen . 2006 . Multilateral comparison and
signifcance testing of the Indo-Uralic question. In Phylogenetic Methods
and the Prehistory of Languages, ed. P. Forster and C. Renfrew , 3342.
Cambridge, UK : McDonald Institute for Archaeological Research .
Mallory , J. P. 1989 . In Search of the Indo-Europeans: Language,
Archaeology and Myth . London : Tames and Hudson .
Nakhleh , Luay , Don Ringe , and Tandy Warnow . 2005 . Perfect phyloge-
netic networks: A new methodology for reconstructing the evolution-
ary history of natural languages . Language 81 : 382 420.
sound correspondences. Tis has the advantages both of being
closer to the traditional comparative method and of generating
correspondences useful for subgrouping and reconstruction.
Disappointingly, however, none of these neo-Greenbergian
techniques found evidence for the deep relations that were
advertised for the original, impressionistic method .
Other new techniques have concentrated on subgrouping.
Lexicostatistics (Swadesh 1955 ) was an early attempt to facilitate
subgrouping and also assign dates to protolanguages. Te idea
was that if languages replace a constant number of words per
century with new words, then by measuring the percentage of
a list of words that is cognate between languages, one could cal-
culate when the languages diverged and even construct a family
tree. Although these assumptions were mostly wrong and were
therefore rejected by most linguists, many people still use lexi-
costatistical techniques as a rough indication of a languages his-
tory in the absence of more compelling data .
Arranging many shared innovations into a binary tree is an
extremely laborious undertaking, especially given the possibil-
ity that some identical innovations are independent ( homoplas-
tic ). Te recent development of computational cladistic
methods similar to those used in biology (e.g., Ringe, Warnow,
and Taylor 2002 ) is a tremendous advance in helping the lin-
guist fnd optimal trees. In addition, several solutions to the
problem of borrowing have emerged in the form of programs
that construct networks instead of trees. Shared innovations
that cannot be cleanly attributed to a shared ancestor are taken
as evidence of contact, obviating somewhat the need to make a
priori judgments about whether borrowing was involved (e.g.,
Bryant, Filimon, and Gray 2005 ; Nakhleh, Ringe, and Warnow
2005 ) .
Te problems of fnding homelands and tracing the spread
of languages still requires one to resort to data that are often
suggestive but not defnitive. Colin Renfrew ( 1987 ) added a new
perspective when he theorized that languages may be spread
by the movement of culture, rather than by the movement of
people. He suggested that Indo-European languages were
spread from Anatolia along with the adoption of agriculture.
Most linguists have not accepted this theory, in part because it
is incompatible with such linguistic data as an Indo-European
word for the wheel, which postdates the spread of agriculture
by millennia. Recently, further data are aforded by genetic
analyses of populations ( genes and language ). Te pres-
ence of a Pontic genetic component in Europe is compatible
with the idea that invaders from the PonticCaspian region
brought Indo-European languages into Europe (Cavalli-Sforza,
Menozzi, and Piazza 1994 ).
Prospects
Recent computer techniques add simplicity, reproducibility,
and quantitative rigor to methodologies for proving relation-
ships between languages, but so far there has been no notice-
able increase in power over what experts are able to do by hand.
Failure to corroborate the sort of deep relationships conceived
by Greenberg may mean that better techniques need to be devel-
oped, or that the languages are not in fact related, or that the
answer is unknowable. Because languages are always changing,
they constantly lose information that links them to their relatives;
Language Families
417
the interplay, in the determination of meaning, between lan-
guage and the actions into which it is woven and to bring into
prominence the fact that the speaking of language is part of
an activity, or of a form of life ([1953] 1997, 23; see forms of
life ).
In fact, Wittgenstein nowhere provides a well-rounded def-
nition of language-game because his employment of the term
evolves and because it is what he calls a family resemblance
concept. He employs the term
i. to circumscribe various more or less broad domains of lan-
guage. Here, we can speak of single and specifc language-games,
such as those we play in our use of particular words or con-
cepts (e.g., fear, game, hand, knowing) or in specifc
activities (e.g., lying, thanking, cursing, making a joke,
following a rule, giving orders and obeying them); but these
specifc language-games are subsumed under more general uses
of the term either a generic use: he calls the whole, consisting of
language and the actions into which it is woven, the language-
game ([1953] 1997, 7) or an anthropological use, that is, what
he calls the human language-game (1977, 554) (as opposed, say,
to the language-game of alien tribes).
ii. to describe the degree of sophistication of language use.
Here, Wittgenstein speaks of primitive or complicated language-
games.
iii. to describe how language works .
It is in its encapsulation of how language works that the
expression language-game is most eloquent. Te expression
is due to the analogy Wittgenstein makes between language and
games, which supersedes the calculus analogy of the Tractatus ,
thereby signaling the switch from his own conception of language
as a fxed and timeless symbolism to a conception of language as
essentially embedded in human practice language as essen-
tially in use . In the Blue Book, he begins to question the idea that
speaking a language is, in all cases, to apply a calculus according
to strict and exact rules; rather, using language is much like play-
ing a game. Te game analogy is more ftting than the calculus
analogy in several ways:
1. Like game, language is a family resemblance concept . Just
as there is a multiplicity of games with nothing common to all,
there is nothing common to all of our uses of language that
makes them into language or parts of language, but they are
related to one another in many diferent ways, and it is because
of this relationship that we call them all language ([1953]
1997, 65) .
2. Language is an activity, and it is essentially connected with
practice or use : our language-game is a piece of behaviour
(1980, 151); a language-game incorporates both language and
the actions into which it is woven ([1953] 1997 , 7). To use lan-
guage meaningfully is analogous to making a move in a game;
to understand a word is to know how to use it. Just as we learn
how to play a game by learning what the permitted moves are
in the game, we learn the meaning of words by learning what
is accepted as a meaningful use of the word. And here, it is not
the application of explicit rules but training (1970, 186) and
repeated exposure that are needed to play a game properly or
to use a word meaningfully. Te concept of language-games
highlights the idea that the mastery of language is an acquired
skill or know-how, not a systematic ( innate or acquired)
Oswalt , Robert L. , 1998. A probabilistic evaluation of North Eurasiatic
Nostratic. In Nostratic: Sifting the Evidence , ed. J. C. Salmons and B. D.
Joseph , 199216. Amsterdam : Benjamins .
Renfrew , Colin. 1987 . Archaeology and Language: Te Puzzle of Indo-
European Origins. London : Pimlico .
Ringe , Don A ., Jr. 1992. On Calculating the Factor of Chance in Language
Comparison . Philadelphia : American Philosophical Society .
Ringe , Don , Tandy Warnow , and A. Taylor . 2002 . Indo-European and
computational cladistics . Transactions of the Philological Society
100 : 59 129.
Swadesh , Morris . 1955 . Towards greater accuracy in lexicostatistic dat-
ing . International Journal of American Linguistics 21 : 121 37.
Tomason , Sarah Grey , and Terrence Kaufman . 1988 . Language Contact,
Creolization, and Genetic Linguistics . Berkeley : University of California
Press .
LANGUAGE-GAME
At the beginning of the Philosophical Investigations , Ludwig
Wittgenstein questions the Augustinian or traditional picture of
the essence of human language, according to which the meaning
of a word is the object for which it stands, so that a word to which
no object corresponds has no meaning. Learning language con-
sists in giving names to objects, and the association between
the word and the object is established by ostensive teaching of
words. Wittgenstein attacks this conception as both confused
and reductive. It makes naming seem like a queer connection of
a word with an object, a kind of baptism of the object, as if mean-
ing existed separately from the word and was attached to it by
a mental process: a remarkable act of mind or occult process.
Moreover, making the correspondence between name and object
a condition of meaning has the absurd consequence that when
the object no longer exists, the word no longer has any meaning.
Here, the meaning of a name is confounded with the bearer of
the name, whereas in fact when Mr. N. N. dies, one says that the
bearer of the name dies, not that the meaning dies (Wittgenstein
[1953] 1997 , 40). Te Augustinian picture of language is also
reductive. Even if we were to correct its conception of ostensive
defnition (e.g., by replacing the occult process with training), it
would still err in being an oversimplifcation: Ostensive defni-
tion is too narrow; it does not describe everything that we call
language. We think that language consists in giving names to
objects, whereas in fact we do the most various things with our
sentences: Tink of exclamations alone, with their completely
diferent functions. Water! Away! Ow! Help! Fine! No!
([1953] 1997, 27) .
Te functions of words, suggests Wittgenstein, are as diverse
as the functions of tools in a toolbox ([1953] 1997, 11). And
these are countless. But his toolbox analogy does more than
underscore the diversity of the uses of language; it also suggests
that words work much like tools in that their meaning resides in
their use : We remain unconscious of the prodigious diversity
of all the everyday language-games because the clothing of our
language makes everything alike ([1953] 1997, 224) and so it
is use that is the distinguishing mark. In reaction to the mental-
ist conception of meaning, which sees it unilaterally as a mental
connection between words and objects, Wittgenstein af rms
that the meaning of a word or sentence resides in the use we
make of it. He introduces the term language-game to highlight
Language-Game
418
of life a human language. Language and form of life are intern-
ally related: To imagine a language means to imagine a form
of life ([1953] 1997, 19), and to imagine a human language is
necessarily to imagine a human form of life, a human way of
being and acting that essentially involves both our biological
makeup and our social behavior. For our language-games are
impacted by the facts of our natural history, such as our bio-
logical human constraints; for example, the language-game
with colors is characterized by what we can and what we can-
not do (1970, 345). Terefore, if certain very general facts of
nature were diferent from what they are, so would our con-
cepts and language-games be. But to say that our language-
games are conditioned by certain facts (1977 , 617), is not to
say that our language-games are justifed by , or answerable to,
these facts.
5. Like games, language and language-games are autono-
mous . By this, Wittgenstein means that although our language-
games are rooted in our form of life, they are not accountable
to, or rationally grounded in, any reality: Te language-game is
not based on grounds. It is not reasonable (or unreasonable). It
is there like our life (1977, 559). Rather than speak of symbols,
words, or sentences as the primary or elementary units of mean-
ing (as logicians, including Wittgenstein himself in the Tractatus ,
had done), the later Wittgenstein views the language-game as
the basic unit in linguistic activity; he urges us to look on the
language-game as the primary thing ([1953] 1997, 656), that
which does not have to be explained by any fact. Here, he can
be viewed as broadening Gottlob Freges context principle: Te
context necessary for meaning is not the proposition but the
language-game (e.g., a sound is an expression only in a language-
game [(1953)1997, 261]) .
In his last work, On Certainty , Wittgenstein dwells on the
importance of unmoving foundations for the possibility of
language-games: It is essential for our language-games that
no doubt appears at certain points (cf. 1977, 524). He argues
that some basic certainties such as The world exists or
This is a hand or Cats dont grow on trees are necessary,
unmoving foundations of our language-games (cf. 1977, 403,
411), that the whole language-game rests on this kind of cer-
tainty (cf. 1977, 446). This kind of certainty, however, is non-
epistemic; he views it as a kind of trust: I really want to say
that a language-game is only possible if one trusts something
(1977, 509) .
Danile Moyal-Sharrock
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baker , G. P. , and P. M. S. Hacker . 1980 . Wittgenstein: Understanding
& Meaning: An Analytical Commentary on the Philosophical
Investigations . Vol. 1. Oxford : Basil Blackwell .
Canfeld , John V. 1993 . Te Living Language: Wittgenstein and
the Empirical Study of Communication Language Sciences 15.3 :
165 93.
Wittgenstein , Ludwig. [1953] 1997 . Philosophical Investigations . Trans.
G. E. M. Anscombe . 2 d ed. Oxford : Blackwell .
. 1970 . Zettel . Ed. G. E. M. Anscombe and G. H. von Wright , trans
G. E. M. Anscombe . Berkeley : University of California Press .
. 1974 . Philosophical Grammar . Ed. R. Rhees , trans. A. Kenny .
Oxford : Blackwell .
application of rules: To understand a language means to be
master of a technique ([1953] 1997, 199).
3. Like games, languages are rule governed , but this does not
mean that there are strict and precise rules for each language-
game: Just as there are not rules to legislate for everything in
a game (e.g., there are no rules for how high or how hard one
throws the ball in tennis), language, too, is not everywhere cir-
cumscribed by rules. And as in games, the learning of explicit
rules is not always necessary someone can have learned a
game without ever learning or formulating rules ([1953] 1997,
31) and so, too, in language can the game be learned purely
practically, without learning any explicit rules (1977, 95) .
Te constitutive rules of language are those of grammar
(1974, 184). What Wittgenstein means by grammar is not, how-
ever, what grammarians mean by grammar: It is neither a tax-
onomy of the structural features of a language nor a science
that describes or prescribes the correct or standard usage of
words or arrangement of words. Wittgensteinian grammar is a
generic term for the publicly determined (though this determi-
nation is not due to a concerted, but to an unconcerted, con-
sensus) conventions or conditions (1974, 138, 88) that govern
our meaningful use of words or expressions. Languages are rule
governed, but the rules that govern them are not metalinguistic
norms that exist in advance of use; learning the meaning of a
word is learning how the word is used . Moreover, if grammat-
ical rules determine what it makes sense to say, they cannot
themselves belong to the language-game: a grammatical rule
is a preparation (1993, 72) for a language-game, except in
heuristic language-games (e.g., pedagogical language-games;
language teaching) where the formulation of some rules is the
object of the language-game (the distinction here is an instance
of the use and mention distinction). To learn grammatical
rules is to learn what the conventional conditions and con-
straints of the uses of a word are, which linguistic moves are
meaningful and which are not. Just as the rules of a game con-
stitute the game and its allowable moves, grammar constitutes
language and its allowable moves :
4. Like games, language is embedded in our social, cul-
tural and natural ways of living that is, in our form of life.
Languages cannot be abstracted from the context in which
they live: words have meaning only in the stream of life
(1982, 913). Language is a normative practice, but it is also a
social practice. Any language is founded on convention ([1953]
1997, 355); it is embedded in the shared activities of the lan-
guage users in a given community: To obey a rule, to make a
report, to give an order, to play a game of chess, are customs
(uses, institutions) ([1953] 1997, 199). To understand a lan-
guage-game, one must be either immersed in the community
in which it is embedded or knowledgeable about that commu-
nitys practices: Someone not accustomed to, or aware of, the
practice in some Semitic cultures of saying Hamsa! (Five!)
a verbal conjuring of the fve fngers of the hand that protects
against the evil eye would not understand the purpose of
the utterance. For a language to emerge or be possible, there
has to be something shared. What is shared is a distinct form
of life: the particular biosocial conditions and activities that
make particular languages possible. Te human form of life
could not have produced a feline language, nor a feline form
Language-Game
419
the basis of their co-occurrence statistics (Redington, Chater,
and Finch 1998 ) or their occurrence in frequent frames (Mintz
2003 ). Moreover, studies using modern infant techniques have
shown that very young children are sensitive to all of these
potential sources of information (Jusczyk 1999 ; Gomez and
Gerken 2000) .
An interesting feature of this kind of work is the extent to which
it sheds new light on classical arguments from the poverty of the
stimulus . For example, John Lewis and Jefrey Elman ( 2001 ) have
shown that it is possible for a simple recurrent network to learn
to obey structure-dependent rules such as those involved in the
formation of complex yes-no questions (e.g., Is the boy who is
dancing singing?) in the absence of exposure to complex yes/
no questions. Tis occurs because the knowledge that the net-
work has acquired by processing simple yes/no questions and
complex declaratives constrains the way in which it deals with
complex yes/no questions .
A fnal strand of research has investigated the relation between
variation in language development and variation in the language
to which children are exposed. For example, research on chil-
dren learning English has found a relation between their aux-
iliary development and mothers use of yes/no questions (e.g.,
Can you kick the ball?) in which the auxiliary occurs in stressed
utterance-initial position (see Richards 1990 for a review). One
problem with this kind of research is that covariance in patterns
of within-language variation often makes it dif cult to distinguish
empirically between alternative explanations of the efects that
are found. One way of avoiding this problem is to focus on the
relation between cross-linguistic diferences in patterns of devel-
opment and cross-linguistic variation in the language to which
children are exposed. For example, Daniel Freudenthal and his
colleagues (2007) have recently shown that it is possible to simu-
late cross-linguistic variation in childrens tendency to use non-
fnite verb forms in fnite contexts in English, Dutch, German,
and Spanish as a function of the interaction between one identi-
cal mechanism that learns from the right edge of the utterance
(MOSAIC, Model Of Syntax Acquisition In Children) and the
statistical properties of child-directed speech in these four lan-
guages . MOSAIC produces high proportions of nonfnite verb
forms in fnite contexts in Dutch, German, and English because
in these languages, the verb forms occurring in utterance-fnal
position in the input are much more likely to be nonfnite than
fnite. However, it produces much lower proportions of nonfnite
verb forms in fnite contexts in Spanish because in this language,
fnite verb forms are much more likely than nonfnite verb forms
to occur at the right edge of the utterance .
Julian Pine
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brent , Michael . 1999 . Speech segmentation and word discovery: A com-
putational perspective . Trends in Cognitive Sciences 3 : 294 301.
Freudenthal , Daniel , Julian Pine , Javier Aguado-Orea , and Fernand
Gobet . 2007 . Modelling the developmental patterning of fniteness
marking in English, Dutch, German and Spanish using MOSAIC .
Cognitive Science 31 : 311 41.
Gomez , Rachel , and LouAnn Gerken . 1999 . Infant artifcial language
learning and language acquisition. Trends in Cognitive Sciences
4 : 178 86.
. 1977 . On Certainty . Ed. G. E. M. Anscombe and G. H. von
Wright , trans. D. Paul and G. E. M. Anscombe . Amended 1st ed.
Oxford : Blackwell.
. 1980 . Remarks on Te Philosophy Of Psychology . Vol. 1. Ed.
G. E. M. Anscombe and G. H. von Wright , trans. G. E. M. Anscombe .
Oxford : Blackwell .
. 1982 . Last Writings on the Philosophy of Psychology . Vol. 1. Ed.
G. H. von Wright and Heikki Nyman , trans. C. G. Luckhardt and
M. A. E. Aue . Oxford : Blackwell .
. 1993 . Moores Wittgenstein lectures in 19301933. In
Philosophical Occasions: 19121951 , ed. J. C. Klagge and A. Nordman ,
46114. Indianapolis : Hackett .
LANGUAGE-LEARNING ENVIRONMENT
Tis term refers to the linguistic and sociocultural environment
in which children learn to talk and, in particular, to the language
to which they are exposed. Te frst systematic studies of the lan-
guage-learning environment were conducted in response to the
claim that, like the language used among adults, the language
heard by children was grossly defective full of false starts, gram-
matical errors, and misleading pauses and as such represented
a very poor sample of the language that the child must eventually
learn. Tese studies showed that speech addressed to children
was largely clear, well formed, and semantically and syn-
tactically simpler than speech addressed to adults, leading
some researchers to argue that in simplifying their speech, par-
ents were presenting their children with graded language lessons
that could bear at least some of the burden of explanation for the
childs remarkably swift progress in language learning (see Pine
1994 for a review).
In fact, there are a number of problems with this view. First,
it confuses the notions of facilitating interaction and facilitating
acquisition. Tus, some of the adjustments made by parents to
facilitate interaction (e.g., the extensive use of questions) prob-
ably have the efect of increasing the complexity of the language
to which children are exposed during the early stages. Second,
it is unclear to what extent the adjustments made by Western
middle-class parents generalize across cultures. Indeed, ethno-
graphic researchers have identifed cultures in which parents
react with mirth or horror to the idea of holding conversations
with young language-learning children (see Lieven 1994 for a
review). Tird, the notion that the language-learning environ-
ment somehow facilitates acquisition is theoretically rather vac-
uous in the absence of a reasonably well-specifed theory of how
the child is learning from this environment .
Cognizant of this last problem, more recent work has focused
on the kind of information that is present in the environment
and the way in which this might be exploited by the childs
language-learning mechanisms. Tus, work in computational
modeling has shown that there is a great deal of information in
the statistical structure of human languages that could, in prin-
ciple, be used to solve particular language-learning problems.
For example, work on segmentation has shown that it is pos-
sible to use information about the stress pattern, the phonot-
actics, and the transitional probabilities between syllables in
a language to identify boundaries between words (Brent 1999 ),
and work on the acquisition of syntactic categories has shown
that it is possible to categorize words quite successfully on
Language-Learning Environment
420
ii) there are semantic issues regarding the meaning or interpreta-
tion of the symbols and data structures. In the case of artifactual
computers, this issue is usually conveniently settled by stipula-
tion : Te artifactor gets to say what the symbols represent (e.g.,
bank balances, chess moves). In the case of natural creatures,
there is, of course, no relevant artifactor, and so the meaning
of the symbols must be determined by some natural facts or
relations.
Tere are two (not necessarily exclusive) kinds of candidates
for a theory of meaning of a LOT:
i. Meaning is spelled out in terms of some of the symbols cru-
cial causal/ conceptual roles in relation to other symbols, mirror-
ing patterns of inference. Tis is a natural suggestion for logical
symbols, such as and and not: A symbol, @, might mean
and because states entokening sentences p and q separately
might cause and in turn be caused by a state entokening p@q
by itself (see Block 1986; Peacocke 1994 ).
ii. Meaning is spelled out in terms of causal relations that
the individual symbols bear to phenomena in the world. For
example, a symbol S might be entokened in a creatures brain in
a way that covaries with the presence of a certain shape before
the creatures eyes in various circumstances, such as under ideal
conditions, under evolutionarily selective ones, or under ones
that display a certain counterfactual structure (see Dretske 1988 ;
Stalnaker 1984 ; Fodor 1987) .
Te chief rivals to the LOT hypothesis are either one or another
form of interpretativism , according to which a creature has prop-
ositional attitudes only because ascription of such states permits
the most rational interpretation of their behavior (see Davidson
[1973] 1984; Dennett 1987 ), and/or the brain structures underly-
ing that ascription are of a radical connectionist nonsyntacti-
cally structured sort (see Smolensky 1988 ). Its sometimes also
thought that some kind of system of imagistic representations
would not only be truer to introspection but also explain various
response-time results that suggest that people think in images
(see Kosslyn 1994 ).
Te main reasons for believing in a LOT as opposed to these
rivals is that it would explain a number of interesting phenom-
ena associated with the mind. Salient among these are the
following:
1. Te propositional structure of attitudes: Te standard
object of, for example, a thought, belief, desire, hope, or expec-
tation is some kind of truth - valuable object, most perspicuously
expressed by a sentence; neither images nor connectionist
networks are able to systematically represent logically com-
plex thoughts, involving, for example, negations and nested
quantifers (what image could express Not everyone loves
someone?) .
2. Te causal ef cacy of thought: A thought can cause bodily
states and movements because it is a structure entokened in the
brain .
3. Te productivity of thought: People can potentially under-
stand an infnitude of diferent thoughts formed by logical com-
binations of simpler ones, for example, Its possible for every
cat to chase some rat that eats some cheese that that lives in
the house that Jack built, since the LOT entokened in their brain
can (under reasonable idealization) produce a corresponding
infnity of sentences .
Jusczyk , Peter . 1999 . How infants begin to extract words from speech .
Trends in Cognitive Sciences 3 : 323 8.
Lewis , John , and Jefrey Elman . 2001 . Learnability and the statistical
structure of language: Poverty of stimulus arguments revisited. In
Proceedings of the Twenty-Sixth Annual Boston University Conference
on Language Development , 35970. Somerville, MA : Cascadilla .
Lieven , Elena . 1994 . Crosslinguistic and crosscultural aspects of
language addressed to children. In Input and Interaction in
Language Acquisition , ed. Clare Gallaway and Brian Richards , 5673.
Cambridge : Cambridge University Press .
Mintz , Toby . 2003 . Frequent frames as a cue for grammatical categories
in child-directed speech . Cognition 90 : 91 117.
Pine , Julian . 1994 . Te language of primary caregivers. In Input and
Interaction in Language Acquisition , ed. Clare Gallaway and Brian
Richards , 1537. Cambridge : Cambridge University Press .
Redington , Martin , Nick Chater , and Steven Finch . 1998 . Distributional
information: A powerful cue for acquiring syntactic categories .
Cognitive Science 22 : 425 69.
Richards , Brian . 1990 . Language Development and Individual
Diferences: A Study of Auxiliary Verb Learning . Cambridge : Cambridge
University Press .
LANGUAGE OF THOUGHT
Tis is a special language that has been postulated by a number
of writers G. Harman ( 1972 ), J. Fodor ( 1975 , 1987) to explain
how humans and many animals represent and think about the
world. Te language of thought (LOT) is claimed to be coded,
or entokened, in their brains, rather than in the way certain
formal languages are entokened in the circuitry of a computer.
What makes the LOT a language is that it possesses semantically
valuable, causally ef cacious logico-syntactic structure : Tat is, it
consists, for example, of names, predicates, variables, quantifers
(all, some) logical connectives (not, and, only if) and
operators (possibly, probably) that are combined to form
complex sentences that can be true or false.
Te LOT need not be the natural language (e.g., English,
Chinese), if any, that a creature speaks , although some writers
have supposed that in adult humans the two may coincide (pro-
viding an interesting perspective in the Sapir-Whorf hypothesis
that thought is determined by language). Indeed, given that the
relevant sorts of intelligent behavior are displayed by many infra-
linguistic creatures infants, chimpanzees its postulation need
not be confned to natural language users. Nor need the LOT be
in the least conscious or introspectible: A persons thought pro-
cesses might be explained by a LOT, while introspectively his or
her mental life might seem to consist wholly of images, feel-
ings, and inarticulate impulses. Most importantly, processing a
LOT need not require any sort of intelligent creature to read and
understand the sentences being processed in the brain. Along
lines set out by Alan Turing, the processing of the LOT symbols
can be executed purely mechanically.
As in discussing any language, there are two issues raised by
such a postulation: i) Tere are syntactic issues regarding the
character of the actual symbols and the computations defned
over them. Tis is the kind of issue regularly addressed in rich
detail by programmers dealing with artifactual computers, and
with psychologists dealing with naturally occurring, living crea-
tures, in their concern with data structures and algorithms for
dealing with them, for example, for vision or reasoning; and
Language of Thought
421
with or claiming authority to modify the practices or beliefs of
other speakers. Practices, beliefs, and management can be stud-
ied separately, although they turn out to be interdependent .
Early language managers were the Sanskrit and Arabic gram-
marians who guarded the purity of sacred texts, the medieval
European rulers who switched from Latin to the vernacular for
legal matters, and the language nationalists in the nineteenth
century who made their national language diferent from that of
their previous ruler.
Einar Haugen ( 1966 ) described how rival political ideolo-
gies supported competing invented varieties of Norwegian and
compromised by requiring schoolchildren to learn both. In the
1960s, scholars became interested in the language-planning
problems of postcolonial Africa and Asia (Fishman, Ferguson,
and Das Gupta 1968 ). Tey focused on status planning (mak-
ing one variety of language of cial or designating it for school
use) and corpus planning (changes in the language itself, such
as standardizing it or providing it with a writing system or new
terminology) (Kloss 1966 ). Others concentrated on the culti-
vation and standardization of developed European languages
(Nekvapil 2007).
In status policy, the problem was to decide between the
demands of contending varieties; as a decision depended on
nonlinguistic values, such as the power of social or ethnic or eco-
nomic groups, language planners had little real infuence. Tere
was work to do in language academies or in writing textbooks
to purify the linguistic usage of schoolchildren. In the 1970s,
some scholars tried to evaluate the efect of corpus planning,
but it proved easier to keep doing it than to study its efective-
ness (Fishman 1977). Language policy expanded when Robert
L. Cooper ( 1989 ) added a third area, language-acquisition plan-
ning, or language-education management, the efort to increase
the number of speakers. Related is language difusion, govern-
ments working to spread their language outside their political
boundaries (Cooper 1982 ) .
Rights and Theories
In the last half century, a further development has been the
study and promotion of human or civil rights associated with
language (Laitin and Reich 2003 ; May 2005 ; Skutnabb-Kangas,
Phillipson, and Rannut 1995 ). Building on principles frst pro-
posed after World War I, several international covenants of lan-
guage rights for minorities have been formulated, and some have
been adopted by international bodies such as UNESCO and the
European Community; a smaller number have been ratifed by
nation-states, and a few have been implemented .
Tere is no generally accepted theory of language policy.
Tomas Ricento ( 2001 ), in fact, argues that there cannot be one.
However, Joshua A. Fishman (1991) has presented a model of
reversing language shift , which includes a graded intergenera-
tional disruption scale intended both to describe the state of a
language and its likelihood of being maintained and to suggest
how to resist further loss or reestablish earlier strength. Bernard
Spolsky ( 2004 ) has proposed that language policy has three
components (language practices, language ideology or beliefs,
and language management), and has sketched a theory based
on this proposal. Ji Nekvapil ( 2006 ), following Bjoern Jernudd
and J. V. Neustupn ( 1987 ), has put forward a theory of language
4. Te systematicity of thought: If people can think some
thought p , then they can also think all logical permutations of
p ; for example, people can think If John leaves, then someone
insulted Mary if and only if they can think If someone leaves, then
Mary insulted John , since they can mechanically recombine the
simple expressions of a sentence .
5. Te intensionality of thought: People can think of things
in one way without thinking of those very things in another; for
example, they can think that the morning star is Venus without
thinking that the evening star is, and they can think about dif-
ferent nonexistent things , such as Zeus and Santa Claus. Indeed,
the LOT can explain the hyperintensionality of thought: People
can think about things that are even necessarily identical , as
when one thinks that Mark Twain but not Sam Clemens is
funny, because they employ correspondingly diferent LOT
symbols .
6. Te multiple roles of attitudes: Diferent attitudes can be
directed upon the very same thought; for example, people can
believe, desire, suspect, doubt the same thought, that God exists .
Fodor ( 1975 , 1987 ) and Georges Rey ( 1997 ) have argued that
these and other phenomena cannot be explained by the rival
views without substantial, additional ad hoc assumptions, for
example, that certain images or networks express logically com-
plex thoughts and are causally related in the aforementioned
ways. For these reasons, the language of thought is to be pre-
ferred on empirical grounds (for more of an a priori argument
see, e.g., Davies 1991 ) .
Georges Rey
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Block. N. 1986 . Advertisement for a conceptual role semantics. In
Studies in the Philosophy of Mind . Ed. P. French, T. Uehling, and
H. Wettstein , 61578. Minneapolis: University of Minnesota Press.
Davidson , D. [1973] 1984. Radical interpretation. In Inquiries into Truth
and Interpretation . Oxford : Clarendon .
Davies , M. 1991 . Concepts, connectionism, and the language of thought.
In Philosophy and Connectionist Teory , ed. W. Ramsey et al., 22956.
Hillsdale, NJ : Erlbaum .
Dennett , D. 1987 . Te Intentional Stance . Cambridge, MA : MIT Press
Dretske , F. 1988 . Explaining Behavior: Reasons in a World of Causes .
Cambridge, MA : MIT Press
Fodor , J. 1975 . Te Language of Tought . New York : Crowell .
. 1987 . Psychosemantics , Cambridge, MA : MIT Press .
Harman , G. 1972 . Tought . Princeton, NJ : Princeton University Press .
Kosslyn , S. 1994 . Image and Brain: Te Resolution of the Imagery Debate .
Cambridge, MA : MIT Press .
Peacocke , C. 1994 . A Study of Concepts . Cambridge, MA : MIT Press .
Rey , G. 1997 . Contemporary Philosophy of Mind . Oxford : Blackwell .
Smolensky , P. 1988 . On the proper treatment of conectionism .
Behavioral and Brain Sciences 11 : 59 74.
Stalnaker , R. 1984 . Inquiry . Cambridge, MA : MIT Press .
LANGUAGE POLICY
Defnition and History
Language policy studies the regular choices among varieties and
variants within a speech community the language practices
of members of the community, their beliefs about the values to
be assigned to the varieties, and eforts by individuals or groups
Language Policy
422
2,000 varieties. Central Europe and the Balkans repeated this
process, as the partition of Czechoslovakia has led to of cial sta-
tus for Czech and Slovak (Neustupn and Nekvapil 2003 ), and
the division of Yugoslavia has now led to eforts to distinguish
Serbian and Croatian (Pranjkovic 2001 ) .
Belgium and Switzerland use territoriality to resolve language
confict. Externally, both are believed to be bilingual. In fact,
Belgium is divided into language regions, some of which are of -
cially French-speaking, others of cially Dutch-speaking, and a
few of cially German-speaking. Only Brussels is of cially bilin-
gual. Te varieties spoken in these regions are neither Dutch nor
French but regional dialects; as a result, 40 percent of Belgian
high school students report that they are taught in a language
that they do not speak at home (Aunger 1993 ) . In Switzerland,
each canton establishes its own language policy, choosing among
German, French, Italian, and Romansch. Knowledge of a second
language (other than the expanding use of English associated
with globalization) is no better than in other European countries
(Harlow 2004 ; Hrdegen 2001 ) .
Te special language problems of Africa were produced by
the fact that the borders drawn by colonizing European powers
in the nineteenth century did not coincide with ethnic, tribal, or
linguistic boundaries. After independence, African states had
to choose among a variety of languages, most of which were
also spoken in a neighboring states (Bamgbose 2000 ). Colonial
educational policy had favored the use of European metropoli-
tan languages, absolutely in the case of French and Portuguese
colonies and, after initial use for a few years of local vernaculars,
in British colonies. Partly because choice of any one vernacular
would provide excessive power to its speakers, partly because
the elite already spoke the metropolitan language, and partly
because of inertia, postindependence eforts to establish the sta-
tus of African languages have generally failed (Phillipson 1992 ) .
Globalization and Local Resistance
Globalization has a major impact on language policy. One efect
has been the unparalleled difusion of English, the most widely
used second language in most of the world. English is the favored
frst foreign language in all European countries, spreading also
into former Soviet nations. In Asia, English is the lingua franca
for intercommunication among Japanese, Chinese, Koreans, and
Tais. International corporations, even those located in European
countries, tend to prefer English. Foreign language teaching is a
topic of interest mainly in English-speaking countries; elsewhere,
the major concern is English language teaching .
Te protection of endangered languages is a recent concern
of many language policy scholars. Tey have noticed the rapid
loss of smaller minority languages, estimating that most of the
current 6,000 languages in the world will disappear in the next
hundred years (Krauss 1991 ). Te threat comes not just from
world languages like Spanish (which has virtually denuded
Latin America of its rich linguistic diversity) or French (with its
strong monolingual ideology) or English (universally feared as
the exemplar of linguistic imperialism) but also from stronger
local languages like Swahili. Fishman (1990) provides a set of
benchmarks for studying loss and suggests how to reverse it. So
far, the most successful eforts at reversal have been associated
with grants of political autonomy, as in Spain (Hofmann 1995 ),
management, ranging from individual self-correction to the
organized management of all micro and macro levels. Detailed
descriptions of language policy such as Grenoble ( 2003 ), Kaplan
and Baldauf ( 2003 ) and Zhou ( 2004 ) have started to clarify the
complex dimensions that a theory must handle .
The Politics of Language Policy
One of the most critical facts or beliefs about language vari-
eties concerns their power. There are many nation-states
that assume monolingualism to be ideal and combine this
assumption with a belief in the value, beauty, efficiency,
and desirability of their own national variety. This is true of
English-speaking nations, although it is challenged by coun-
terassertions in South Africa, which has a long tradition of
claims for Afrikaans and has recently extended nominal rec-
ognition to nine African languages, and in Canada by the lan-
guage-related claims of Quebec for independence. The belief
was first manifested in Spain, which continued its search for
purity after the expulsion of Moors and Jews with a procla-
mation carried to the New World of the value of Spanish; this
resulted in the virtual destruction of Native American lan-
guages. The belief in the importance of a single national vari-
ety was adopted by the Jacobins during the French Revolution
and gradually implemented in France and French territories
(Ager 1999 ): The difficulty of its implementation continues
to be demonstrated by the need to pass new laws and regula-
tions. German Romanticism and nationalism (Fishman 1973 )
provided an ideological base with the proclamation of the
truth of one nation, one language. Another example of ideo-
logical monolingualism is Japan, which during its period of
colonial expansion required conquered peoples to switch to
Japanese, and which has only recently taken note of minority
languages (Katsuragi 2005 ).
Commonly, the existence of two or more major languages
within a single nation-state or confederation is associated with
political confict. One resolution is to favor a single variety, either
that spoken by the majority or that controlled by the dominant
elite. In the Soviet Union, a Lenin-inspired policy of recognizing
minority languages to speed up the spread of communism was
replaced under Stalin by a Russifcation policy (Grenoble 2003 ;
Lewis 1972 ). After the collapse of the Soviet Union, most of the
newly independent states reasserted the signifcance of their
territorial languages, so that currently each of the former Soviet
states (including Russia itself) appears to be working toward
monolingualism in the territorial language (Landau and Kellner-
Heinkele 2001 ; Ozolins 2003 ) .
Territoriality
A second solution to problems associated with having multiple
major languages within a single nation-state is territoriality.
India tried to base its internal political divisions on language.
Te partition into a Hindu-dominated India and Moslem-
dominated Pakistan was paralleled by a language-management
efort to divide what was previously considered a single lan-
guage, Hindustani, into two Hindi written in Devanagari script
and Urdu with Perso-Arabic script (Annamalai 2001 ). Te split-
ting up of India into states refected major language diferences,
although it could not capture the complexity of a nation with
Language Policy
423
Fishman , Joshua A. , Charles A. Ferguson , and Jyotirinda Das Gupta . 1968 .
Language Problems of Developing Nations . New York : Wiley .
Grenoble , Lenore A. 2003 . Soviet Language Policy . Dordrecht, the
Netherlands : Kluwer Academic Publishers.
Harlow , Ray. 2004 . Switzerland. In Encyclopedia of Linguistics , ed.
P. Strazny . London : Taylor and Francis .
Haugen , Einar. 1966 . Language Confict and Language Planning: Te
Case of Modern Norwegian . Cambridge : Harvard University Press .
Hofmann , Charlotte . 1995 . Monolingualism, bilingualism, cultural
pluralism and national identity: Twenty years of language planning
in contemporary Spain . Current Issues in Language and Society
2.1 : 59 90.
Hornberger , Nancy H. , and Kendall A. King . 2001 . Reversing language
shift in South America. In Can Treatened Languages be Saved? ed.
J. A. Fishman , 16694 . Clevedon and Avon : Multilingual Matters .
Hrdegen , Stephan . 2001 . Te Fribourg linguistic case controversy
about the language of instruction in schools in the light of freedom
of language and equal educational opportunities in Switzerland .
European Journal for Educational Law and Policy 5 : 73 82.
Jernudd , Bjoern , and J. V. Neustupn . 1987 . Language planning: For
whom? In Proceedings of the International Colloquium on Language
Planning , ed. L. LaForge , 7184 . Quebec : Presses de lUniversite Laval .
Kaplan , Robert B. , and Richard B. Baldauf . 2003 . Language and
Language-in-Education Planning in the Pacifc Basin . Dordrecht, the
Netherlands : Kluwer Academic Publishers.
Katsuragi , Takao . 2005 . Japanese language policy from the point of view
of public philosophy . International Journal of the Sociology of lan-
guage 175 /176 : 41 54.
Kloss , Heinz . 1966 . German-American language maintenance eforts.
In Language Loyalty in the United States , ed. J. Fishman , 20652 . Te
Hague : Mouton .
Krauss , Michael . 1991 . Te worlds languages in crisis . Language
68.1 : 4 10.
Laitin , David D. , and Robert Reich . 2003 . A liberal democratic
approach to language justice. In Language Rights and Political
Theory , ed. W. Kymlicka and A. Patten , 80104 . Oxford : Oxford
University Press .
Landau , Jacob , and Barbara Kellner-Heinkele . 2001 . Politics of Language
in the Ex-Soviet Muslim States: Azerbaijan, Usbekistan, Kazakhstan,
Kyrgyzstan, Turkmenistan and Tajikistan . London and Ann Arbor :
C.Hurst & Co. and University of Michigan Press .
Lewis , E. Glyn . 1972 . Multilingualism in the Soviet Union . Te
Hague : Mouton .
May , Stephen. 2005 . Language rights: Moving the debate forward .
Journal of Sociolinguistics 9.3 : 319 47.
May , Stephen , and Richard Hill . 2005 . Mori-medium education: Current
issues and challenges . International Journal of Bilingual Education
and Bilingualism 8.5 : 377 403.
McCarty , Teresa L. 2003 . Revitalising indigenous languages in homoge-
nising times . Comparative Education 29.2 : 147 63.
Nekvapil , Ji. 2006 . From language planning to language management .
Sociolinguistica 20 : 92 104.
. 2007 . Language cultivation in developed contexts. In Handbook
of Educational Linguistics , ed. B. Spolsky and F. M. Hult , 25165 .
Oxford : Blackwell .
Neustupn , J.V. , and Ji Nekvapil . 2003 . Language management
in the Czech republic . Current Issues in Language planning
4.3 /4: 181 366.
Laoire , Muiris . 1996 . An historical perspective of the revival of Irish
outside the Gaeltacht, 18801930, with reference to the revitaliza-
tion of Hebrew. In Language and State: Revitalization and Revival in
Israel and Eire , ed. S. Wright , 5175 . Clevedon and Avon : Multilingual
Matters .
the United Kingdom ( Laoire 1996 ; Coupland et al. 2006), and
Canada (Bourhis 2001 ). Tere are also eforts in New Zealand
(Spolsky 2005; May and Hill 2005 ), in South America (Hornberger
and King 2001 ), and among other indigenous peoples (McCarty
2003 ; Omoniyi 2003 ) .
Speakers of major languages also fear language loss. Tis
can be seen in Spain, with the sensitivity of its Academy to lan-
guage change; in France, with its growing number of regulations
and language agencies; in Russia, with its refusal to recognize
non-Cyrillic alphabets for minority languages and its claim of
defending the language rights of Russian-speakers in former
Soviet states; and even in the United States, where an English-
only movement is struggling against what it sees as the threat of
Spanish and other immigrant languages to the survival of what
most people believe to be the strongest language in the world
(Baron 1990 ).
Language policy is a new and rapidly developing feld, the
urgency and seriousness of which has resulted in activism by
scholars who feel responsible for correcting what they see as
injustices or blindness to the potential loss of linguistic diversity,
as well as in academic attempts to develop theories to explain
data from increasingly detailed descriptions of situations and
policies .
Bernard Spolsky
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ager , Dennis E. 1999 . Identity, Insecurity and Image: France and Language .
Clevedon and Philadelphia : Multilingual Matters.
Annamalai , E. 2001 . Managing Multilingualism in India: Political and
Linguistic Manifestations . New Delhi : Sage .
Aunger , Edmund A. 1993 . Regional, national and of cial languages
in Belgium . International Journal of the Sociology of Language
104 : 31 48.
Bamgbose , Ayo . 2000 . Language and Exclusion: Te Consequences of
Language Policies in Africa, Beitrage zur Afrikanistik . Mnster and
Hamburg : LIT Verlag .
Baron , Dennis E. 1990 . Te English Only Question . New Haven, CT : Yale
University Press .
Bourhis , Richard Y. 2001 . Reversing language shift in Quebec. In Can
Treatened Languages be Saved? ed. J. A. Fishman 101 41. Clevedon
and Avon : Multilingual Matters .
Cooper , Robert L. 1989 . Language Planning and Social Change .
Cambridge : Cambridge University Press .
Cooper , Robert L. , ed. 1982. Language Spread: Studies in Difusion and
Social Change . Bloomington : Indiana University Press .
Coupland , Nicholas , Hywel Bishop , Betsy Evans , and Peter Garrett . 2006 .
Imagining Wales and the Welsh language: Ethnolinguistic subjectivi-
ties and demographic fow . Journal of Language and Social Psychology
25.4 : 351 76.
Fishman , Joshua A. 1973 . Language and Nationalism: Two Integrative
Essays . Rowley, MA : Newbury House .
. 1977 . Comparative study of language planning: Introducing a sur-
vey. In Language Planning Processes , ed. J. Rubin , B. H. Jernudd , J. Das
Gupta , J. A. Fishman and C. A. Ferguson , 3140 . Te Hague : Mouton .
. 1990 . What is reversing language shift (RLS) and how can it suc-
ceed? Journal of Multilingual and Multicultural Development 11 .1/2:
5 36.
. 1991 . Reversing Language Shift: Teoretical and Empirical
Foundations of Assistance to Treatened Languages .
Clevedon : Multilingual Matters .
Language Policy
424
Laws in the Study of Language and Text
In quantitative linguistics , the exact science of language
and text, distributional and functional kinds of laws are known.
Te frst kind takes the form of probability distributions; that is,
it makes predictions about the number of units of a given prop-
erty. A well-known example of this kind is the Zipf-Mandelbrot
Law. Te status of the corresponding phenomenon has been
discussed since the days of George K. Zipf , who was the frst to
systematically study quantitative properties of language from
a scientifc point of view. Te law relates a) the frequency of a
word in a given text (in any language) to the number of words
with the given frequency (called frequency spectrum ) and b) the
frequency of a word in relation to its rank (called rank-frequency
distribution ). Te frst formulation by Zipf was later modifed
and corrected by Benoit Mandelbrot , who derived the law from
the assumption that languages optimize their lexicons with
respect to code-production efort in the long run. Tis resulted
in the famous formula (1), which has the form of a rank-fre-
quency distribution: If the words are arranged according to their
frequency, the most frequent word is assigned rank one, and so
on. Te formula gives the frequency that a word should have at
a given rank:
(1)
f (r) =
K
(b+r)
where f ( r ) is the frequency, r the rank, b and empirical param-
eters, and K a normalizing constant that makes the probabilities
sum up to 1.0.
Since the seminal works of Zipf and Mandelbrot, numerous
laws have been found. Other examples of distributional laws
are (in morphology and lexicon) the distribution of length,
polysemy, synonymy, age, part of speech (see word classes ),
and so on, (in syntax ) the frequency distribution of syntac-
tic constructions, the distribution of their complexity, depth of
embedding, information, and position in mother constituent; (in
semantics ) the distribution of the lengths of paths in semantic
networks (see also semantic fields ), semantic diversifca-
tion, and so on. Any property and any linguistic unit studied so
far displays a characteristic probability distribution .
Te second kind of law is called the functional type, because
these laws link two (or more) properties. An illustrative example
of this kind is Menzeraths Law (also called Menzerath-Altmann
Law), which relates the size of linguistic constituents to the size
of the corresponding construct. Tus, the (mean) length of the
syllables of a word depends on the number of syllables the
word consists of; the (mean) length of the clauses in a sentence
depends on the length of the sentence (measured in terms of the
number of clauses it consists of). Te most general form of this
law is given by formula (2):
(2)
y = Ax
b
e
-cx
where y is the mean length of the constituents, x the length of the
construct, and A, b , and c are parameters. (Tis law predicts the
function [2] but not the values of its parameters. Tey are esti-
mated empirically on the data under analysis. Future research
may provide an enhanced version of the law that will also deter-
mine these parameters.) Experience shows that the parameters
are determined mainly by the level of the units under study. Tey
Omoniyi , Tope . 2003 . Local policies and global forces: Multiliteracy and
Africas indigenous languages . Language Policy 2.2 : 133 52.
Ozolins , Uldis . 2003 . Te impact of European accession upon language
policy in the Baltic States . Language Policy 2.3 : 217 38.
Phillipson , Robert . 1992 . Linguistic Imperialism . Oxford : Oxford
University Press .
Pranjkovic , Ivo . 2001 . Te Croatian standard language and the Serbian
standard language . International Journal of the Sociology of Language
147 : 31 50.
Ricento , Tomas , ed. 2001. Ideology, Politics and Language Policies: Focus
on English. Amsterdam and Philadelphia : John Benjamins .
Skutnabb-Kangas , Tove , Robert Phillipson , and Mart Rannut . 1995 .
Linguistic Human Rights: Overcoming Linguistic Discrimination . Berlin
and New York : Mouton de Gruyter .
Spolsky , Bernard . 2004 . Language Policy, Key Topics in Sociolinguistics .
Cambridge : Cambridge University Press .
. 2005 . Maori lost and regained. In Languages of New Zealand , ed.
A. Bell , R. Harlow, and D. Starks , 6785 . Wellington : Victoria University
Press .
Zhou , Minglang , ed. 2004. Language Policy in the Peoples Republic
of China: Teory and Practice since 1949 . Dordrecht, the
Netherlands : Kluwer Academic Publishers .
LAWS OF LANGUAGE
The Concept of Law
Te philosophy of science defnes the term scientifc law as a
meaningful universal hypothesis that is systematically connected
to other hypotheses in the feld and, at the same time, well cor-
roborated on relevant empirical data (cf. Bunge 1967 ). A law is
called universal because it is valid at all times, everywhere, and
for all objects of its scope.
A system of laws is called a theory . Te construction of a the-
ory is the highest and most demanding goal of scientifc research
and can be undertaken only if and when a number of interre-
lated laws have been found. Tere is much confusion about the
term theory, especially in linguistics, where all kinds of formal-
isms, thoughts, approaches, descriptive tools, defnitions, and
concepts are called theories. Te philosophy of science distin-
guishes two kinds of theories: 1) the axiomatic theories of logics
and mathematics and 2) the empirical theories in the factual sci-
ences. While the frst ones make statements only within a given
axiomatic system and can be used only to construct analytical
truths, the latter ones make statements about parts of the world.
Te truth of an empirical theory and of its elements, the laws,
depends not only on internal correctness but also on the cor-
respondence with the facts of reality although every empirical
theory must have an axiomatic kernel .
Te value of theories and their components, the laws, lies not
only in their role as the containers of scientifc knowledge but
also in the fact that there can be no explanation without at least
one law: A valid scientifc explanation (the so-called deductive-
nomological explanation; cf. Hempel and Oppenheim 1948 ) is a
subsumption under laws taking into account boundary condi-
tions. Laws must not be confused with rules, which are either
prescriptive or descriptive tools without any explanatory power;
hence, grammars and similar formalisms also cannot explain
anything. Another signifcant diference is that rules can be vio-
lated laws (in the scientifc sense) cannot .
Laws of Language
425
(/vart/ > /vurde/) in the time period from 1445 to 1925. As the
graph shows, the replacement was very limited for the frst 200
years, but even a much shorter time span can provide enough
information to predict the development over the next several
hundred years .
Another variant of this third kind of law is based on (discrete)
linguistic instead of (continuous) physical time. Te simplest
way to operationalize linguistic time is the reference to text
position. In oral texts, there is a direct correspondence of the
sequence of linguistic units to physical time intervals.
Several linguistic characteristics can be investigated using
indices that relate their frequency to current text position, among
them the type-token-ratio (TTR). At each text position, the num-
ber of types occurred to that point is counted, which yields a
monotonously increasing curve, because the number of words
used before a given text position cannot decrease in the course
of the rest of the text. A straightforward theoretical derivation of
this law was given by Gustav Herdan (1966), represented by the
simple formula (4):
(4) y =ax
b
where y is the number of types, x the number of tokens (= text
position), and b a text characteristic. Te parameter b is also an
indicator of the morphological type of the language under study
if word forms are considered because morphologically rich lan-
guages display a faster increase in word-form types than isolat-
ing languages.
A problem of the TTR, if used for text comparison, is that it is
not independent of the overall text length. Terefore, more com-
plicated formulae are used to take this infuence into account or
quite diferent models (cf. Popescu and Altmann 2006 a, 2006b)
are applied.
Recent investigations have found that other linguistic units
show a similar behavior in their text dynamics (letters, morphs,
syntactic constructions, syntactic function types, etc.). However,
depending on the size of their inventory in language (which may
vary over several orders of magnitude compare, e.g., the size
of an alphabet or a phoneme system to the size of a lexicon),
diferent models have to be used. Te TTR of syntactic units, for
example, is shown in Figure 3 .
Theory Construction
Currently, there are two approaches to the construction of a lin-
guistic theory (in the sense of the philosophy of science): 1) syn-
ergetic linguistics and 2) Gejza Wimmer and Gabriel Altmanns
unifed theory.
The basic idea behind synergetic linguistics (cf. Khler 1986 ,
2005 ) is the aim to integrate the separated laws and hypoth-
eses found so far into a complex model that not only describes
the linguistic phenomena but also provides a means to explain
them. This is achieved by introducing the central axiom that
language is a self-regulating and self- organizing system .
An explanation of the existence, properties, and changes of
linguistic, more generally semiotic, systems is not possible
without the aspect of the (dynamic) interdependence of struc-
ture and function. The genesis and evolution of these systems
must be attributed to repercussions of communication upon
structure (cf. Bunge 1998 and Khler and Martinkov 1998) .
increase from the level of sound length gradually to the sentence
and suprasentence level. Figure 1 gives an impression of a typical
curve. Other examples are the dependence of word (or morph)
frequency on word (or morph) length, and the frequency of syn-
tactic constructions on their complexity, of polysemy on length,
of length on age, and so on .
A special variant of a functional law is the developmental one.
Here, a property is related to time. Te best-known example is
the Piotrowski Law, which represents the development (increase
and/or decrease) of the portion of new units or forms over time.
Tis law is a typical growth process and can be derived from a
simple diferential equation with the solution (3):
(3)
bt
ae
c
p =
+ 1
where p is the proportion of new forms at time t, c is the
saturation value, and a and b are empirical parameters.
Figure 2 shows the increase of the forms with /u/ at the cost
of the older form with /a/ in the German word ward > wurde
0 2 4 6 8
2.0
2.2
2.4
2.6
2.8
3.0
3.2
Figure 1. The functional dependence of mean syllable length ( y -axis) on
word length ( x -axis) in Hungarian. The line represents the prediction; the
marks show the empirical data points.
Figure 2. Typical curve representing the replacement of a linguistic unit
by a new one.
Laws of Language
426
. 1998 . Semiotic systems. In Systems: A New Paradigm for the
Human Sciences , ed. Gabriel Altmann and Walter A. Koch , 33749.
Berlin and New York : Walter de Gruyter .
Haken , Hermann . 1978 . Synergetics . Berlin and Heidelberg : Springer .
Haken , Hermann , and R. Graham . 1971 . Synergetik. Die Lehre vom
Zusammenwirken . Umschau 6 : 191 .
Hempel , Carl G. , and P. Oppenheim . 1948 . Aspects of scientifc explana-
tion . Philosophy of Science 15 : 135 75.
Herdan , Gustav . 1966 . Te Advanced Teory of Language as Choice and
Chance . Berlin : Springer .
Hebek , Ludek . 1997 . Lectures on Text Teory . Prague : Oriental
Institute .
Khler , Reinhard . 1986 . Zur linguistischen Synergetik. Struktur und
Dynamik der Lexik . Bochum, Germany : Brockmeyer .
. 1990 . Elemente der synergetischen Linguistik . Glottometrika
12 : 179 88.
. 1995 . Bibliography of Quantitative Linguistics = Bibliographie zur
quantitativen Linguistik = Bibliografja po kvantitativnoj lingvistike .
Amsterdam : Benjamins .
. 2005 . Synergetic linguistics. In Quantitative Linguistik. Ein
internationales Handbuch. Quantitative [Linguistics: An International
Handbook] , ed.Reinhard Khler , Gabriel Altmann , and Rajmund G.
Piotrowski , 76075. Berlin and New York : Walter de Gruyter .
Khler , Reinhard , and Zuzana Martinkov . 1998 . A systems theoreti-
cal approach to language and music. In Systems: A New Paradigm for
the Human Sciences , ed. Gabriel Altmann and Walter A. Koch , Berlin,
51446. New York : Walter de Gruyter .
Mautek , Jn , and Gabriel Altmann . 2007 . Discrete and Continuous
Modeling in Quantitative Linguistics . Journal of Quantitative
Linguistics 14 : 8194.
Popescu , Ioan-Iovitz , and Gabriel Altmann . 2006a . Some aspects of
word frequencies . Glottometrics 13 : 23 46.
. 2006 b. Some geometric properties of word frequency distribu-
tions. Gttinger Beitrge zur Sprachwissenschaft 13 : 8798.
Wimmer , Gejza , and Gabriel Altmann . 2005 . Unifed derivation of
some linguistic laws. In Quantitative Linguistik. Ein internationales
Handbuch. [Quantitative Linguistics: An International Handbook] ,
ed. Reinhard Khler , Gabriel Altmann , and Rajmund G. Piotrowski ,
760 75. Berlin and New York : de Gruyter .
Zipf , George Kingsley . [1935] 1968 . Te Psycho-Biology of Language: An
Introduction to Dynamic Philology . 2d ed. Boston: Houghton-Mif in.
Cambridge, MA : MIT Press .
. 1949 . Human Behaviour and the Principle of Least Efort . Reading,
MA : Addison-Wesley .
Synergetic modeling in linguistics starts from axiomatically
assumed requirements that a semiotic system must meet: the
coding requirement (semiotic systems have to provide a
means to create meaningful expressions) and the require-
ments of coding and decoding efficiency, of memory saving,
of transmission security, of minimization of effort, and many
others.
Te other approach at theory construction in linguistics is
Wimmer and Altmanns unifed theory. Integration of sepa-
rately existing laws and hypotheses starts from a very general
diferential (alternatively: diference) equation, as well as two
very general assumptions: 1) If y is a continuous linguistic vari-
able (i.e., some property of a linguistic unit), then its change
over time or with respect to another linguistic variable will be
determined in any case by its temporary value. Hence, a cor-
responding mathematical model should be set up in terms of
its relative change ( dy/y ). 2) Te independent variable that
has an efect on y also has to be taken into account in terms
of its relative change (i.e., dx/x ). Te discrete approach is ana-
logical; one considers the relative diference y
x
/y
x
. Hence, the
general formulas are dy/y = g(x)dx and y
x-1
/ y
x-1
= g(x). Te
solutions of these equations are quite interpretable linguisti-
cally and yield the same results as the synergetic approach. Te
great majority of laws known up to now can be derived from
these equations.
Both models, the unifed and the synergetic, turn out to be
two representations of the same basic assumptions. Te syner-
getic model allows easier treatment of multiple dependencies for
which partial diferential equations must be used in the unifed
model .
Reinhard Khler
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Altmann , Gabriel . 1980 . Wiederholungen in Texten . Bochum, Germany :
Brockmeyer .
Bertalanfy , Ludwig van . 1968 . General System Teory: Foundations,
Development, Applications . New York : George Braziller .
Bunge , Mario . 1967 . Scientifc Research I, II . Berlin and Heidelberg :
Springer .
0 300 600 900 1200 1500
Tokens
0
50
100
T
y
p
e
s
Figure 3. The TTR of syntactic constructions in a text.
The smooth line corresponds to the prediction; the
irregular line represents the empirical data.
Laws of Language
427
L0, which is an infnite set consisting of the symbol a repeated an
arbitrary number of times:
L0 = { a, aa, aaa, ... }
Otherwise, a language Li in the set consists of all strings of a s
shorter than, and including, a repeated i times. Learning a lan-
guage in this set just means converging to the i that is the index
for the language Li.
It is easy to see that this set of languages is not learnable from
positive-only evidence. Suppose that the longest string that the
learner has seen to date is of length n . Nothing about the text will
allow the learner to distinguish between L0 and Ln, and so the
learner will be incapable of converging. Tus, this class of lan-
guages is not learnable.
Golds result might seem to spell disaster for the learnability
project. Te derivational machinery needed for the languages
that Gold used to prove his theorem is much simpler than what
would be required for the set of natural languages, yet Golds set
is unlearnable. Gold seems to have shown that the natural lan-
guages are not learnable in the sense outlined here.
We neednt fret for too long over this particular bugbear since
another way of thinking about Golds result is near at hand. In
particular, the set of learnable languages does not contain the set
that Gold constructed for his proof. Some have found it tempt-
ing to massage Golds result into an argument that the learner
is equipped with prior information ( innate knowledge) about
the set of natural languages, although this goes beyond the actual
content of the result.
Tere are a number of interesting responses to Golds result.
Complexity bounds can be placed on the grammars that the
learner can consider. M. Kanazawa ( 1998 ) has shown that
restricted sets of categorial grammars are string learn-
able learnable using the kind of text presentation that we have
considered. Golds result entails that the entire class of categorial
grammars cannot be learned using this kind of evidence. Note
that Kanazawas result does not confict with Golds theorem
since it holds only of a particular subset and not the entire class,
the latter case being what Golds theorem excludes .
Alternatively, learners might receive more evidence about the
target language than is present in a positive-only text. K. Wexler
and P. Culicover ( 1980 ) developed a proof of the learnability of
the set of rules in the transformational component of a 1970s style
transformational grammar . In their system, the learner is
presented with pairs consisting of the surface syntactic string
along with a base structure. Tis base structure is akin to the level
of deep structure (see underlying structure and sur-
face structure ), where the syntactic representation is a kind
of mentalese the language of thought, and would be invariant
across languages (the Universal Base Hypothesis). Te learner
is presented occasionally with both a grammatical sentence and
its meaning. Te proof shows both that the transformation rule
component could be learned and that a complexity bound could
be placed on the input evidence that the learner needed in order
to converge (their Degree 2 Learnability result).
S. Pinker ( 1984 ) considered the cases whereby the learner has
access to the string and a representation of its semantic content.
Tis process, called semantic bootstrapping, uses a set of heu-
ristic rules to link semantic categories to syntactic categories.
LEARNABILITY
generative grammar shifted the foundations of theoretical
linguistics away from discovery procedures the automatic con-
struction of an optimal grammar to the problem of language
learnability, the question of how a natural language could be
learned in principle. Tis statement of the problem is so vague
as to be useless; so let us break the question into subparts and
consider them in turn .
Te study of language learning might proceed from the obser-
vation and investigation of actual human children engaged in the
process of learning their mother tongue or tongues. However fas-
cinating and useful this approach is, it is fraught with a number
of dif culties. Real children are engaged in a number of diferent
tasks and are changing along a number of diferent dimensions
while in the process of learning their frst language. Te develop-
mental psycholinguist must, then, be careful of all these diferent
factors.
Te investigation of learnability seeks to circumvent these dif-
fculties by considering language learning as a problem in com-
putational logic. Te researcher in language learnability seeks to
construct an explicit algorithm that will produce a grammar for
a target language after fnite exposure to evidence from that lan-
guage. Such a researcher takes the rarefed view of language as
a set of strings, corresponding to the grammatical sentences
of that language. He or she supposes that the learner is actually
an algorithm that takes as input a text, an infnite sequence of
sentences. Te text is constructed by drawing strings from the
language and presenting them, one at a time, to the learner. In
this case, the learner is presented with positive only evidence;
he/she is given information about sentences that are in the lan-
guage but no information about sentences that are outside the
language. An alternative learning setting would be to allow the
learner to be tutored by giving him/her strings that are marked
for grammaticality . Such information demonstrably simpli-
fes the learning task, but in real learning, the child is unlikely
to receive systematic evidence about grammaticality. As a result,
learnability research has generally proceeded from the assump-
tion of positive-only evidence.
After each example is presented to the learner, that learner
makes a guess about the grammar of the target language. A
learner is said to converge to a grammar for a language just in
case the learner hypothesizes that grammar after fnite exposure
to the text and never alters the hypothesis after that. If the gram-
mar is correct, then the learner is said to have learned the target
language; truly, learning a language means that the learner has
hit the correct grammar and never changes his/her mind after
that. We only require, at this point, that the grammar generate
the correct set of sentences; we have not said anything about
how the grammar does so, and so we place no constraints on the
structural descriptions assigned to sentences.
A language is learnable if there exists a learner who, upon
fnite exposure to the language, learns the language in the afore-
mentioned sense. A set of languages is learnable if the leaner can
learn every language in the set .
We can turn to an early learnable result from E. M. Gold
( 1967 ; see also Osherson, Stob, and Weinstein 1986 ). Imagine a
set of languages defned as follows: We defne a language, call it
Learnability
428
only then would the learner set the parameter to the exemplifed
value.
Charles Yang (2002) has imported a number of mathematical
tools from population biology to develop a sophisticated model
of parameter setting that is clearly inspired by evolutionary the-
ory. Equally interesting work has been done on the learnability of
optimality theory (Tesar and Smolensky 2000 ) using tech-
niques drawn from statistical machine learning . Although no full
proof of the learnability of parametric approaches yet exists, the sta-
tistical approaches as well as work in conventional machine learn-
ing (see Manning and Schutze 1999 ) promise to yield new insights
in language learning, language variation, and language change .
Robin Clark
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Clark , R. 1992 . Te selection of syntactic knowledge . Language
Acquisition 2 : 83 149.
Clark , R. , and I. Roberts . 1993 . A computational model of language
learnability and language change . Linguistic Inquiry 24 : 299 345.
Dresher , E. , and J. Kaye . 1990 . A computational learning model for met-
rical phonology . Cognition 34 : 137 95.
Florencio , C. C. 2003 . Learning categorial grammars. Ph.D. diss.,
Universiteit Utrecht.
Frank , R. , and S. Kapur . 1996 . On the use of triggers in parameter set-
ting . Linguistic Inquiry 27 : 623 60.
Gibson , E. , and K. Wexler . 1994 . Triggers . Linguistic Inquiry 25 : 407 54.
Gold , E. M. 1967 . Language identifcation in the limit . Information and
Control 10 : 447 74.
Kanazawa , M. 1998 . Learnable Classes of Categorial Grammars . Stanford,
CA : CSLI Publications .
Kapur , S. 1991 . Computational learning of languages. Ph.D. Tesis,
Cornell University.
Manning , C. D. , and H. Schutze . 1999 . Foundations of Statistical Natural
Language Processing . Cambridge, MA : MIT Press .
Niyogi , P. 2006 . Te Computational Nature of Language Learning and
Evolution . Cambridge, MA : MIT Press .
Osherson , D. , M. Stob , and S. Weinstein . 1986 . Systems Tat Learn: An
Introduction to Learning Teory for Cognitive and Computer Scientists .
Cambridge, MA : MIT Press .
Pinker , S. 1984 . Language Learnability and Language Development .
Cambridge : Harvard University Press .
Tesar , B. , and P. Smolensky . 2000 . Learnability in Optimality Teory .
Cambridge, MA : MIT Press .
Wexler , K. , and P. Culicover . 1980 . Formal Principles of Language
Acquisition . Cambridge, MA : MIT Press .
Yang , C. D. 2002 . Knowledge and Learning in Natural Language .
Oxford : Oxford University Press .
LEFT HEMISPHERE LANGUAGE PROCESSING
Some History
Te left hemisphere (LH) has been considered to be the primary
locus of language-specifc processing for centuries. We now
know that the right hemisphere (RH) has considerable lan-
guage abilities and should no longer be considered the minor
hemisphere. Recent research has demonstrated that both the left
and right hemispheres contribute to varying aspects of language
processing (Beeman and Chiarello 1997 ). Even so, historical and
current work still regards the left hemisphere as having a primary
and signifcant role in language processing.
Costa Florencio ( 2003 ) has shown that the full range of categorial
grammars can be learned if the learner is presented with unla-
beled structures .
Another response to Golds result is to impose diferent types
of constraints on the learners hypothesis space. For example,
the idea that universal grammar consists of a set of invariant
principles whose expression is regulated by a fnite set of param-
eters has played a seminal role in linguistic theory over the past
quarter of a century (see principles and parameters the-
ory and language acquisition ). Te learner is taken as
being faced with the fnite task of discovering the correct value
of each parameter, where each parameter can take on one of a
fnite set of parameters, given a text consisting of simple gram-
matical sentences .
Te parametric approach usually assumes that the learner
is given positive-only input. After a sentence is presented to
the learner, it produces a hypothesis, possibly by changing the
value(s) of one or more parameters. It is obvious that even a rela-
tively small set of parameters could produce an enormous space
of languages. Te space of languages might, for example, con-
tain local maxima . A local maximum might look correct to the
learner since it would always yield a structural analysis for any
input sentence, but it would systematically give the sentence an
incorrect structural analysis, for example. If this happened, then
the set of languages defned by that parameter space would not
be learnable relative to a simple learning device with positive-
only evidence.
In response to this problem, a number of diferent algorithms
were proposed. R. Clark ( 1992 ) proposed using a kind of artifcial
evolution to converge to the target. A population of grammars
would be exposed to the input text, with the best performers
allowed to combine and produce ofspring that had inherited
properties from the parent grammars. Tis approach uses the
parallelism implicit in a population to avoid the problem of local
maxima. Tis approach to learning falls into the class of probably
approximately correct (or PAC) learning; in this framework, the
learner is guaranteed to converge within a margin of error, where
the margin of error can be made arbitrarily small, but not zero.
Clark and I. Roberts (1993) extended this work to try to account
for language change . P. Niyogi ( 2006 ) has developed a more
sophisticated computational approach to this problem (see also
Yang 2002 ) .
E. Gibson and Wexler ( 1994 ) tried to develop a learner that
used an algorithm that tests to see if resetting a parameter to
a new value actual improves the learners performance on the
input example. In order to avoid local maxima, they proposed
ordering the parameters according to a maturational sequence.
Readers should consult Frank and Kapur ( 1996 ) for an extensive
critique. S. Kapur ( 1991 ) developed a learning algorithm that
avoids local maxima by using a statistical model of indirect nega-
tive evidence. Tis algorithm is, once again, clearly within the
PAC learners.
Others have proposed using triggering evidence to set param-
eters (Dresher and Kaye 1990 ). On this model, each value of a
parameter would be associated with the abstract description of
a piece of triggering evidence that would cause the parameter
to be set to that value. Te learner would scan the input text,
searching for examples that matched the description of a trigger;
Learnability Left Hemisphere Language Processing
429
Tere is evidence that this exhaustive access is controlled imme-
diately by the LH (Burgess and Simpson 1988 ). Soon after, a
lexical choice is made on the basis of frequency and context.
Evidence from aphasia also supports the role of the left anterior
frontal cortex in lexical access. For example, individuals with
Brocas aphasia appear to show a slow rise time of the initial
activation of multiple meanings, while those with Wernickes
aphasia evince normal patterns (Prather et al. 1991 ). Other acti-
vation accounts suggest that individuals with Brocas aphasia
(with damage to LIFG) underactivate lexical forms, while those
with Wernickes aphasia (with damage to STG) overactivate
(Blumstein and Milberg 2000 ) .
Te role of LIFG in real-time processing also appears to extend
to the comprehension of sentences that contain displaced argu-
ments or those with fller-gap dependencies, for example, in
object relative (OR) constructions (e.g., Te audience liked the
wrestler that the priest condemned *____ for foul language) where
a direct object argument or fller (e.g., wrestler) has been dis-
placed from its canonical, post-verb position or gap (noted by *).
Individuals with Brocas aphasia do not activate the fller at the
gap in real time, unlike what is observed for neurologically intact
individuals and those with Wernickes aphasia or RH lesions
(Swinney et al. 1996 ). Tus, a real-time processing defcit may
underlie the inability for Brocas individuals to ultimately com-
prehend these constructions when they are probed with simple
sentence-picture matching tasks or grammaticality judgments .
Variability
Much of the work detailing the role of the LH in language pro-
cessing has been based on descriptions of neuroanatomy con-
ducted in the later third of the nineteenth and early twentieth
centuries. K. Brodmann ( 1909 ) suggested that the most func-
tionally relevant parcellation of the brain is by cytoarchitecton-
ics (cellular composition), but the map of Brodmanns Areas
is based on manually drawn borders of a single brain. K. Amunts
and colleagues (1999) examined 10 postmortem brains, and the
borders for each brain were automatically drawn and super-
imposed on a template to produce a group cytoarchitectonic
map. Large intersubject variability was uncovered, perhaps par-
tially explaining why so much variability exists in the mapping
between behavior and anatomy from both lesion and functional
imaging studies.
Another possible contributor to intersubject variability is the
assumption of dead tissue only in and around the lesion. For
many years, investigators have assumed that structural lesions
(and the ischemic penumbra surrounding the lesion) were the
primary loci contributing to language defcits. With the advent
of more refned neuroimaging technology, such as perfusion
weighted and difusion tensor imaging, researchers have been
investigating areas of the brain that are found to be structurally
intact yet not receiving an optimal supply of blood fow. Tese
hypoperfused regions give way to functional lesions inside seem-
ingly intact neural tissue (Hillis 2007 ; Love et al. 2002 ) .
The Role of Functional Neuroimaging in the Investigation
of Left Hemisphere Language
Lesion studies alone must be interpreted with caution as these
can only provide information regarding a specifc (i.e., the
One of the frst functional accounts of brain-language rela-
tions geared to the LH, the Wernicke-Lichteim model, separated
language into activities such as listening, reading , writing, and
speaking. Based on classic lesion localization eforts, these activ-
ities were thought to be localized in diferent LH brain regions.
Geschwinds (1965) model proposed that brocas area a
region located in the left inferior frontal gyrus (LIFG) at the foot
of the motor strip near regions controlling mouth and tongue
movements was the seat of speech production , while
wernickes area in the posterior superior temporal regions
adjacent to the primary auditory cortex was the seat of audi-
tory comprehension. While this simple model was appealing, it
has been clear since at least the 1970s that this view of language
is likely inaccurate . For example, individuals diagnosed with
Brocas aphasia have auditory comprehension defcits that are
exposed on simple experimental probing and have production
defcits that go well beyond those described by fuency measures
(Zurif and Caramazza 1976 ; Friedmann 2006 ).
Beginning in the 1970s, eforts were made to characterize lan-
guage in the LH by reference to linguistic levels of analysis along
the lines of syntax, semantics , and phonology . Te LIFG
has been suggested to be critical for syntax, while the temporal
lobe has been suggested to be important for the normal func-
tioning of word -level semantics. Moreover, the posterior supe-
rior temporal gyrus (pSTG) has been suggested to play a critical
role in phonology. Even within these linguistic divisions, eforts
have been made to discover exquisitely detailed neurological
instantiations. Te trace deletion hypothesis (Grodzinsky 2006 ),
for example, has taken a minimalist position on syntax-brain rela-
tions (see syntax, neurobiology of ) and has suggested that
only sentence constructions that are derived from the displace-
ment of an argument (e.g., a noun phrase) and that yield a trace
(see also movement ) rely on an intact Brocas area, and that
other hypothesized aspects of syntax rely on more widely distrib-
uted anatomical regions . Other accounts of the relation between
syntax and the brain suggest that only those constructions that
are defned as complex rely on an intact LIFG. Alternative theories
suggest that syntax, broadly defned, requires a neuroanatomical
language network consisting of Brocas region as well as STG, the
middle temporal gyrus (MTG), and the white matter fber tracks
(arcuate fasciculus) connecting these regions .
A Real-Time Perspective
Te most current approach to brain-language relations as we
near the end of the frst decade of the twenty-frst century is for-
mulated in terms of processing metaphors such as activation
and maintenance that require a real-time analysis . It has been
suggested that within the LH, Brocas area is required for fast-
acting, relatively automatic and refexive processing routines;
more frontal areas are critical for executive functions underly-
ing language processing, including selecting among alternatives;
left posterior temporal areas seem important for activating and
maintaining argument structure, an aspect of lexical-semantics
or conceptual structure.
Take, for example, the activation of meaning-related word
forms during sentence comprehension. It has been argued that
unimpaired individuals initially activate multiple meanings of
ambiguous words, regardless of the context of the sentence.
Left Hemisphere Language Processing
430
that are observed, may help illuminate brain-behavior map-
ping in both the left and the right hemispheres (e.g., Tompson
and Shapiro 2007 ) .
Tracy Love and Lewis P. Shapiro
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Amunts , K. , Axel Schleicher , U . Brgel , Hartmut Mohlberg , Harry Uylings ,
and Karl Zilles . 1999 . Brocas region revisited: Cytoarchitecture
and intersubject variability . Journal of Comparative Neurology 412
(August): 319 41.
Beeman , M. , and C. Chiarello , eds. 1997 . Right Hemisphere Language
Comprehension: Perspectives from Cognitive Neuroscience . Hillsdale,
NJ : Lawrence Erlbaum .
Blumstein , S. , and W. Milberg . 2000 . Comprehension in Brocas and
Wernickes aphasia: Singular impairment. In Language and the
Brain , ed. Y. Grodzinsky , L. P. Shapiro and D. Swinney , 16783. San
Diego : Academic Press .
Brodmann , K. 1909 . Vergleichende Lokalisationslehre der Grohirnrinde
in ihren Prinzipien dargestellt auf Grund des Zellenbaues . Leipzig :
Barth JA .
Burgess , C. , and G. Simpson . 1988 . Cerebral hemispheric mechanisms
in the retrieval of ambiguous word meanings . Brain and Language 33
(March): 86 103.
Caplan , D. , and G. Waters . 1999 . Verbal working memory capacity and
language comprehension . Behavioral and Brain Sciences 22.1 : 77 94.
Friederici , A. 2002 . Towards a neural basis of auditory sentence
processing . TRENDS in Cognitive Sciences 6.2 : 78 84.
Friedmann , N. 2006 . Speech production in Brocas agrammatic
aphasia: Syntactic tree pruning. In Brocas Region , ed. Y. Grodzinsky
and K, Amunts , 6382. New York : Oxford University Press .
Geschwind, N. 1965. Disconnection syndromes in animals and man.
Brain 88 : 585644.
Grodzinsky , Y. 2006 . A blueprint for a brain map of syntax. In Brocas
Region , ed. Y. Grodzinsky and K. Amunts , 83107. New York : Oxford
University Press .
Heiss , W. , J. Kessler , A. Tiel , et al. 1999 . Diferential capacity of left
and right hemispheric areas for compensation of poststroke aphasia .
Annals of Neurology 45 .4: 430 8.
Hickok , G. , T. Love-Gefen , and E. Klima . 2002 . Left temporal lobe
supports sign language comprehension . Brain and Language
82 .2: 167 78.
Hillis , A. 2007 . Magnetic resonance perfusion imaging in the study of
language . Brain and Language 102.2 : 165 75.
Humphries , C. , T. Love , D. Swinney , and G. Hickok . 2005 . Response
of anterior temporal cortex to syntactic and prosodic manipu-
lations during sentence processing . Human Brain Mapping
26.2 : 128 38.
Kinsbourne , M. 1971 . Te minor cerebral hemisphere as a source of
aphasic speech . Archives of Neurology 25.4 : 302 6.
Love , T. , D. Swinney , E. Wong , and R. Buxton . 2002 . Perfusion imaging
and stroke: A more sensitive measure of the brain bases of cognitive
defcits . Aphasiology 16.9 : 873 83.
Prather , P. , L. Shapiro , E. Zurif , and D. Swinney . 1991 . Real-time exami-
nations of lexical processing in aphasics . Journal of Psycholinguistic
Research 20.4 : 271 81.
Stowe , L. , A. Paans , A. Wijers , F ., et al. 1999 . Sentence comprehension
and word repetition: A positron emission tomography investigation .
Psychophysiology 36.6 : 786 801.
Stromswold , K. , D. Caplan , N. Alpert , and S. Rauch . 1996 . Localization
of syntactic comprehension by positron emission tomography . Brain
and Language 52.3 : 452 73.
damaged) neural regions necessity to perform a particular
language task. Functional neuroimaging patterns, on the other
hand, describe the level of recruitment of specifc area(s), not
the necessity of only the lesioned area for the process itself (Hillis
2007 ). It is the fusion of these and other methodologies that best
aides in the understanding and modeling of the brain basis of,
and networks involved in, language processing .
Neuroimaging research has demonstrated that the LH is par-
ticularly well suited for language processing, regardless of the
modality of language input (auditory or visual, as is found in
sign languages ; Hickok, Love-Gefen, and Klima 2002 ). Te
literature has demonstrated a LH bias for the neural circuitry
involved in the processing of complex versus simple sentence
constructions. More specifcally, it has been argued that there
exists active recruitment of BA 44 and BA 45 (pars opercularis
[Stromswold et al. 1996 ] and pars triangularis [Caplan and
Waters 1999 ]) of the LH during the parsing of complex sentence
constructions (e.g., the fller-gap constructions described ear-
lier). Yet other reports examining sentence comprehension have
found anterior temporal cortex activation, including STG and
MTG (e.g., Humphries et al. 2005 ; Stowe et al. 1999 ). It is quite
likely that the discrepancies found in the imaging literature are
due to varying methods of presentation, and difering behavioral
requirements of the participants, as well as experimental design
issues and analysis procedures .
The Integration of Language Processing in the Left
Hemisphere
Work from multiple tasks and methodological techniques have
been integrated to form the basis of neurocognitive models of
language processing. Tese models capture the choreographed
workings of neural regions during language processing. One
such model posited by A. Friederici ( 2002 ) argues for an LH
biased temporofrontal network. According to this model, iden-
tifcation of a word into a grammatical category (e.g., noun, verb,
determiner, etc.) begins at about 200 milliseconds after the word
is encountered in the speech stream, localized in the regions sur-
rounding the anterior superior temporal cortex. At this temporal
point, such grammatical categories are placed into a hierarchical
syntactic form, which relies on the regions surrounding Brocas
area. At about 300500 milliseconds after a word is encountered,
the lexical entry is accessed, which allows for subsequent syntac-
tic integration via Brocas area and semantic integration via
temporal lobe regions .
Translational Research
Finally, the investigation of language processing in the LH has
yielded a growing enterprise devoted to mapping recovery of
function in aphasia following brain damage. Converging evi-
dence from clinical studies along with functional neuroimag-
ing studies have demonstrated that, depending on individual
factors such as size and extent of lesion, premorbid handed-
ness, and so on, recovery of language function may include the
undamaged regions of the LH language processing network or
homologous RH regions (Heiss et al. 1999 ; Kinsbourne 1971 ).
Treatment of language disorders for individuals with brain
damage, and the subsequent behavioral and neural changes
Left Hemisphere Language Processing
431
holding to follow deductively from some set of higher or more
general principles. Te answer to the present problem then could
be deduced from those principles. Te deductive process incor-
porates an efort to ensure that the application of a precedent in
the case at hand cohere with the entire set of abstract principles
that, taken together, constitute the common law. (Interpretation
of the codes in civil law systems is said to follow the same model,
except that the general principles are found in the codes provi-
sions, rather than extracted by interpretation from prior decided
cases.)
Te formalist approach to precedent was subjected to with-
ering criticism in the twentieth century, primarily by the rule
skeptics associated with American legal realism (Rumble 1968 ).
Tey argued, and to much of the legal community demonstrated,
that the purported deductions never satisfed minimal standards
of deductive reasoning. Te legal realists argued that what courts
actually did was to interpret precedents with an eye to the poli-
cies advanced by the rules articulated by the courts: Te present
problem would be resolved by determining how the policies
embodied in the rules articulated in the precedents would best be
advanced. In the nineteenth century, for example, courts barred
an employee injured by the negligence of another employee
from recovering damages from their joint employer, in part
because the injured employee was said to be in a good position
to notice whether the other employee was a careful worker. Later
courts had to decide whether that policy was applicable where
the negligent worker labored in a diferent department, or was
the injured workers supervisor. Critics of the strongest versions
of legal realism wondered why the policy-oriented approach
should be described as involving interpretation at all. Policy-
oriented decisions, they argued, were entirely forward looking,
and the precedents did no more than provide a convenient heu-
ristic to guide thinking about the present problem .
Te temporal dimension of legal interpretation is apparent
on the surface when courts interpret prior decisions. Te insti-
tutional dimension is revealed when we ask, particularly of the
policy-oriented interpreter, Why should a court today give any
weight to the rules articulated by courts in the past? Answers
vary, but most combine a Burkean ideal that judges today should
not be overly confdent that they know better than their prede-
cessors what good policies are, with a pragmatic sense that some
degree of reliance on prior decisions conserves judicial efort .
Statutory Interpretation
Questions of statutory interpretation as such arise only when
someone an enforcement of cial or a judge, for example has
some question about what a statutes terms mean. Where statu-
tory terms are thought to be unambiguous, of cials simply apply
the statutes, an operation that to them seems preinterpretive
(see also philology and hermeneutics ). Application rather
than interpretation is likely to be more common soon after a
statutes adoption, because people will generally be familiar with
what the statutes enactors were trying to do unless, as happens
with some frequency, the adopters deliberately left specifc pro-
visions in the new statute unclear.
Tere are three prominent approaches to statutory interpre-
tation in the United States, with parallels in other legal systems.
(For an overview of the contemporary discussion in the United
Swinney , D. , E. Zurif , P. Prather , and T. Love . 1996 . Neurological dis-
tribution of processing operations underlying language comprehen-
sion . Journal of Cognitive Neuroscience 8.2 : 174 84.
Tompson , C. , and L. P. Shapiro . 2007 . Complexity in treatment of
sentence defcits in aphasia . American Journal of Speech-Language
Pathology 16 : 30 42.
Zurif , E. , and A. Caramazza . 1976 . Psycholinguistic structures in
aphasia: Studies in syntax and semantics. In Studies in Neurolinguistics ,
ed. N. Avakian-Whitaker and H. Whitaker , 26092. New York : Academic
Press .
LEGAL INTERPRETATION
Te judicial practice of legal interpretation provides the model
for other versions of legal interpretation, such as the interpreta-
tion of law by practicing lawyers. Interpretation of three types
of texts prior judicial decisions (which serve as precedents for
a present interpretive exercise), statutes, and constitutions
provides the model for interpretation of other texts, such as
regulations and treaties. It has a temporal and an institutional
dimension. Decision makers interpret texts written earlier,
sometimes substantially earlier, and produced by institutions
diferent from the ones engaged in the interpretive enterprise
(if only because of changes in personnel). Tese characteristics
generate some problems common to all three forms of legal
interpretation.
Case Law and the Interpretation of Precedents
Prior judicial decisions play an important part in every system
of legal interpretation. Civil law systems, such as those in France
and Germany, are committed to a legal ideology in which judges
revert to statutory texts directly, without reference to any prior
judicial decisions. Judicial decisions in such systems rarely refer
to prior decisions, but even in such systems, precedent plays an
important role before courts issue their opinions (Lasser 2004 ).
Te common law system of unwritten law, the foundation of
law in Great Britain and the United States, was developed by the
courts themselves to regulate large portions of the law of prop-
erty, contracts, and torts (accidents, among other topics). Te
common law is unwritten only in the sense that the texts on
which it is based are prior judicial decisions, rather than legis-
lative enactments .
Interpreting a judicial decision in order to apply it to a new
problem involves several analytic steps. Typically, a decision will
describe a cases facts and articulate several legal rules that the
court says lead it to its conclusion. Te interpreter later must dis-
tinguish the decisions holding from any obiter dicta found in the
decision. On standard accounts, the holding is the rule or rules
necessary to support the conclusion, dicta any rule or rules that
could be eliminated from the courts discussion without altering
its conclusion (Marshall 1997 ). Later courts do not always disre-
gard dicta, however, sometimes fnding that they provide useful,
though not binding, guidance.
Te prior decisions holding, once identifed, must be applied
to the new problem. Again typically, that problem will difer in
some respects from that presented by the precedent. Courts apply
the precedents rule in two ways. For much of the nineteenth
century, and to some extent today, courts applied precedents
formalistically . Tey took the rule supporting the precedents
Legal Interpretation
432
singular when discussing the institutions that adopt statutes, but
it cannot be avoided so easily .
Te fnal prominent approach, usually called purposivism ,
implicitly shifts the focus from the enacting legislature to the
interpreter. In a classic formulation, the purposivist assumes
that the legislature was composed of reasonable people seeking
to pursue reasonable purposes reasonably. Ambiguous statu-
tory terms are to be interpreted so that the goals imputed to the
legislature are most likely to be achieved. Purposivism avoids
most of the problems associated with intentionalism, because,
although its proponents ordinarily refer to the legislatures
purposes, they are not truly concerned with a real institution
stafed by real people. Rather, purposivists construct an ideal-
ized legislature to which they impute purposes that they then
seek to implement. Yet purposivism typically lacks an account
of whether the interpreter should posit abstract or more con-
crete purposes.
Purposivism makes the institutional dimension of legal inter-
pretation clear. It allocates efective decision-making authority to
courts, at least once the legislature does something that licenses
the courts to engage in the interpretive enterprise. Its proponents
believe that purposivism contributes to the good functioning of
the government overall, as courts and legislatures collaborate in
accomplishing good for the society. Critics respond with some
skepticism about the very idea of the public good as something
independent of the choices made by legislatures, and with the
observation that what the purposivists are doing cannot fairly be
described as interpretation. Rather, they suggest, the judges are
reading into the law their own policy preferences and then attrib-
uting those purposes to the statute.
Statutory interpretation also involves the use of canons of
statutory construction, which might be thought as well to con-
strain the judges power to interpret statutes merely to advance
their policy preferences. One example is the rule of lenity,
according to which criminal statutes should be construed where
fairly possible to limit the scope of criminal liability. Another
example, in legal systems with some form of judicial review for
constitutionality, is the canon that statutes should be construed,
again where fairly possible, to make them consistent with the
constitution or basic human rights. Scholars divide canons of
construction into two groups, substantive and legislative-intent
canons. Substantive canons embody policies that courts seek
to pursue independent of what legislatures actually sought
to accomplish in enacting particular statutes. Te rule of len-
ity and the rule that statutes should be construed to limit their
impact on background law are examples. Legislative-intent
canons are rebuttable presumptions about what legislatures
seek to accomplish in enacting particular statutes. Te canon
dealing with avoiding constitutional questions can be justifed
on the ground that courts should not assume that legislators
sought to enact unconstitutional statutes. Karl Llewellyn ofered
a classic critique of canons of interpretation, suggesting that
for each thrust built into one canon, there was a parry from
another equally well-established canon of statutory interpreta-
tion (Llewellyn 1950 ). So, for example, the canon Every word
and clause must be given efect was parried by the canon If
inadvertently inserted or if repugnant to the rest of the statute,
they may be rejected as surplusage.
States, see Vermeule 2006 .) Te textualist approach interprets a
statutes terms by asking what the words would mean to an ordi-
nary reader (usually, a reader at the time the statute was enacted)
who is reasonably well informed about the meaning of the tech-
nical terms and about the entire statutory environment within
which the contested term is located.
Textualism is a rather bare-bones interpretive approach,
which to its critics requires the interpreter to ignore real and
accurate information about what a statute is designed to do.
Proponents of textualism often claim more clarity for the out-
comes they reach than there actually is. Where ambiguity per-
sists after considering the sources to which textualists limit
themselves, some other basis is needed for resolving the contro-
versy. Te most prominent candidate emerges from the assump-
tion that ambiguous legislation should not be taken to disturb
the status quo. Tis assumption is sometimes expressed as a
canon of construction that ambiguous statutes should not be
construed to change the common law (that is, the background
rules that would apply if the legislature took no action). An alter-
native defense of textualism is comparative: that it reaches better
outcomes overall than alternative approaches that ask interpret-
ers to assess information that they are not well equipped to han-
dle, even though in particular cases the use of one or the other
approaches might produce a better result than textualism .
Te intentionalist approach shifts the focus from the reason-
able reader to the enacting legislature. It asks what the legisla-
ture intended to accomplish by enacting the statute. In its least
controversial version, the intentionalist approach directs the
interpreters attention to the problem the statute was designed
to solve, producing an interpretation that, in the judges view,
solves the problem as well as possible within the bounds set by
the statutes words as reasonably understood. Intentionalists in
the United States, more than in the United Kingdom, are will-
ing to consult documents produced as the statute proceeded
through the enactment process (the statutes legislative his-
tory), such as reports by the committees that considered the
legislation and statements by the statutes supporters and oppo-
nents, to determine what the legislature meant by the terms it
used.
Textualists criticize these more expansive versions of inten-
tionalism. Most narrowly, they note that materials drawn from
the legislative history are readily deployed strategically by advo-
cates, who present only the materials that support the interpreta-
tion that will yield the result they favor, and selectively by judges,
who refer only to those parts of the legislative history that favor
that result the judges prefer for reasons independent of the inter-
pretive enterprise. Critics also observe that referring to legisla-
tive history gives some degree of authority to committees and
individual members, whereas only the entire legislature has any
authority to enact law. Finally, critics question the coherence of
invoking intentionalist terms with respect to multimember legis-
lative bodies. Some legislators might have favored the adoption
of a statutory provision because they thought that it solved an
important public policy problem, others because their constit-
uents favored it, still others because important contributors to
their campaigns did so. How can these varying states of mind be
aggregated into an intent of the legislature? Continental legal
theorists elide this question by referring to the legislator in the
Legal Interpretation
433
One can rely on concrete intentions to rule out the possibility
that a practice they understood to be constitutionally permis-
sible would later be found to be constitutionally impermissible,
but even then the reliance on concrete intentions or understand-
ings requires a defense that goes outside the terms set by origi-
nalism itself. An alternative, similar to that ofered by canons
of statutory interpretation, is to hold a practice constitutionally
permissible unless it is clearly precluded by the constitution
as originally understood. Te justifcation for this alternative is
institutional: Te constitution of a liberal democracy taken as a
whole should be understood to commit decision-making author-
ity to democratically elected legislatures, unless the constitution
clearly takes that authority away from the legislatures and gives
it to the courts .
Te second family has no standard name, but probably can be
best described as including varieties of perfectionism . According
to this group of approaches, general and abstract constitutional
provisions should be interpreted in accordance with some over-
arching principles of good government and individual liberty.
Tese principles can be relatively modest, as in a commitment
to democratic self-governance (Ely 1980 ), or more robust, as in
a commitment to justice broadly understood (Dworkin 1996 ).
Germanys constitutional court fnds the perfectionist approach
to interpretation embodied in its constitutions commitment
to what it calls a basic order of values. (Te Muslim con-
cept of ijtiha
-
d might be thought to have a similar underlying
structure.)
Perfectionist approaches to constitutional interpretation
resemble purposivist approaches to statutory interpretation.
Interpreters, it appears, are to rely on their own best understand-
ing of what the basic order of values is. Tis, some believe, is
inconsistent with democratic self-governance because it allows
judges to substitute their judgments about what justice or equal-
ity requires for the judgments made by elected representatives.
Critics suggest, in this setting as well, that what perfectionists
do cannot be called interpretation. Te sting of that observation
might be reduced by responding that a constitutions text as such
has no authority anyway; only the long-standing practices that
people have come to accept have authority, and perfectionist
practices have been widely accepted for many years.
Another response, suggested by Stanley Fish ( 1994 ), is that
perfectionist interpretation is not as unconstrained as its critics
think. Judges are part of an interpretive community whose
shared understandings place signifcant limits on what even the
most willful judge will take as a responsible interpretation of a
constitutional provision. One important version of this view
describes a common law constitution, in which what judges
interpret is not primarily the constitution as written but the prior
decisions interpreting the constitution (Strauss 1996 ). Here,
constitutional interpretation reproduces common law interpre-
tation. It is worth noting that these two defenses of perfectionist
interpretation do not preclude interpretations that are norma-
tively unattractive, if the people or the interpretive community
settle on unattractive practices .
Consistent with the idea that judges are members of inter-
pretive communities, judges in diferent nations take diferent
approaches to interpreting their constitutions (Goldsworthy
2006 ). Te practice in the United States is quite eclectic, with
Canons of interpretation can ft into each of the interpretive
approaches. For the textualist, the canons are part of the general
background the ordinary reader is assumed to know as he or she
reads a contested statutory text. Te intentionalist can defend the
substantive canons on the ground that legislatures typically do
not intend to infringe on the policy goals embodied in substan-
tive canons. Te purposivist has an easy time with substantive
canons, which directly embody judgments about good policy,
and can treat the other canons as similarly refecting good policy
judgments, rather than imputations of legislative intent .
Constitutional Interpretation
Questions of statutory interpretation typically involve specifc
and detailed provisions of complex statutes. Constitutional
interpretation, in contrast, typically involves the application of
general and abstract constitutional terms, such as freedom of
speech, due process of law, and equal protection of the laws
(to take examples from the U.S. Constitution) to specifc prob-
lems. Here, too, there are two primary families of approaches.
Te frst family includes varieties of originalism . One ver-
sion holds that constitutional provisions should be interpreted
to conform to the intent of the constitutions adopters. As with
intentionalism in statutory interpretation, original-intent
approaches run into many dif culties, such as the problem of
aggregating individuals intentions. In addition, when, as with
the U.S. Constitution, major provisions were adopted two centu-
ries earlier, the task of identifying what any particular individual
understood a provision to mean is extremely dif cult.
Finally, the abstract terms that constitutions use pose an addi-
tional problem: Should the provisions be interpreted according
to the abstract or the concrete understandings of their adopters?
Ronald Dworkin uses the term concept to describe the abstract
understanding, conception the concrete one (Dworkin, 1977).
Consider, for example, a constitutional provision dealing with
equality. Does that provision enact into fundamental law the
particular understandings the adopters had about equality, such
as the understanding that laws could treat men and women dif-
ferently while still providing equality, or does it enact equality
itself, that is, the best understanding an interpreter can devise at
the moment of interpretation?
Some proponents of originalism responded to these and
other problems with an original-intent approach by arguing that
constitutional provisions should be interpreted according to the
original public meaning of their terms. Te interpreter should
identify what the terms meant to a reasonable and well-informed
member of the public when the provisions were adopted. Tis
does not completely eliminate the evidentiary problems associ-
ated with original-intent approaches, because one is still search-
ing for what the terms meant to individuals, but it substantially
expands the range of relevant materials to include uses of the
terms in newspaper discourse and the like. Similarly, it shifts ter-
minology about abstract versus concrete intentions to references
to abstract versus concrete public meanings.
Te temporal problem is perhaps the most dif cult one fac-
ing originalist approaches. Te problem is that the adopters or
the general public years ago could have no intentions about, or
understanding of, what the terms they used meant, in connec-
tion with developments they did not and could not anticipate.
Legal Interpretation
434
partonomy, as well as with the meanings of verbs, adjectives,
and prepositions. Or, alternatively, they have argued that chil-
dren rely on many of the same pragmatic principles as adults
in making inferences in context about possible meanings. Under
this view, childrens initial inferences are limited only by what
they know and the words they have already acquired (Bloom
2000 ; Clark 1993 ).
How do children assign some meaning to an unfamiliar
word? Once adult and child are both attending to the same
object or action, for example, the child can infer that that object
or actionat their locus of joint attention in the physical con-
textis the adults intended referent. Tat is, the child draws on
both physical and conversational context in assigning meanings
to unfamiliar words, regardless of word class (Clark 2009).
Moreover, once an object or action has been labeled, the child
can often infer that subsequent utterances are also relevant to
the newly identifed referent. And these utterances, in turn, may
supply added information about properties (size, texture; man-
ner of motion), relations (role of the object as agent, location, or
entity-afected, say; see thematic roles ), function (common
uses, use on that occasion), and so on. Te inferences children
make about meanings are guided by adult usage (a way of fnd-
ing out the conventional way to designate each category) and by
the fact that new words must contrast in meaning with whatever
vocabulary they already know .
As children acquire more vocabulary, they build up seman-
tic domains words for food, clothing, cars, animals; types of
motion and location; and relations in space, for example and
they organize and reorganize each domain as they add new
members. Members of a domain are typically linked by semantic
relations like X is a kind of Y, Z is part of A, B is made of C,
or D is used for E. But not all relations hold in every domain.
Tey depend on the meanings of individual lexical items. Among
verbs, for instance, the relations typically include their argu-
ment roles. A locative verb like put , for instance, is accompanied
by three arguments an agent, an object, and a location, as in
Miranda put the cup on the shelf . But a verb of motion like run
requires only one argument role: the doer or agent, as in Robert
ran fast (Clark 2003).
Building up each semantic domain also involves identifying
words that co-occur (dogs bark , but horses neigh ), and common
collocations (compare disappearing ink and vanishing cream ). It
requires working out the semantic relations that link such terms
as tiger, predator , and mammal , or tree, aspen , and gingko , on
the one hand, and throw, toss, twirl or break, tear, and cut , on the
other. It also requires that children learn the terms for parts and
wholes ( thumb, fnger, hand ), for groups ( fock, pod, herd; crowd,
reunion, meeting ), for complex events ( circus, opera, play ), for
cycles (days of the week, months of the year), for relations ( in,
above, behind; before, after; if, because ), abstractions ( justice,
equality, goodness ), and much more.
Learning words is also the frst step in learning constructions.
Many constructions are linked initially to specifc verbs, and only
later extended to others that can take the same construction.
Children may learn want frst with a direct object, as in I want
that . Ten they start to use nouns in place of demonstrative that
( I want the ball, I want a spoon ), and only sometime later do they
start to use want with a to -complement, as in I want to go out .
judges using originalism and perfectionism relatively unsystem-
atically; the practice in Germany is more perfectionist, and that
in Australia is formalist .
Legal theorists have recurrently been attracted to the idea
that law, and legal interpretation, could become a science. Te
direct invocation of sciences, such as linguistics, psychology, and
more recently neuroscience, to understand legal interpretation
has produced relatively little enlightenment, to the point where
it seems more likely than not that whatever science of law there
might eventually be, it will not be a science on the model of the
physical or biological sciences .
Mark Tushnet
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Dworkin , Ronald. 1996 . Freedoms Law: Te Moral Reading of the
American Constitution . Cambridge : Harvard University Press .
. 1977 . Taking Rights Seriously . Cambridge : Harvard University
Press .
Ely , John Hart . 1980 . Democracy and Distrust: A Teory of Judicial Review .
Cambridge : Harvard University Press .
Fish , Stanley E. 1994 . Teres No Such Ting as Free Speech: and Its a
Cood Ting, Too . New York : Oxford University Press .
Goldsworthy , Jef . 2006 . Interpreting Constitutions: A Comparative Study .
New York : Oxford University Press .
Lasser , Mitchel de S.-O.-LE. 2004 . Judicial Deliberations: A Comparative
Analysis of Judicial Transparency and Legitimacy . New York : Oxford
University Press .
Llewellyn , Karl . 1950 . Remarks on the theory of appellate decision and
the rules or canons about how statutes are to be construed . Vanderbilt
Law Review 3 : 395 406.
Marshall , Geofrey . 1997 . What is binding in a precedent. In Interpreting
Precedents: A Comparative Study , ed. Neil MacCormick and Robert S.
Summers , 503 17. Brookfeld, VT. : Ashgate/Dartmouth .
Rumble , Wilfrid E. 1968 . American Legal Realism: Skepticism, Reform,
and the Judicial Process . Ithaca, NY. : Cornell University Press .
Strauss , David A. 1996 . Common law constitutional interpretation .
University of Chicago Law Review 63 : 877 935.
Vermeule , Adrian . 2006 . Judging Under Uncertainty: An Institutional
Teory of Legal Interpretation . Cambridge : Harvard University Press .
LEXICAL ACQUISITION
Children start to produce their frst recognizable words between
12 and 18 months of age, and typically understand more than
they produce. Tis asymmetry is a lifetime efect. Te forms of
their earliest words often depart radically from the adult versions
(consider ga for squirrel) and may be hard to understand. But
as their pronunciation becomes more skilled (see phonology,
acquisition of ), children add rapidly to the vocabulary at their
disposal. Tey add words for people, animals, everyday objects,
toys, food, and various activities and by age two generally pro-
duce between 200 and 800 distinct words.
Researchers have taken one of two main approaches to the
study of word acquisition in the last two decades: With the frst
approach, they have postulated built-in constraints that would
limit the hypotheses children entertain about possible mean-
ings, typically limited to noun meanings only (Markman 1989 ),
constraints that must later be overridden since they are incom-
patible with semantic relations, such as inclusion, overlap, and
Lexical Acquisition
435
A crucial underpinning idea is that any meaningful linguistic
element has associated with it diferent types of linguistic infor-
mation, for instance, information about prosodic structure, about
category and constituent structure (see phrase structure ),
about grammatical relations (also referred to as functions), and
about semantic structure (see semantics ) . It is furthermore
assumed that the organizational principles for these dimensions
may vary and that the formalisms used to represent the diferent
types of information should capture this variation. Te informa-
tion is represented in diferent dimensions, for instance c-struc-
ture (for category and constituent), f-structure (for functional),
a-structure (for argument structure), and i-structure (for infor-
mation structure). Each dimension operates with its own funda-
mental categories and principles. Te diferent dimensions are
related through mapping principles.
C-structure is represented in terms of a version of x-bar syn-
tax, employing both lexical categories, such as noun , adjective ,
verb , and preposition , and functional categories, such as comple-
mentizer , infection , and determiner . Quite a restrictive approach
to functional categories tends to be taken; they are used for ele-
ments expressing crucial functional features whose distribution
is limited to a certain position within a phrase. Hence, a category
such as I, assumed in some transformational theories to form a
part of every clause in every language, is motivated within LFG
for languages where an infected verbal element occupies a par-
ticular structural position. One example would be verb second
languages, where the fnite verb occurs in the second position in
the clause and, hence, fniteness can be associated with this posi-
tion. Tis can then be captured through a functional category IP,
headed by the fnite verb in I and the initial element placed in the
specifier position. In English, fnite auxiliaries have properties
that motivate the use of a functional category I (see Dalrymple
2001 , 534).
A principle of economy of expression applies to constituent
structure to yield trees that look rather unorthodox from a stan-
dard X-bar perspective. Tis principle states that any constituent
introduced by a rule is optional unless some separate principle
requires its presence. Tis can be illustrated by the analysis of
fnite auxiliary verbs in English. Te category I, to which these
verbs belong, is assumed to be introduced by a rule I I VP;
however, in sentences that do not contain a fnite auxiliary, the I
is not present, giving a tree such as that in (1).
Tey take even longer to add a subject to the complement, as in I
want Anna to come . Constructions often appear to be built up on
single lexical items on a one-by-one basis. Tis takes time (Clark
and Kelly 2006 ; Tomasello 2003 ).
Finally, adult usage plays a crucial role in acquisition.
Children track the frequencies of constructions in parental
speech and acquire frst those constructions that occur most
often. It is adult speakers who model word use, who ofer chil-
dren conventional terms for talking about types of objects,
activities, and relations. And it is adults who continually check
up with young children on just what meanings the children
intended to convey .
Eve V. Clark
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bloom , Paul . 2000 . How Children Learn the Meanings of Words .
Cambridge, MA : MIT Press . Tis book reviews how meaning acquisi-
tion is linked to speaker intentions and theory of mind.
Clark , Eve V. 1993 . Te Lexicon in Acquisition . Cambridge: Cambridge
University Press . Tis book examines childrens word learning and
their ability to coin words to fll gaps in their current vocabulary.
. 2009 . First Language Acquisition . 2
nd
ed. Cambridge: Cambridge
University Press . A review of frst language acquisition, and how cogni-
tive and social factors interact in a usage-based approach.
Clark , Eve V. , and Barbara F. Kelly , eds. 2006 . Constructions in Acquisition .
Stanford, CA : CSLI . Tis book reports on studies of how children
acquire constructions.
Markman , Ellen M. 1989 . Categorization and Naming in Children .
Cambridge, MA : MIT . Tis book presents a constraints-based account
of lexical acquisition.
Tomasello , Michael . 2003 . Constructing a Language . Cambridge : Harvard
University Press . Tis book presents a usage-based approach to early
syntactic acquisition (see syntax, acquisition of ).
LEXICAL-FUNCTIONAL GRAMMAR
Lexical functional grammar (LFG) is what is known as a con-
straint-based parallel correspondence architecture for a theory of
language (Bresnan 1982 , 2001 ; Dalrymple 2001 ). It was called lexi-
cal because certain relations between elements, like that between
an active and a passive verb, were dealt with in the lexicon, as a
relation between lexical items. Tis contrasts with the approach
in transformational theories. Functional ambiguously refers
to grammatical relations , which are prominent in the theory, and
mathematical functions , which are used in the LFG-formalism.
Te LFG-formalism can be mathematically modeled and, hence,
analyses expressed within it are susceptible to computational test-
ing (see Dalrymple et al. 1995 ). LFG is often referred to as a syn-
tactic theory , but like many other syntactic theories, it is actually
a framework within which theories of language can be expressed .
As one of the founders of LFG puts it: [T]he formal model of LFG
is not a syntactic theory in the linguistic sense. Rather, it is an
architecture for syntactic theory. Within this architecture, there
is a wide range of possible syntactic theories and sub-theories,
some of which closely resemble syntactic theories within alterna-
tive architectures, and others of which difer radically from famil-
iar approaches (Bresnan 2001, 43) .
IP
DP I
VP
(1)
Tere is assumed to be a fair amount of typological
variation in c-structure among languages. A language such as
Wambaya, which has relatively free word order apart from the
constraint that an auxiliary-like element has to appear in second
position, is assumed to have a functional category IP, where the
Lexical-Functional Grammar
436
fll them from those that are restricted in this way; a subject, for
instance, can be associated with a large number of thematic
roles, whereas an oblique is restricted as to which thematic
role can fll it. LMT associates feature values to thematic roles
intrinsically for example, an agent is intrinsically associated
with [o] and by default for instance, the highest thematic
role according to a typologically motivated thematic hierarchy
is associated with [r]. Grammatical functions are then defned
in terms of these two features; a subject, for instance, is [-o] in
not being object-like and [-r] in not being restricted to any par-
ticular thematic role, whereas an oblique is [-o] but [+r]. Two
well-formedness constraints apply to the mapping; the func-
tion - argument bi-uniqueness condition which states that each
thematic role must be associated with exactly one function and
each function with exactly one thematic role and the subject
condition which states that every predicate must have a sub-
ject. For examples of how LMT works and how it can analyze
constructions such as locative inversion or complex predicates
in interesting ways, see, for instance, Dalrymple ( 2001 , Chap. 8)
or Bresnan (2001, Chap. 14) .
The mapping between c-structure and f-structure can be
structurally defined or identified through morphological ele-
ments. English is an example of a language wherein functions
are determined through their hierarchical position; the sub-
ject appears in the specifier of the IP. This is then captured
formally through the phrase structure rule in (3a), where the
up arrow should be read as the f-structure associated with
my mother node and the down arrow as the f-structure
associated with this node. The resulting tree can be found in
(3b), where indices have been inserted to identify f-structures;
the f-structure associated with the IP is referred to as f
1
, and
so on.
(3)
a.
I takes an exocentric category S as its complement (for a discus-
sion of the relevant data and potential constraints on the typo-
logical variation in constituent structure, see Nordlinger 1997 ).
F-structure is assumed to be reasonably invariant among
languages. It takes the shape of feature-value matrices, where
the features capture grammatical relations and functional fea-
tures. Te simplest features are those with atomic values, such as
[ number plu ]. Grammatical relations such as subj( ect), obj (ect)
or adj (unct) are represented as features that take f-structures as
their values. Each element that has lexical semantic content has
associated with it a feature pred , which has a semantic form as
its value. A verb such as tickle , for instance, has the feature-value
pair [ pred tickle < ( subj ) ( obj ) >], that is, in the pred value of this
verb, tickle requires a subject and an object (for more details
on semantics within LFG, see especially Dalrymple 2001 , 21754,
who develops an approach to semantic composition called glue
semantics). Te pred feature also captures selectional properties,
which are based on functions and functional properties, rather
than on syntactic categories; a transitive verb selects for a subject
and an object, not for two noun phrases. An f-structure for the sen-
tence Oscar tickled the cat can be found in (2).
(2)
PRED tickle SUBJ OBJ
SUBJ
PRED Oscar
GEND
( ) ( )
masc
OOBJ
PRED cat
NUM
SPEC
TENSE
sg
def
past
Tree well-formedness conditions apply to f-structures. Te
general uniqueness condition requires each feature to have a
unique value. Te completeness and coherence conditions ensure
compatibility between the requirements of a pred feature and its
local f-structure. Completeness requires that all functions spec-
ifed by an elements pred feature be present in the f-structure
built up around that element; if, for instance, there had been
no obj in (2), completeness would have been violated. It also
requires those functions to have a semantic value, which prevents
an argument position from being flled by an expletive pronoun,
for instance. Te coherence condition requires that all functions
present in a local f-structure be licensed by another elements
pred feature. If, for instance, there had been an obl( ique) in (2),
coherence would have been violated since no such function is
licensed by tickle . Tis is one example of the way that constraints
accounted for in terms of structure in other approaches are
expressed through f-structure in LFG .
Te information captured in the pred feature is not a primi-
tive of the theory. Te syntactic valency is, in fact, derived from
the semantic roles associated with the verb; hence, this aspect
of the f-structure is derived from the a-structure of the element.
Te relation between semantic and syntactic valency is speci-
fed through lexical mapping theory (LMT). LMT works in terms
of two features [o(bject)] and [r(estrictive)]. Te feature [o]
captures the fact that certain thematic roles cannot fll an
object function, for instance agents , which can be subjects or
obliques, but not objects. Te feature [r] distinguishes those
functions that are not restricted as to which thematic roles can
b. IP f
1
DP
(SUBJ)=
f I'
=
f
3
DP f
2
Te arrows can now be replaced by the indices, and we get the
equations in (4).
(4)
1
f =
2
f
1
f =
3
f
SUBJ
Tese equations refer to three f-structures and defne their rela-
tions; the f-structure f
1
contains a feature subj which has as its
value the f-structure f
2
. Te second equation states that f
1
is iden-
tical to f
3
, which means that any featurevalue pair associated
with either of the two will also be associated with the other; the
two nodes IP and I will then be associated with one f-structure.
In fact, the categories that form the clausal backbone CP, IP
and VP will always share f-structure, so that featurevalue pairs
introduced to any of them will also be shared by the others. Tese
Lexical-Functional Grammar
437
Information about dimensions of information not discussed
here can be found in publications listed in the bibliography on
the LFG Web site. At this site, information can also be found on
extensions and applications of LFG, such as combining LFG with
optimality theory (Bresnan 2000 ), as well as work by Rens
Bod and Ron Kaplan ( 2003 ), which combines linguistic theory
and statistical methods to create an exemplar-based theory of
syntax .
Kersti Brjars
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bod , Rens , and Ron Kaplan . 2003 . DOP model for lexical-functional
grammar. In Data-Oriented Parsing , ed. Rens Bod , Remko Scha , and
Khalil Simaan , 21133. Stanford, CA : CSLI Publications .
Bresnan , Joan . 2000 . Optimal syntax. In Optimality Teory: Phonology,
Syntax and Acquisition , ed. Joost Dekkers , Frank van der Leeuw , and
Jeroen van de Weijer , 33485. Oxford : Oxford University Press .
. 2001 . Lexical-Functional Syntax . Oxford : Blackwell .
Bresnan , Joan , ed. 1982 . Te Mental Representation of Grammatical
Relations . Cambridge, MA : MIT Press .
Dalrymple , Mary . 2001 . Lexical Functional Grammar . San Diego,
CA : Academic Press .
Dalrymple , Mary , Ronald M. Kaplan , John T . Maxwell III , and Annie
Zaenen . 1995 . Formal Issues in Lexical-Functional Grammar . Stanford,
CA : CSLI Publications .
Falk , Yehuda . 2001 . Lexical-Functional Grammar: An Introduction to
Parallel Constraint-Based Syntax . Stanford, CA : CSLI Publications .
Nordlinger , Rachel . 1997 . Constructive Case: Evidence from Australian
Languages . Stanford, CA : CSLI Publications .
Te LFG Web site is located online at: http://www.essex.ac.uk/
linguistics/external/LFG/.
LEXICAL LEARNING HYPOTHESIS
According to this hypothesis, childrens grammatical develop-
ment is incremental and driven by the learning of lexical ele-
ments (see Pinker 1984 ; Clahsen 1996 ; and Eisenbeiss 2000 ,
2003 , 2009 for overviews and references). Te lexical learn-
ing hypothesis was developed by proponents of generative
grammar in order to address the poverty-of-the-stimulus
argument: In order to produce and understand new sentences,
children must generalize beyond individual input utterances.
However, they do not have reliable access to systematic correc-
tions that would allow them to reject incorrect generalizations
about the target language . Terefore, generative linguists have
postulated an innate language acquisition device, uni-
versal grammar (UG), that constrains childrens hypothesis
space. According to the principles and parameters the-
ory , UG contains i) principles that constrain all grammatical
representations and ii) open parameters that provide a fnite
set of values, that is, options from which learners can choose
(Chomsky 1981 ). For instance, generative linguists assume
that all sentences contain subjects, but that languages may dif-
fer with respect to the positioning of subjects and their overt
realization (e.g., optional subjects in Italian versus obligatory
subjects in English). In such a model, language acquisition
only involves i) setting parameters to their target values and ii)
acquiring the lexicon .
categories are referred to as co-heads (Bresnan 2001, 102). Te
equations in (4) give rise to the partial f-structure in (5) .
(5)
f
1
f
3
f
2
SUBJ [ ]
As further elements are added, the information contributed by
their lexical entries or by functional equations associated with
structure will be inserted into the f-structure as dictated by the
functional equations. Tis mapping procedure from c-structure
to f-structure, like all LFG mapping relations, has the property
of monotonicity; information can be added but never deleted,
moved, or changed.
For a language like Latin, there are no arguments for an
elaborate hierarchical clause structure, but rather a fatter exo-
centric structure is appropriate. In such a language, functions
are not defned structurally and there is no structural equation
of the kind illustrated in (3a). Instead, functions are identi-
fed through case marking, and this is captured directly in LFG
through an association between the value for the feature case
and a function. For Latin, there would then be a global equation
as in (6).
(6)
Tis equation can be inserted at any noun phrase node and is
read as if the f-structure associated with this node contains the
featurevalue pair [ case = nom ], then the f-structure associated
with the node above contains the feature subj and the f-structure
associated with this node is the value of that subj feature. Or in
less formal language, if this node is nominative, then it is the sub-
ject of the node above.
Te mapping principles permit non-one-to-one correspon-
dences between dimensions of information. For instance, an
f-structure can contain a subj function without there being
a noun phrase in the corresponding c-structure. Tis is how
constructions generally referred to as pro-drop are analyzed,
though in LFG they are more appropriately named pronoun
incorporation . Te verb in the Italian sentence Rido I laugh,
for instance, is analyzed as consisting of just a verb. Tis verb
contains in its f-structure description the type of informa-
tion contributed by the subject pronoun in the corresponding
English sentence. Te verb form rido would be associated with
equations such as those in (7).
(7)
The crucial part of (7) is the equation that introduces a pred
feature with a pronominal value for its subject . This means
that the principle of completeness is satisfied by the verb
itself.
Some of the fundamental properties of LFG have been illus-
trated here mainly through reference to f- and c-structure.
(CASE) = nom (SUBJ) =
Lexical-Functional Grammar Lexical Learning Hypothesis
438
but to their individual grammatical features (e.g., tense ), which
are stored in lexical entries for grammatical morphemes and
project to phrases whenever these morphemes are combined.
According to such models, children should be able to acquire
individual features independently of one another, integrate them
into lexical entries for individual lexical/morphological elements
in an item-by-item fashion, and project each of these features
into phrases when these elements are combined. Tus, whether
or not a childs utterance involves a realization of a particular
grammatical feature and the corresponding syntactic operations
does not depend on a global parameter value. Rather, it depends
on the individual lexical items that the child has acquired so far.
Hence, developmental dissociations between individual lexical
items and individual features are expected. For instance, def-
nite and indefnite articles are diferent lexical realizations of the
functional category determiner , and German children acquire
indefnite articles before defnite articles. Similarly, when they
start producing defnite articles, German children use feminine
forms correctly, but then incorrectly combine masculine forms
of articles with both masculine and neuter nouns. Tis suggests
that German children acquire the [FEMININE] distinction
before they instantiate the feature [MASCULINE] that distin-
guishes masculines from neuters.
Tus, in sum, the lexical learning hypothesis, that is, the idea
that syntactic development is driven by lexical development,
can provide accounts for the incremental nature of syntactic
development, as well as for the observed correlations between
lexical and syntactic development and the developmental dis-
sociations that have been observed in childrens grammatical
development .
Sonja Eisenbeiss
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , Noam . 1981 . Lectures on Government and Binding . Dordrecht,
the Netherlands : Foris .
. 1986 . Barriers . Cambridge, MA : MIT Press .
. 1989 . Some notes on economy of derivation and representa-
tion. In MIT Working Papers in Linguistics 10, 4374, Massachusetts
Institute of Technology, Cambridge .
Clahsen , Harald , ed. 1996 . Generative Perspectives on Language
Acquisition: Empirical Findings, Teoretical Considerations and
Crosslinguistic Comparisons . Amsterdam : Benjamins . With relevant
contributions by Harald Clahsen, Sonja Eisenbeiss, and Martina
Penke; Jrgen Meisel and Maria-Jose Ezeizabarrena; Andrew Radford;
and Tomas Roeper.
Eisenbeiss , Sonja . 2000 . Te acquisition of the determiner phrase in
German child language. In Te Acquisition of Syntax: Studies in
Comparative Developmental Linguistics , ed. M.-A. Friedemann and
L. Rizzi , 2662. London : Longman .
. 2003 . Merkmalsgesteuerter Grammatikerwerb. Eine Untersuchung
zum Erwerb der Struktur und Flexion der Nominalphrase. Available
online at: http://www.ub.uni-duesseldorf.de/home/etexte/diss/
show?dissid=1185.
. 2009 . Generative approaches to language learning. Linguistics
47 .2: 273310.
Manzini , Rita , and K. Wexler . 1987 . Parameters, binding theory, and
learnability . Linguistic Inquiry 18 (July): 413 44.
Pinker , Steven . 1984 . Language Learnability and Language Development .
Cambridge : Harvard University Press .
If one assumes such a powerful acquisition device, one must
explain why children need several years to acquire their tar-
get grammar and initially produce non-target-like sentences
for example, subjectless sentences in English. Faced with this
developmental problem, proponents of the lexical learning
hypothesis argue that UG is available from the onset of grammat-
ical development, but in order to set parameters, children still
need to learn the grammatical properties of the lexical elements
associated with these parameters.
Tese assumptions are in line with lexicalist generative
models: Initially, parameters referred to a heterogeneous set of
linguistic properties, for example, subject omissions, word order,
or morphological marking. However, cross-linguistic (parametric)
variation is closely linked to lexical properties, in particular to
properties of grammatical morphemes (see, e.g., Manzini and
Wexler 1987 ). For instance, Germanic languages with postverbal
negation exhibit a morphological distinction between frst and
second person. Proponents of lexicalist models argue that this sug-
gests a relationship between parameter values for word order and
the person specifcations of subject-verb- agreement markers.
In recent generative models, such markers or function words (e.g.,
auxiliaries) are analyzed as realizations of functional categories
that project to phrases, just like the lexical categories verb and
noun . For instance, subject-verb-agreement markers are viewed
as realizations of the functional category INFL (Chomsky 1986).
Proponents of lexical learning regard functional categories
as the only source of parametric variation (Chomsky 1989), and
they argue that children should fx parameters and build up pro-
jections of functional categories by learning the properties of the
lexical elements that encode the respective functional categories.
Hence, one should fnd developmental correlations between the
acquisitions of lexical items and the acquisition of the syntactic
properties associated with the projections of the corresponding
functional categories. Such correlations have been documented
for instance, a correlation between the acquisition of the German
subject-verb-agreement paradigm and the target-like ordering
of subjects, verbs, and negation (Clahsen 1996 ). Moreover, if one
assumes incremental phrase-structure building, one can explain
developmental dissociations between realizations of diferent
functional categories for instance, the observation that German
children master the use of agreement markers associated with
INFL before they consistently produce complementizers, that is,
realizations of the functional category COMP .
Children show even more complex dissociations, however
(Eisenbeiss 2003): First, they start to realize diferent features of
the same category at diferent points. For instance, for the cat-
egory case , German children mark the nominative/accusative
distinction before the accusative/dative distinction. Second, chil-
dren do not acquire all instantiations of the same features simul-
taneously. For example, German children show case distinctions
on pronouns earlier than on articles. Tird, childrens realiza-
tions of functional categories show lexeme-specifc restrictions.
For instance, German children initially restrict the possessive -s
to some familiar names (e.g., Mamas mommys).
Tese observations can be captured in feature-based, lexi-
calist versions of the lexical learning hypothesis (see Eisenbeiss
2003, 2009 for discussion): In these models, cross-linguistic var-
iation is not so much related to functional categories, as such,
Lexical Learning Hypothesis
439
A broad perspective on the neural correlates of spoken word
recognition is provided by P. Indefrey and A. Cutler ( 2004 ), who
report a meta-analysis of 55 experiments in which subjects pas-
sively listened to tones , pseudowords, words, or sentences . It
was found that all of the diferent types of auditory stimuli reliably
activate overlapping, as well as partially diferentiated, central
and posterior regions of the superior temporal gyri in both hemi-
spheres. In addition, the following hierarchical organization was
observed: As the linguistic complexity of the stimuli increases,
there is recruitment of progressively more anterior regions of the
left superior temporal sulcus. Tus, moving anteriorly, there is
frst an area responsive to pseudowords but not tones, then an
area responsive to words but not pseudowords, and fnally an
area responsive to sentences but not words. Te anterior area
that is selectively activated by words may contribute to the res-
olution of the lexical competition process described here; how-
ever, it is also conceivable that this operation is subserved by one
of the more posterior word-specifc areas (Orfanidou, Marslen-
Wilson, and Davis 2006 ). After the phonological form of a word
has been recognized, its semantic and syntactic components are
retrieved. As summarized by Indefrey and Cutler ( 2004 ), these
processes may be executed by a wide distribution of predomi-
nantly left hemisphere brain regions, including most notably
the middle and inferior temporal gyri and the posterior inferior
frontal gyrus.
Turning to spoken word production, one of the most infu-
ential theories is that proposed by W. J. M. Levelt, A. Roelofs,
and A. S. Meyer ( 1999 ). According to their model, the produc-
tion of spoken content words depends on multiple processing
stages, each of which generates its own characteristic output
representation ( Figure 1). First, conceptual preparation involves
identifying the meaning of the word to be produced. Second,
lexical selection involves activating the lemma for the word
that is, a unit that intervenes between semantics and phonology
and that serves as the gateway to syntactic features (e.g., gram-
matical category, number, tense , etc.; these features are not
shown in Figure 1). Tird, form retrieval involves calling up the
phonological code for the word. Fourth, syllabifcation involves
determining segmental clusters and metrical assignments. Fifth,
phonetic encoding involves transforming syllabic units into
motor instructions. And sixth, articulation involves the fnal pro-
gramming of overt speech .
Te neural correlates of the frst stage, conceptual prepara-
tion, remain mysterious, largely because this stage constitutes
the complex interface between language and thought and is also
heavily infuenced by social-cognitive perspective-taking abili-
ties for example, the same piece of real estate can be called the
coast, the shore, or the beach, depending on ones communicative
goals (Tomasello 1999 , 119). Future research may show that con-
ceptual preparation is subserved by widespread cortical struc-
tures that underlie semantic processing (Kemmerer 2010; see
semantics, neurobiology of ). Te next two stages, lemma
selection and phonological form retrieval, both involve core
lexical processes, and their neural correlates are beginning to be
understood. In a meta-analysis of 58 functional imaging studies
including several studies employing magnetoencephalography,
which has excellent temporal resolution Indefrey and Levelt
( 2004 ) found that lemma selection is linked with the midsection
LEXICAL PROCESSING, NEUROBIOLOGY OF
Te lexicon is the store of words in the mental dictionary.
A typical English-speaking high school graduate knows about
60,000 words, a literate adult perhaps twice that number (Miller
1991 , 138). A word can be regarded as a long-term memory
association of semantic , syntactic , phonological , and
orthographic structures. For example, the lexical entry for rose
includes the following components, with the semantic compo-
nent symbolized by a picture for convenience:
rose
meaning:
part of speech: noun
phonology: /roz/
orthography: ROSE
During the past two decades, there has been remarkable prog-
ress in understanding the neural substrates of lexical processing,
mainly because of advances in two complementary approaches for
investigating the functions of specifc brain structures: 1) the lesion
method, which, when used with ample numbers of patients who
are carefully studied both neuropsychologically and neuroanatomi-
cally, can yield indispensable insights about the neural systems that
are necessary for particular abilities; and 2) functional imaging tech-
niques, such as fMRI, which allow researchers to identify with more
fne-grained spatial resolution the brain structures that are engaged
during the normal performance of certain tasks (see neuroimag-
ing ). Much more has been learned about the neural substrates of
lexical processing than can be summarized here, and so this review
concentrates on cortical regions that have been linked with the rec-
ognition and production of spoken and written word forms.
Neural Substrates of Spoken Word Recognition and
Production
It is well established that the sensorimotor aspects of spoken
word processing depend on the left perisylvian cortex , and
there is growing evidence that both the posterior superior tem-
poral (auditory-related) and the posterior inferior frontal
(motor-related) sectors of this large anatomical territory contrib-
ute to both speech perception and speech production (Imada
et al. 2006 ; Okada and Hickok 2006 ; Pulvermller et al. 2006 ;
Skipper et al. 2008). Tese two regions interact not only via direct
connections but also via an indirect pathway mediated by the
inferior parietal lobule (Catani, Jones, and Ffytche 2005 ).
To understand spoken words, listeners must frst use the audi-
tory input to activate stored representations of lexical-phono-
logical form. It is only after this process of lexical access has been
achieved that the semantic and syntactic properties of words can
be activated and used to construct higher-level representations of
the utterance. Numerous behavioral studies suggest that speech
information is continuously projected to the lexicon, so that an
initial sequence like bla will activate all the words in the listen-
ers lexicon that begin with those sounds ( black, bland, blanket,
etc.); as the input accumulates, the set of activated words dimin-
ishes until only a single one matches the input, at which point
recognition can be said to occur (McQueen, Dahan, and Cutler
2003 ). Pseudowords (e.g., blash ) also activate partially matching
candidate words, but ultimately no winner is selected.
Lexical Processing, Neurobiology of
440
related brain structures; however, the exact neural correlates of
each stage are not yet clear (Bohland and Guenther 2006 ).
Independently of Levelt, Roelog, and Meyers (1999) model,
a great deal of neuroscientifc research has focused on the pro-
cess of mapping the meanings of words onto their correspond-
ing phonological forms during speech production . One
important line of work, conducted by Hanna Damasio, Daniel
Tranel, and their colleagues (2004), suggests that this process
of the left middle temporal gyrus and typically occurs during a
time window of 150225 milliseconds (ms) post-stimulus in oral
picture-naming tasks (see Color Plate 4 ). Tey also found that
phonological form retrieval is linked with the posterior portions
of the left middle and superior temporal gyri and occurs during
a time window of either 200400 or 275400 ms, depending on
the studies that are considered. Te three postlexical stages of
spoken word production are known to rely on a variety of motor-
Figure 1. The LRM (i.e., Levelt, Roelofs, and Meyer) model of spoken word production. Left column: Word pro-
duction tasks involving lead-in processes that enter the central word production architecture at different stages.
Middle column: Core processes of word production and their characteristic output. Right column: Example fragments
of outputs generated at each stage. Reprinted by permission from Elsevier, copyright 2004, from P. Indefrey and
W. Levelt, The spatial and temporal signatures of word production components, Cognition 92: 10144.
Lexical Processing, Neurobiology of
441
the VWFA was reported by R. Gaillard et al. ( 2006 ; see also Martin
2006 ). In brief, prior to surgery for intractable epilepsy, the patient
exhibited normal single-word reading, including a lack of increase
in reading time for common words varying in length from three to
eight letters; moreover, fMRI revealed his VWFA to have normal
functional-anatomical characteristics, and local feld potentials
recorded from implanted electrodes showed that this area was
sensitive to word frequency but not word length, again within
normal parameters. After excision of tissue just posterior to the
VWFA, the patients epileptic seizures were successfully elim-
inated, but his reading was markedly slow and inaccurate, with
reading times increasing linearly with word length (i.e., letter-
by-letter reading). In addition, the VWFA no longer responded
to printed words, even when they were contrasted with a simple
fxation point. Tis study, therefore, provides powerful new evi-
dence that the VWFA is in fact necessary for access to the stored
orthographic forms of words during reading.
Writing also depends on a large network of widely distributed
brain regions (Hillis and Rapp 2004 ; Rapcsak and Beeson 2002 ).
Information about the neural basis of lexical access during writ-
ten word production comes primarily from patients with lexical
agraphia , a disorder in which words with regular mappings
between phonology and orthography are spelled correctly, but
words with irregular mappings (e.g., choir ) are misspelled. Te
errors are usually phonologically plausible (e.g., circuit serkit )
and afect low-frequency words more than high-frequency
ones. Lexical agraphia is typically caused by damage to the left
temporo-parieto-occipital junction (Brodmann areas 37 and/or
39), although in some cases, there is involvement of the left ven-
tral occipitotemporal region, close to, if not encompassing, the
VWFA. Several functional imaging studies with normal subjects
provide further support for a role of these cortical regions in writ-
ten word production (e.g., Petrides Alvisatos, and Evans 1995 ;
Nakamura et al. 2000 ) .
Conclusion
When people recognize and produce the spoken and written
forms of words, they usually concentrate on the meanings being
expressed and remain blithely unaware of the complex compu-
tations being executed by their brains in order to rapidly and
efectively process the lexical structures themselves. Cognitive
neuroscience is beginning to reveal the intricasies of these neural
systems, and dramatic advances are likely to happen in the com-
ing years. Exciting new discoveries are appearing in the literature
almost daily, and this explosion of research will undoubtedly
provide fresh insights into the neurobiology of lexical process-
ing, with signifcant implications for understanding and treating
disorders that result from brain injury .
David Kemmerer
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bohland , J. W. , and F. H. Guenther . 2006 . An fMRI investigation of sylla-
ble sequence production . NeuroImage 32 : 821 41.
Catani , M. , D. K. Jones , and D. H. Ffytche . 2005 . Perisylvian language
networks of the human brain . Annals of Neurology 57 : 8 16.
Damasio , H. , T. J. Grabowski , D. Tranel , R. D. Hichwa , and A. R. Damasio .
1996 . A neural basis for lexical retrieval . Nature 380 : 499 505.
is subserved by intermediary units that are analogous to lem-
mas insofar as they function as relays, taking lexical-semantic
structures as input and then pointing to the appropriate lexical-
phonological structures. It is interesting that these intermediary
units may be neurally organized according to both semantic and
grammatical principles. For example, lesion data suggest that,
contrary to Indefrey and Levelts ( 2004 ) proposal, the retrieval
of nouns for diferent categories of concrete entities may hinge
on intermediary units that do not reside in the left middle tem-
poral gyrus but, rather, in the left temporal pole (TP) and inf-
erotemporal (IT) cortices. Specifcally, studies in which oral
picture-naming tasks have been administered to large cohorts of
brain-damaged patients have shown that 1) impaired access to
proper nouns for unique persons (e.g., Jennifer Aniston ) is asso-
ciated with left TP lesions, 2) impaired access to common nouns
for animals (e.g., horse ) is associated with damage to the anterior
sector of left IT, and 3) impaired access to common nouns for
tools (e.g., hammer ) is associated with damage to the posterior
sector of left IT, a region called IT+ (Damasio et al. 1996; Damasio
et al. 2004). Crucially, the patients have intact object recognition
and conceptual knowledge since they can accurately describe
the entities they cannot name; in other words, the disorders are
purely anomic. Furthermore, functional imaging data indicate
that the same cortical regions are activated in normal subjects
in the same category-specifc ways when concrete entities are
orally named from either pictures (Damasio et al. 1996 , 2004 )
or characteristic sounds (Tranel et al. 2003 ; Tranel et al 2005 ).
Tere is also increasing evidence from several methodologies
that the process of retrieving action verbs engages a quite difer-
ent neural pathway that includes the left ventrolateral premotor/
prefrontal cortex (Damasio et al. 2001 ; Shapiro and Caramazza
2004 ; Tranel et al. 2001 ; Tranel et al. 2008). Tis region is reliably
activated when action verbs are accessed, and damage to it fre-
quently impairs the production of verbs but not nouns .
Neural Substrates of Written Word Recognition and
Production
Reading and writing are recent inventions in human history
and must be explicitly taught. For literate individuals, how-
ever, word representations include not just a phonological
component but also an orthographic component that is ef -
ciently processed by neural circuits that are gradually being
elucidated.
Te activity of reading recruits numerous brain regions in
the temporal, parietal, and frontal lobes (Hillis and Rapp 2004 ;
Hillis and Tuf ash 2002 ). Perhaps the most controversial region,
however, has been the visual word form area (VWFA), located
in the left occipitotemporal sulcus bordering the fusiform gyrus
(McCandliss, Cohen, and Dehaene 2003 ; Dehaene 2005 ). Tis
area responds more strongly to printed words than to other types
of visually presented objects, such as faces, animals, and tools.
Also, disruption of the input projections to this area can induce
pure alexia without agraphia , a disorder in which reading can be
accomplished only in a laborious letter-by-letter manner, while
writing and all other linguistic skills are unafected. Despite these
fndings, the question of whether the VWFA plays a genuine causal
role in reading has been hotly debated (e.g., Price and Devlin
2003 ). Recently, however, a compelling case study supporting
Lexical Processing, Neurobiology of
442
Price , C. J. , and J. T. Devlin . 2003 . Te myth of the visual word form area .
NeuroImage 19 : 473 81.
Pulvermller , F. , M. Huss , F. Kherif , F. M. del Prado Martin , O. Hauk ,
and Y. Shtyrov . 2006 . Motor cortex maps articulatory features of
speech sounds . Proceedings of the National Academy of Sciences
103 : 7865 70.
Rapcsak , S. Z. , and P. M. Beeson . 2002 . Neuroanatomical corre-
lates of spelling and writing. In Te Handbook of Adult Language
Disorders: Integrating Cognitive Neuropsychology, Neurology, and
Rehabilitation , ed. A. E. Hillis , 71100. Philadelphia : Psychology
Press .
Rapp , B. , and M. Goldrick . 2006 . Speaking words: Contributions of
cognitive neuropsychological research . Cognitive Neuropsychology
23 : 39 73.
Shapiro , K. , and A. Caramazza , A . 2004 . Te organization of lexical
knowledge in the brain: Te grammatical dimension. In Te Cognitive
Neurosciences. Vol. 3. Ed. M. Gazzaniga , 80314. Cambridge, MA : MIT
Press .
Skipper , J. I. , V. van Wassenhove , H. C. Nusbaum , and S. L. Small . 2008 .
Hearing lips and seeing voices: How cortical areas supporting speech
production mediate audiovisual speech perception . Cerebral Cortex
18: 243949 .
Tomasello , M. 1999 . Te Cultural Origins of Human Cognition .
Cambridge : Harvard University Press .
Tranel , D. , R. Adolphs , H. Damasio , and A. R. Damasio . 2001 . A neural
basis for the retrieval of words for actions . Cognitive Neuropsychology
18 : 655 70.
Tranel , D. , H. Damasio , G. R. Eichhorn , T. J. Grabowski , L. L. B. Ponto ,
and R. D. Hichwa . 2003 . Neural correlates of naming animals from
their characteristic sounds . Neuropsychologia 41 : 847 54.
Tranel , D. , T. J. Grabowski , J. Lyon , and H. Damasio , H. 2005 . Naming
the same entities from visual or from auditory stimulation engages
similar regions of left inferotemporal cortices . Journal of Cognitive
Neuroscience 17 : 1293 1305.
Tranel , D. , K. Manzel , E. Asp , and D. Kemmerer . 2008 . Naming static and
dynamic actions: Neuropsychological evidence . Journal of Physiology,
Paris 102 : 8094.
LEXICAL RELATIONS
Lexical relations are ways in which words , or lexemes, share
something in common. Tis broad defnition includes relations
based on phonological relatedness, such as rhyming , and
morphological relatedness, like being the range of tensed
forms of a verb. However, in most contexts, the term is used to
refer specifcally to semantic relations among words and, most
frequently, to paradigmatic semantic relations among words,
including synonymy, hyponymy, and antonymy. Such relations
are sometimes called sense relations (Lyons 1977 ), as it is usually
a single denotative sense of a word rather than every sense and
every aspect of the lexical item that is relevant to the relation.
Tus, we expect the postal, rubber, and stomp senses of
stamp to enter into lexical relations with diferent sets of words .
Paradigmatic and Syntagmatic Relations
Semantic relations are generally divided into two types, usu-
ally called paradigmatic and syntagmatic . Syntagmatic rela-
tions are relations of combination that is to say, words that
fll diferent slots in a phrase, like book and read or delicious
and food . Tese can be grouped into relational types like mod-
ifermodifed or event verbagent. Some theories of lexical
Damasio , H. , T. J. Grabowski , D. Tranel , L. L. B. Ponto , R. D. Hichwa , and
A. N. Damasio . 2001 . Neural correlates of naming actions and of nam-
ing spatial relations . NeuroImage 13 : 1053 64.
Damasio , H. , D. Tranel , T. Grabowski , R. Adolphs , and A. N. Damasio .
2004 . Neural systems behind word and concept retrieval . Cognition
92 : 179 229.
Dehaene , S. 2005 . Evolution of human cortical circuits for reading and
arithmetic: Te neuronal recycling hypothesis. In From Monkey
Brain to Human Brain , ed. S. Dehaene , J.-R. Duhamel , M. Hauser , and
G. Rizzolatti , 13358. Cambridge, MA : MIT Press .
Gaillard , R. , L. Naccache , P. Pinel , S. Clemenceau , E. Volle , D. Hasboun ,
S. Dupont , M. Maulac , S. Dehaene , C. Adam , and L. Cohen . 2006 .
Direct intracranial, fMRI, and lesion evidence for the causal role of
left inferotemporal cortex in reading . Neuron 50 : 91 204.
Hillis , A. E. , and B. C. Rapp . 2004 . Cognitive and neural substrates of
written language: Comprehension and production. In Te Cognitive
Neurosciences. Vol. 3. Ed. M. Gazzaniga , 77587. Cambridge, MA : MIT
Press .
Hillis , A. E. , and E. Tuf ash . 2002 . Neuroanatomical aspects of reading.
In Te Handbook of Adult Language Disorders: Integrating Cognitive
Neuropsychology, Neurology, and Rehabilitation , ed. A. E. Hillis , 1526.
Philadelphia : Psychology Press .
Imada , T. , Y. Zhang , M. Cheour , S. Taulu , A. Ahonen , and P. K.
Kuhl . 2006 . Infant speech perception activates Brocas area:
A developmental magnetoencephalographic study . NeuroReport
17 : 957 62.
Indefrey , P. , and A. Cutler . 2004 . Prelexical and lexical processing.
In Te Cognitive Neurosciences. Vol. 3. Ed. M. Gazzaniga , 75974.
Cambridge, MA : MIT Press .
Indefrey , P. , and W. J. M. Levelt . 2004 . Te spatial and temporal signa-
tures of word production components . Cognition 92 : 101 44.
Kemmerer , D. 2010 . How words capture visual experience: Te per-
spective from cognitive neuroscience. In Words and the Mind: How
Words Capture Human Experience , ed. B. Malt and P. Wolf, 289329 .
Oxford : Oxford University Press .
Levelt , W. J. M. , A. Roelofs , and A. S. Meyer . 1999 . A theory of lexical
access in speech production . Behavioral and Brain Sciences 22: 138.
Martin , A. 2006 . Shades of Dejerine forging a causal link between the
visual word form area and reading . Neuron 50 : 173 5.
McCandliss , B. D. , L. Cohen , and S. Dehaene . 2003 . Te visual word form
area: Expertise for reading in the fusiform gyrus . Trends in Cognitive
Sciences 7 : 293 9.
McQueen , J. M. , D. Dahan , and A. Cutler . 2003 . Continuity and
gradedness in speech processing. In Phonetics and Phonology in
Language Comprehension and Production: Diferences and Similarities ,
ed. N. Schiller and A. Meyer , 3776. New York : Mouton de Gruyter .
Miller , G.A. 1991 . Te Science of Words . New York : Freeman .
Miozzo , M. , and A. Caramazza , eds. 2008. Lexical processing. Cognitive
Neuropsychology 25 .4 (Special Issue).
Nakamura , K. , M. Honda , T. Okada , T. Hanakawa , K. Toma , H. Fukuyama ,
J. Konishi , and H. Shibasaki . 2000 . Participation of the left posterior
inferior temporal cortex in writing and mental recall of kanji orthogra-
phy: A functional MRI study . Brain 123 : 954 67.
Okada , K. , and Hickok , G. 2006 . Left posterior auditory-related cortices
participate both in speech perception and speech production: Neural
overlap revealed by fMRI . Brain and Language 98 : 1121 7.
Orfanidou , E. , W. D. Marslen-Wilson , and M. H. Davis . 2006 . Neural
response suppression predicts repetition priming of spoken words
and pseudowords . Journal of Cognitive Neuroscience 18 : 1237 52.
Petrides , M. , B. Alivisatos , and A. C. Evans . 1995 . Functional activation
of the human ventrolateral frontal cortex during mnemonic retrieval of
verbal information . Proceedings of the National Academy of Sciences
92 : 5803 7.
Lexical Processing, Neurobiology of Lexical Relations
443
synonyms, for example, when the same object is named inde-
pendently by two subcommunities, either one word falls into dis-
use or one or both of the words become specialized to a slightly
diferent sense or context type. English, through its history of
contact with other languages, has come to have many such
near-synonyms, such as riseascend, smartintelligent , and dead
deceased . While these are substitutable for each other in many
contexts, they all difer in meaning and use. For example, while
a balloon could ascend or rise , a person rises , rather than ascends ,
from a chair. And while people can be dead or deceased , plants
can only be said to be dead . Many other cases of not-quite-synon-
ymy involve words that refer to similar, but not identical, things
for example, tapas and hors doeuvres or shovel and scoop .
Te second problem with a truth-conditional defnition of
synonymy is that it allows as synonyms items that have difer-
ent non-truth-conditional content. For example, the nouns dog ,
doggy , and pooch may all be truthfully applied to a particular
animal, but choosing pooch implies diferent things about the
animal and the speakers relation to it than dog does. Tus, some
apply a more restrictive substitution test that takes into account
connotational and social aspects of meaning, as well as the truth
conditional. To the extent that goodness of synonym relations
can be afected by non-truth-conditional issues like register
and morphological complexity, synonymy can be considered a
lexical relation, as well as a semantic relation.
Synonymy currently receives much attention in compu-
tational linguistics , as language generators and machine
translators require principled means for selecting the most
appropriate word for a context from a lexicon flled with near-
synonyms. Such studies can be particularly concerned with dis-
cerning ways in which synonyms or near-synonyms can difer
(see, e.g., DiMarco, Hirst, and Stede 1993 ) .
HYPONYMY. Hyponymy, the type of relation, is the relation of
sense inclusion, although it is often defned in terms of referen-
tial, or categorial, inclusion. In sense terms, beer is a hyponym
of beverage because the sense of beer includes all the informa-
tion that is in the sense of beverage , plus additional information
that identifes beers as special types of beverages. In catego-
rial terms, everything that beer denotes is included in the set of
everything that beverage denotes. Hyponym relations are, thus,
linguistic refexes of categorial inclusion relations, and, as such,
some theorists consider them to be semantic, but not lexical,
relations .
Unlike synonymy, hyponymy is an asymmetrical relation, in
that beer is a type of beverage but beverage is not a type of beer .
Hyperonymy is the term for the converse relation from beverage
to beer. Hyponymy is usually said to be a transitive relation. For
example, if beer is a hyponym of beverage and beverage is a hyp-
onym of liquid , then beer is a hyponym of liquid. However, tran-
sitivity holds only in cases of proper sense inclusion not in all
cases that pass the is a type/kind of test (thus proving that the test
can be misleading). For example, speed-reading is a type of read-
ing , and reading is a type of leisure activity , but speed-reading is
not usually considered to be a leisure activity . While reading can
function as a leisure activity, it is not defned in terms of being
leisurely, and thus reading is not a hyponym of leisure activity in
the logical sense of the term .
semantics build such relations into lexical entries, for exam-
ple in the selectional restrictions of Jerrold J. Katz and Jerry A.
Fodor ( 1963 ) and the lexical functions of meaning - text theory
(Meluk 1996 ).
Paradigmatic relations are relations of substitutability ; the
words in a semantic paradigm are diferent options for flling the
same phrasal slot. For example, red/yellow/blue are three options
for subject position in X is a primary color. Paradigmatic relations
are studied because of their role in logical relations among sen-
tence meanings, such as entailment, and because of what they
might tell us about how the mental lexicon is organized .
Semantic versus Lexical Relations
Te term lexical relation is ambiguous, in that it can refer either
a) to [semantic] relations among words or b) to [semantic] rela-
tions among words that are represented in the mental lexicon,
as information in or links between the lexical entries for those
words. Some authors reserve lexical relations for the b mean-
ing and use semantic relations for the a meaning. For exam-
ple, Derek Gross, Ute Fischer, and George A. Miller ( 1989 ) claim
that while large and little are semantically opposite, they are not
directly related to each other as lexical antonyms, whereas large
and small are both semantically and lexically related. In other
words, large and small are not only semantically opposed; we
have also learned through linguistic experience that the words
themselves are opposed. Tis distinction between lexical and
nonlexical relations is intended to explain why some word sets
are particularly strongly linked to each other, both in terms of co-
occurrence in corpora and psycholinguistic behavior for
example, in word association experiments .
Types of Paradigmatic Relations
Te most studied paradigmatic lexical-semantic relations are
synonymy, hyponymy, and antonymy/contrast, because a) sub-
stitution of members of these sets typically results in regular con-
sequences for truth conditional semantics , and thus b)
they are central organizational principles in many lexicological
theories (see semantic fields ) . While each of these relations
is easily exemplifed, defnitions and the role of the relations in
the mental lexicon is still the subject of debate. Traditionally,
defnitions that depend on the logical consequences of substi-
tution have been used (e.g., Lyons 1977 ). More recently, D. A.
Cruse (1994) has proposed prototype -based defnitions of
these relations, and M. Lynne Murphy ( 2003 ) has proposed a
pragmatic approach. Next, we look at the relations in turn and
highlight some research issues associated with each.
SYNONYMY. Synonymy, or sameness of meaning, is usually
defned in terms of a substitution test. If word X can be substi-
tuted for a particular sense of word Y in any sentence with no
change to the truth conditions of the sentence, then X and Y are
synonyms. However, this defnition does not include many of the
things that are called synonyms in thesauruses and everyday
discourse, for a few reasons.
First, languages generally avoid synonymy since it is eco-
nomical (both in lexical acquisition and in any interaction) to
assume that diferent forms signal diferent meanings (see, e.g.,
Clark 1992 ). When a language variety chances on a pair of perfect
Lexical Relations
444
in all being direct hyponyms of color but difer in the part of
the spectrum they designate. Still, they are not truly incompat-
ible since they may overlap some shades of turquoise may be
considered to be both blue and green . Te fact that most people
would insist that it must be one or the other, however, indicates
our preference for acting as if contrast sets are incompatible .
OTHER RELATIONS. Te foregoing types of paradigmatic rela-
tion are generally held to be the most important to lexical and
semantic theories, but many more have been noted in the lit-
erature. Te most discussed of these is meronymy, the part of
relation, though this can be thought of as a cover term for many
other types of relation, such as segmentwhole ( slicecake ),
materialwhole ( woodtable ), leaderorganization ( captain
team ), and so on.
Whether a precise taxonomy of relations is possible or neces-
sary is an open question. While semantic feld views of the lex-
icon rely on a small number of well-defned relations, theories
employing looser semantic networks or semantic domains might
allow for as many types of lexical relations as there are possible
relations among entities in the world. As the subtypes of anton-
ymy and meronymy show, there is also the question of the level
of specifcity that needs to be employed in representing these
relations in linguistic theory .
M. Lynne Murphy
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Clark , Eve V. 1992 . Conventionality and contrast. In Frames, Fields, and
Contrasts , ed. Adrienne Lehrer and Eva Feder Kittay, 17188. Hillsdale,
NJ: Erlbaum.
Cruse , D. A. 1986 . Lexical Semantics . Cambridge : Cambridge University
Press .
. 1994 . Prototype theory and lexical relations . Rivista di Linguistica
6 : 167 88.
DiMarco , Crysanne , Graeme Hirst , and Manfred Stede . 1993 . Te seman-
tic and stylistic diferentiation of synonyms and near- synonyms .
Proceedings, AAAI Spring Symposium on Building Lexicons for Machine
Translation, Stanford, CA : 114 21.
Gross , Derek , Ute Fischer , and George A. Miller . 1989 . Te organi-
zation of adjectival meanings . Journal of Memory and Language
28 : 92 106.
Jones , Steven . 2002 . Antonymy . London : Routledge .
Katz , Jerrold J. , and Jerry A. Fodor . 1963 . Te structure of a semantic
theory . Language 39 : 170 210.
Lyons , John . 1977 . Semantics . 2 vols. Cambridge : Cambridge University
Press .
Meluk , I. A. 1996 . Lexical functions: A tool for the description of
lexical relations in a lexicon. In Lexical Functions in Lexicography
and Natural Language Processing , ed. Leo Wanner , 37102.
Amsterdam : Benjamins .
Murphy , M. Lynne . 2003 . Semantic Relations and the Lexicon .
Cambridge : Cambridge University Press .
LEXICAL SEMANTICS
Lexical semantics is often loosely described as the study of word
meaning , but both word and meaning require more precise
defnition in this context. Te term is usually used to describe the
study of lexical, or content, words or lexemes (including nouns,
Inclusion relations, along with contrast relations, are central
to most approaches to lexical meaning, particularly in the treat-
ment of noun meaning. For other word classes, hyponym rela-
tions are less frequent or less clearly paradigmatic. For example,
the adjectives happy and sad describe states that belong to the
category designated by emotion , but since emotion is a noun, it is
not substitutable for happy and sad .
ANTONYMY/CONTRAST. While perfect synonymy involves words
that always refer to the same thing, logical incompatibility
involves words that never overlap in reference. Tis logical rela-
tion contributes to two lexical-semantic relations, antonymy and
contrast.
Binary opposition seems to have special status in language
and conceptualization since even where more than one incom-
patible alternative is available, binary relations may be dis-
cerned. For example, although there are many emotions ( happy,
sad, angry, afraid, disgusted ), happy is generally understood as
having a single antonym, sad .
Logical approaches to antonymy distinguish various types.
Complementary (or contradictory ) antonyms perfectly bisect
a semantic domain. For example, in relation to electrical
items, if something is not on , then it is necessarily of and vice
versa . Contrary antonyms designate the extremes of a scale.
For instance, something that is large is necessarily not small ,
but something that is not small is not necessarily large . (Some
authors, including Lyons 1977 and Cruse 1986 , restrict the term
antonym to just contraries.) Converse antonyms indicate difer-
ent perspectives on the same relationship for example send/
receive , teacher/student, north/south . So, if X is north of Y, then Y
is south of X. Other types of antonymy have received some atten-
tion, but these tend not to show the same kinds of logical rela-
tions as those mentioned earlier. For example, kinship terms can
be opposed on the basis of gender ( brother/sister ) or generation
( mother/daughter ) .
In the broad sense of the term, antonymy is often defned as
a relationship of minimal diference ; that is, antonymous words
share most of their semantic content, but for one diference that
makes the two terms incompatible. On such a defnition, mother
and daughter are opposites because they only difer in the gen-
eration they refer to, while mother and son are not opposites
because they difer in both gender and generation.
As mentioned, the lexical relation antonymy is sometimes
contrasted to the semantic relation opposition , and antonym
pairs present the best evidence in favor of the position that rela-
tions between lexical items, not just senses, are represented in
the mental lexicon. Antonym pairings are particularly strong in
experiments such as word association tasks and lexical prim-
ing , and antonyms co-occur frequently in text leading some
(e.g., Jones 2002 ) to question whether antonymy is also a syntag-
matic relation.
Larger sets of incompatible items exist, for example solid/
liquid/gas , but more linguistic-semantic attention is usually paid
to the not-necessarily-incompatible relation of co-hyponymy
or contrast , which, along with hyponymy, is basic to the taxo-
nomic organization at the basis of semantic feld and network
approaches . Such relations can also be defned in terms of min-
imal diference. For example, the basic color terms are similar
Lexical Relations Lexical Semantics
445
are a major force in determining the form of a language and that
lexicon and grammar do not constitute completely separate
types of linguistic knowledge. Tus, such approaches are usually
lexico-centric. 4) Lexical concerns have also been at the forefront
of computational linguistics , in part because lexically
driven approaches to grammar have proved most computation-
ally practicable. Furthermore, the goals of most computational
linguistics programs involve the creation of systems that can use
language meaningfully thus requiring models of how meanings
might be represented and means to acquire and use such seman-
tic information. 5) Meanwhile, advances in computer hardware
and software led to the growth and development of corpus
linguistics , which is particularly suited to the study of words
and their use and has become one of the major methodological
tools of lexical semantics and lexicography.
Tis confuence of diverse motivations and assumptions has
contributed to the variety of approaches to lexical semantics and
to contrary positions on major questions in the feld .
Major Distinguishers of Theoretical Approaches
Any semantic theory must say something about the representa-
tion of lexical meaning, and theories of lexical meaning must ft
into theoretical accounts of the meaning and grammar of larger
constituents. Tus, in a sense, there is no such thing as a free-
standing lexical semantic theory, but instead there are theories
of meaning that pay more or less attention to representation at
the lexical level.
Teories thus difer in the extent to which they take the lexical
or sentential meanings as their starting point. Tose that start at
the lexical level of analysis, for example, Anna Wierzbickas (e.g.,
1996) Natural Semantic Metalanguage approach, are sometimes
criticized for lack of attention to the way that word meanings
combine in order to create sentence meanings . On the other
hand, those that start with complete propositions in mind, for
example, Ray Jackendofs (e.g., 1990 ) Conceptual Semantics , tend
to focus on the aspects of word meaning that interact with each
other in sentential contexts (particularly the relation between
predicate and argument ), but not with more detailed
nuances of meaning, as would interest a lexicographer .
One of the most basic issues for a lexical semantic approach
is the issue of whether lexical meanings should be distinguished
from concepts . Tat is to say, is the meaning of a content word,
like apple , diferent from our conceptualization of the category
apple? I s knowing about apples diferent from knowing the
meaning of apple ? Tose theorists who think that lexical meaning
is diferent from conceptualization generally make the distinc-
tion between defnitional and encyclopedic aspects of meaning.
On this view, only defnitional meaning that which is suf cient
to identify a referent and allow for grammatical interpretation of
the sentential context is relevant to lexical semantics. To use
a simple example, the defnition of girl would be young female
human. Encyclopedic information, on the other hand, includes
other information that comes from our experience of the things
and situations referred to by words. For girl , this might include
information such as may wear pigtails and associated with
the color pink. Jerrold J. Katz and Jerry A. Fodors (1963) desid-
erata for a theory of linguistic semantics takes the position that
encyclopedic information should not be represented as part of
verbs, adjectives), rather than grammatical words (conjunctions,
determiners), which are more usually studied in the context of
sentential (or propositional) semantics. Lexical semantics can
also refer to the semantics of non-word lexical items, such as
idioms . Te meaning aspect of lexical semantics most often
refers to denotative sense in particular that is, determining
what such words can and cannot refer to, as opposed to their
connotation or social import. Some of the main issues that con-
cern lexical semanticists are the following:
How is the meaning of a word best represented in a model of
the mental lexicon?
Are diferent types of representation required for diferent
kinds of meaning?
How should multiple interpretations of a single word be
described and explained?
How are diferent words meanings related to one another?
Lexical Semantics in Linguistic Theory: Historical Context
Word meaning is a long-standing area of interest in philosophy
and, of course, lexicography , but attention to it in theoretical
linguistics has varied by time and place. In recent decades, lex-
ical semantics (and lexicology more generally) has experienced
revitalization after a slow period in the early to midtwentieth
century. For instance, Noam Chomsky ( 1965 , 84) described the
lexicon as simply an unordered list of all lexical formatives,
and Leonard Bloomfeld claimed that we have no way of defn-
ing most meanings and therefore the linguist cannot defne
meanings, but must appeal for this to students of other sciences
(1933, 1446). Of course, infuential lexical semantic work was
pursued in this period, but much of the work in the generative
tradition (e.g., Katz and Fodor 1963 and generative seman-
tics in the 1970s) ran into problems of internal consistency or
explanatory insuf ciency. Other semantic work was pursued in
Europe by practitioners of structuralism and functional
linguistics or by philosophers of language.
Nowadays, the lexicon is central to most major theories of
language. Tere are several reasons (presented here in no par-
ticular order) for the (re)emergence of the lexicon and lexical
semantics in linguistic study: 1) Psychological experimentation
in the 1970s (particularly by Eleanor Rosch) provided evidence
that word meaning is represented in the mind quite diferently
from the way that it is represented on a dictionary page that
some lexical meanings might be based on prototype repre-
sentations. Since the goals of general linguistics were, by this
time, mostly concerned with the mental representation of
language, such evidence needed to be integrated with linguistic
theory more generally. 2) Since the 1980s, theories of grammar
have become more lexically driven (e.g., head-driven phrase
structure grammar, lexical-functional grammar ,
and minimalism ). In such theories, the main constraints on
sentence structure come from the syntactic and semantic
requirements of the lexemes in the sentence, and lexical struc-
tures and rules account for the types of things that transformations
did in earlier Chomskyan approaches (cf. transformational
grammar ) . 3) Around the same time, the group of theoreti-
cal approaches called cognitive linguistics was emerging.
Cognitive linguistic theories hold that semantic considerations
Lexical Semantics
446
Chomsky , Noam . 1965 . Aspects of the Teory of Syntax. Cambridge,
MA : MIT Press .
Cruse , D. A. 1986 . Lexical Semantics . Cambridge : Cambridge University
Press .
Fellbaum , Christiane , ed. 1998 . WordNet: An Electronic Lexical Database .
Cambridge, MA : MIT Press .
Jackendof , Ray . 1990 . Semantic Structures . Cambridge, MA : MIT Press .
Katz , Jerrold J. , and Jerry A. Fodor . 1963 . Te structure of a semantic
theory . Language 39 : 170 210.
Pustejovsky , James . 1995 . Te Generative Lexicon . Cambridge, MA : MIT
Press .
Ravin , Yael , and Claudia Leacock , eds. 2000 . Polysemy: Teoretical and
Computational Approaches . Oxford : Oxford University Press .
Wierzbicka , Anna . 1996 . Semantics: Primes and Universals . Oxford : Oxford
University Press .
LEXICOGRAPHY
While lexicography is often thought of as a subfeld of linguis-
tics, it is a scholarly discipline in its own right with its own prin-
ciples and practices. Tis discipline is divided into two subfelds,
practical lexicography and theoretical lexicography. Practical
lexicography is concerned with compiling, writing, and editing
dictionaries, which serve a double function: as a record of the
vocabulary of the language and as a reference work to meet the
needs of users for information about words, their usage, and
their spelling . Teoretical lexicography is concerned with the
defnition of general principles governing the compilation of
dictionaries .
Dictionaries difer in their selection of vocabulary and other
items that the editors believe merit inclusion, given the size
and purpose of the volume. While most dictionaries use alpha-
betized word lists, certain others such as Rogets Tesaurus of
English Words and Phrases are arranged by topic. It lists words
according to the ideas that they express, for example, abstract
relations (existence, relation, quantity, etc.), space (generally,
dimensions, etc.), and intellect (formation of ideas, communica-
tion of ideas, etc.), among others. Words and phrases are listed in
the main body of a thesaurus according to their word class, but
without a defnition or any information about pronunciation or
etymology.
Dictionaries come in various formats. Besides traditional
print dictionaries, online dictionaries and dictionaries on solid
state media (such as CD or fash memory) have become increas-
ingly popular during the last quarter of the twentieth century
because they facilitate rapid access to information, cross-refer-
encing, and immediate updates with the latest vocabulary.
General-purpose monolingual dictionaries are organized
alphabetically and use the same language for both the object and
the means of description. While all dictionaries aim for compre-
hensiveness, the number and structure of lexical entries depends
upon the target audience, as well as constraints in funding and
time to complete the dictionary. General-purpose dictionaries
focus on the description of a standard language, aim to provide
an exhaustive coverage of the words in a language (abridged
dictionaries focus on somewhat shorter lists), and are typically
more linguistic than encyclopedic.
Dictionaries are usually divided into three parts: an intro-
duction (including instructions on how to use the dictionary),
linguistic (including lexical) meaning. However, many, if not
most, theories of meaning have moved away from this assump-
tion and treat lexical semantics as involving the interface between
language and conceptualization. Such approaches are less likely
to hold that meanings are represented in the mental lexicon (i.e.,
the repository of linguistic information about words), but instead
see lexical meaning as represented in the conceptual realm. In
this case, lexical semantic theories become intertwined with the-
ories regarding the representation of concepts and must explain
a) whether/how lexicalized concepts difer from nonlexicalized
concepts, and b) how the formal aspects of language interact
with the conceptual representations of meaning .
Another issue that has divided lexical semanticists (and lexi-
cologists) is whether word meanings can be defned on their own
or whether lexical meaning is derived (at least in part) through
semantic relations ( lexical relations ) among words. Tat
is, does a words meaning emerge (to some degree) through the
relations between words in a languages lexicon? Lexical feld
theorists and some computational linguists (e.g., the WordNet
project Fellbaum 1998 ) take the position that meaning emerges
from relations, whereas most working within a componential
framework presume that lexical semantic relations should be
explained by a lexical semantic theory, rather than being primi-
tive elements of the theory .
Finally, there is the very big question of how the senses of
words are to be represented in a linguistic model. Te most com-
mon approach is to devise a componential semantic metalan-
guage , which provides a limited and precise vocabulary for
representing elements of meaning and some form of grammar
for combining those elements into more complex meanings. Te
form that such metalanguages take varies considerably among
theories, and some cognitive linguistics theories eschew meta-
languages as such in favor of representations (for example, the
image schema ) with more visual-schematic elements than lin-
guistic ones. What all of these approaches have in common is the
aim to represent meaning using a restricted set of meaning ele-
ments. Tat is, meanings are composed from smaller meaning-
ful parts (often semantic primitives [primes] ) , and the set of
meaningful parts available to a semantic theory is smaller than
the set of lexical items that could be described by such a repre-
sentational system (cf. language of thought ) .
Conclusion Lexical Semantics Today
Te range of lexical semantic research today is extremely varied
in the topics studied, the methodologies applied, and the theo-
retical assumptions behind them. Unlike some other subdisci-
plines of linguistics, no particular theoretical approach can be
said to be the clear leader in the feld. Te advent and develop-
ment of corpus linguistics means that much lexical work is now
based on empirical rather than just introspective evidence,
and continuing developments in that feld strengthen the value
of that evidence .
M. Lynne Murphy
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bloomfeld , Leonard. 1933 . Language . New York : Holt, Rinehart, and
Winston .
Lexical Semantics Lexicography
447
dictionary to get an idea of the microstructure of lexical entries,
reviewers generally conduct an arbitrary sampling of diction-
ary entries. Depending on the time constraints, this procedure
may, for example, lead the reviewer to scrutinize every 10th main
entry on every 20th page for completeness, clearness, accuracy,
simplicity, and modernity. To ensure that the review is based on
a representative sample, it is necessary to check that the difer-
ent parts of speech are adequately represented, that polysemy is
taken into account, and that there is a balance between words
from both the general and specifc domains. In reviewing dic-
tionaries and devising methods for improving the structure of
dictionaries, the reviewer notes the presentation of the text on
the page, as it plays a signifcant role in infuencing the acces-
sibility of the content. Te range of vocabulary is also important
since users typically expect dictionaries to ofer the latest words
from the domains of fashion, technology, and business, among
others, while at the same time including variants. Polysemy, the
structure of defnitions, usage notes, examples, and etymologi-
cal information are also important criteria for the evaluation of a
dictionarys content .
Hans C. Boas
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Atkins , Beryl T. S. , and A. Zampolli . 1994 . Computational Approaches to
the Lexicon . Oxford : Oxford University Press .
Bjoint , Henri . 2000 . Modern Lexicography: An Introduction .
Oxford : Oxford University Press .
Hartmann , R. R. K. , and G. James . 1998 . Dictionary of Lexicography .
London and New York : Routledge .
Zgusta , Ladislav . 1971 . Manual of Lexicography . Te Hague : Mouton .
LINGUISTIC RELATIVISM
Linguistic relativism refers to the idea that language infuences
thought and worldview (see also language of thought ). In
essence, thinking and worldview are relative to the language one
learns to speak in childhood. Language, thought, worldview, and
infuence are thus key concepts in linguistic relativism.
Various understandings of language have led to propos-
als for several levels of relativism, from semiotic relativity when
referring to the general faculty of language, to structural relativ-
ity when referring to the grammatical properties of languages,
and to functional relativity when referring to communicational
patterns of interaction within and across languages (Lucy 1997;
Hymes 1966 ). Most past and current research has concentrated
on structural relativity.
Several levels of thought have been posited as potentially
under the infuence of language, including the neurologi-
cal, cognitive, and propositional levels. To date, little work has
investigated neurological variation across speakers of diferent
languages (Gilbert et al. 2006 ). Most work addresses conceptu-
alization by examining cognitive processes, such as memory ,
categorization , inference, analogy , and emotions (Lucy
1992 ; Levinson 2003 ; Boroditsky, Schmidt, and Phillips 2003 ).
Very few studies, if any, have proposed that language may infu-
ence the actual content, or propositional level, of thought. Tis
idea, known as linguistic determinism, is commonly discredited
as scientifcally untenable.
the body (the alphabetically ordered list of entries), and appen-
dices (other information, such as weights and measures, punc-
tuation, etc.). Te arrangement of the entries of a dictionary
is referred to as its macrostructure. Dictionaries difer in the
placement of homonyms, derived words, compounds, and
phrases, which can be given independent entries or included
in an entry. Te layout and organization of the individual entry
is referred to as the microstructure of the dictionary. Each dic-
tionary difers in its conventions for structuring lexical entries.
Typically, the headword at the beginning of an entry is in bold
and indented by a few spaces. Bold or italic typeface may be
used to mark the part of speech at the beginning of a lexical
entry. Some dictionaries include the standard pronunciation of
headwords and spelling variants. When a word is polysemous,
its senses are often numbered, with the most frequently occur-
ring sense frst. Similarly, when a sense or a group of senses
belong to a diferent word class or subclass, the sense(s) are
labeled accordingly, in combination with defnitions, exam-
ples, and usage notes. More technical, archaic, or obsolete
senses and idiomatic phrases usually appear toward the end of
the lexical entry.
Te headword of a lexical entry consists of a lemma (the basic
word form) that conventionally represents all of the infected
forms of the unit. Following the headword, lexical entries pro-
vide a defnition of it in the form of a paraphrase (in the same
language), which is semantically equivalent to it. Lexicographers
aim to ofer defnitions that are simpler than the word itself.
Another goal is to avoid circularity of defnitions, that is, to not
defne two or more lexemes in terms of each other. Te most
common type of defnition is the analytic defnition since it
aims at maximal inclusion and independence from the context.
Synonyms or antonyms are sometimes used as alternatives to
analytic defnitions because they are short. However, they may
require the dictionary user to look up other defnitions to under-
stand their meanings. Depending on the scope and aim of the
dictionary, lexical entries also provide examples that illustrate
the headwords syntactic behavior or ofer additional semantic
information .
Metalexicography has the goal to improve the structure
and content of dictionaries. One way of achieving this goal is
to be involved in research about lexicography. Tere are sev-
eral professional organizations and academic journals devoted
to metalexicography. Another way is to suggest methods and
criteria for reviewing and evaluating dictionaries. Evaluating
and assessing dictionaries, also known as dictionary criticism,
is dif cult because it is not always clear what types of criteria
should be applied. One solution to this problem has been to
take large sets of dictionary reviews and determine the range
of criteria applied by diferent reviewers in their evaluations.
Tese include reversibility, alphabetization, directionality,
coverage, reliability, currency, redundancy, retrievability, and
equivalents .
Another dif culty of dictionary criticism is the large number
of entries. Reviewers analyze only a small sample of entries or
focus on particular features of dictionaries. To begin with, they
typically focus on the preface of the dictionary that explains how
to use it, who the intended audience is, and what types of infor-
mation are included in the main body. After fipping through the
Lexicography Linguistic Relativism
448
of the code and sociocultural emphases in the cross-linguistic
study of literacy learning.
Te code approach is the most prominent research frame-
work because of its focus on the universal and language-spe-
cifc features that can explain the cognitive-linguistic processes
underlying
decoding mastering a languages written code relative to its
spoken language units. For English, this means that children
must become aware of how letter patterns (graphemes) cor-
respond to the smallest segment of their spoken language, the
phoneme , as the means for achieving automatic and fuent
word-level recognition.
comprehension deriving an overall interpretation of an
authors intended meanings as actively constructed through
interactions with the textual medium.
composition the generation and organization of ones own
ideas as expressed through interactions with the textual
medium.
A contentious debate concerns whether profcient word-decod-
ing abilities must occur before reading comprehension skills can
develop (known as the simple view of reading) (Vellutino et al.
2007 ) or whether reading comprehension develops concurrently
with general spoken language comprehension (Cain and Oakhill
2007 ) .
A second perspective integrates facets of the code and social-
ization frameworks in highlighting purposes and types of literacy,
especially for alphabetic knowledge. Te basic level of literacy is
alphabetic and functional. Individuals who break the alphabetic
code are able to negotiate daily activities that involve recogniz-
ing and accessing known meanings from their spoken language,
such as reading street signs or writing familiar food items for a
grocery list. Functional literacy is inadequate for meeting current
standards in either educational or workplace contexts. In con-
trast, critical literacy stresses profciency. Individuals must be
capable of using literacy tools competently for learning how to
learn. Profciency includes knowing how to analyze critical link-
ages among ones prior knowledge the meaning or signifcance
of a read or written text relative to perspectives expressed and,
at the highest level, integrating this information with other texts
as the process for generating new questions. Tis ability to draw
on and contrast multiple sources of information to formulate
new understandings entails intertexuality .
A third stance broadens the concept of literacy profciency
from the traditional code and socialization emphases to mul-
tiple literacies. Tis construct, rooted in the profound sociocul-
tural changes in communication brought about by the digital
age, encompasses computer literacy, information literacy, and
digital media literacy as components .
Tese three literacy frameworks are not mutually exclusive.
Moreover, notions of being literate and their associated stan-
dards will continue to evolve as outcomes of sociocultural inter-
actions with new technologies.
Since literacy knowledge originates from spoken language
knowledge, the study of literacy crosses multiple disciplines and
subareas. Language studies range from language-learning
environment ; theory of mind and language acqui-
sition; word meaning; the mental lexicon; and acquisition
Concerning the relationship between language and thought,
language has variably been suggested to infuence, impact,
shape, mould, condition, limit, or channel thinking.
An important issue, then, concerns the scope of these efects.
Current research has been asking whether language efects are
on line in the process of producing and comprehending lan-
guage (see Slobin 1996 on thinking for speaking) or whether
language efects pervade human cognition (i.e., efects exist
whether or not one is engaged in linguistic acts). Most research
has assumed the latter type of language efects on thought.
Finally, note that linguistic relativity was originally formulated
as a scientifc principle by Benjamin Lee Whorf in 1940 (1956,
214, 221). Te principle has since then been relabeled the Sapir-
Whorf hypothesis , following an article by Harry Hoijer (1954)
referring to Edward Sapir and Whorf, who contributed to the
early development of linguistic relativity in the 1920s and1930s
(Sapir 1985 ; Whorf 1956 ) .
Stphanie Pourcel
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Boroditsky , Lera , Lauren Schmidt , and Webb Phillips . 2003 . Sex, syntax
and semantics. In Language in Mind , ed. Dedre Gentner and Susan
Goldin-Meadow , 6179. Cambridge, MA : MIT Press .
Gilbert , Aubrey , Terry Regier , Paul Kay , and Richard Ivry . 2006 . Whorf
hypothesis is supported in the right visual feld but not the left .
Proceedings of the National Academy of Sciences 103 : 489 94.
Hoijer , Harry . 1954 . Te Sapir-Whorf hypothesis. In Language in
Culture , ed. Harry Hoijer , 92105. Chicago : University of Chicago
Press .
Hymes , Dell . 1966 . Two types of linguistic relativity. In Sociolinguistics ,
ed. William Bright , 11457. Te Hague : Mouton .
Levinson , Stephen . 2003 . Space in Language and Cognition: Explorations
in Cognitive Diversity . Cambridge : Cambridge University Press .
Lucy , John . 1992 . Grammatical Categories and Cognition .
Cambridge : Cambridge University Press .
. 1997 . Linguistic Relativity . Annual Review of Anthropology
26 : 291 312.
Sapir , Edward . 1985 . Selected Writings in Language, Culture and
Personality . Berkeley : University of California Press .
Slobin , Dan . 1996 . From thought and language to thinking for speak-
ing. In Rethinking Linguistic Relativity , ed. John Gumperz and
Stephen Levinson , 7096. Cambridge : Cambridge University Press .
Whorf , Benjamin Lee . 1956 . Language, Tought, and Reality . Cambridge,
MA : MIT Press .
LITERACY
Prominent Research Frameworks
Written symbolic codes of languages may be alphabetic, syl-
labic , morphosyllabic, (Perfetti 2003 ; Ho et al. 2007), or alpha-
syllabic (Mishra and Stainthrope 2007 ). In each case, children
must frst learn how the written code of their language embodies
spoken language units (Perfetti 2003 , 17). Children must also
have language socialization experiences that promote thinking
and talking in more literate ways if they are to achieve academic
language profciency (Wilkinson and Silliman 2000 ) or linguis-
tic literacy (Ravid and Tolchinsky 2002 ). Contemporary research
frameworks and their related studies refect diferent aspects
Literacy
449
their interactions for profcient literacy learning remain uniden-
tifed (Cain and Oakhill 2007 ) .
Current State of Research
OVERVIEW OF THEORY. Contemporary research on literacy has
been catalyzed by two general theoretical traditions on the
human capacity for knowledge: Sociocultural science and cog-
nitive science . Each involves an efort to build a comprehensive
and coherent account of human knowledge capacity, but difer
in their view of how knowledge accrues to individuals.
Te sociocultural tradition, which encompasses the language
socialization framework, counts its origins in American prag-
matism , such as that of William James and John Dewey . Te
key concept is that human knowledge is embedded in the social
and physical context; socioculturalists view the individual within
a specifc social context as the fundamental unit of analysis for
studying human learning and development . Diferent strands of
the sociocultural tradition emphasize alternate ways of social/
cultural analysis (e.g., interpersonal exchanges versus broad cul-
tural patterns versus local sociopolitical power hierarchies). Two
of these variations, social constructivism and participation/prac-
tice theory, have played signifcant roles in language and literacy
research. Sociolinguistic approaches have been prominent in
academic discourse studies, whereas critical theory, which advo-
cates for social change and empowerment, has played a major
role in literacy studies.
Te cognitive science tradition traces its origins to the nine-
teenth-century studies of individual diferences in perceptual
processing (e.g., Wilhelm Wundt). Modern-day approaches,
however, highlight the mechanisms by which people process
and integrate information, and the unit of analysis tends to be the
individual. Te cognitive science tradition also represents a vari-
ety of conceptual approaches sharing the premise that the indi-
viduals real-time information processing should be the focus
of inquiry. Diferent approaches vary from a strong nativist (see
innateness and innatism ) and representational perspec-
tive to the primacy of the emergence of knowledge from system
experience interactions, as stated by the connectionists . For
example, neuroimaging studies range from information process-
ing, which emphasizes the processing constraints and multilevel
integration of information, to connectionism, which investigates
parallel processing and the extraction of inherent regularities
from the input .
ELABORATION OF APPROACHES. Te cognitive science approach
in which the code framework is embedded has been most infu-
ential in revealing a) the precursory phonological awareness
knowledge needed for beginning reading across alphabetic
and nonalphabetic languages and necessary language-specifc
knowledge (e.g., how the consonant cluster patterns of spoken
Czech infuence phonemic awareness development) (Caravolas
and Bruck 1993 ); b) the instructional design and content that
best facilitates beginning reading in struggling readers; and
c) neurobiological signatures of dyslexia. Te focus in instructional
studies is on experimental investigations as the scientifc basis for
determining the treatment validity of instruction to prevent read-
ing problems in grades 1 to 3. Tese studies, conducted primarily
in the United States, employ randomized controlled feld trials,
of syntax (see syntax , acquisition of), to language varia-
tion and second language acquisition . Literacy learning
processes are also subjects of developmental language study
from difering viewpoints. Subareas include phonological
awareness, reading , composition, spelling , distinctions
between the spoken and written communication modes, and the
efectiveness of teaching reading and teaching writing
in educational programs that, internationally, span alphabetic
and nonalphabetic languages. Furthermore, literacy research
has expanded to incorporate neuroimaging in order to iden-
tify neurobiological correlates of dyslexia and efects on brain
function of scientifcally based reading (SBR) interventions.
Behavioral studies have examined associations among oral lan-
guage impairment, dyslexia, and text comprehension and related
disorders of reading and writing (for reviews, see Cain
and Oakhill 2007 ; Scarborough 2005 ) .
Modern History
Research on literacy learning is relatively new, initiated in
the 1970s. Since its inception, one major principle has guided
this work: Children should have signifcant home and school
opportunities for the integration of oral and written language
experiences. Tese experiences support the development of lit-
erate stances in comprehension, speaking, reading, and writing.
Becoming literate is a social process, infuenced largely by chil-
drens search for meaning. Prior to school-based reading and
writing, childrens engagement in literacy-like actions in play,
such as scribbling on paper with crayons, and with adults, for
example, storybook reading, forms the foundation for later liter-
acy learning.
Initial research on literacy was strongly shaped by studies of
classroom language. Nearly four decades ago, sociolinguists
launched a new direction for inquiry into language and literacy
learning, focusing on oral language use in classrooms. Te frst
research concentrated on language functions, the communi-
cative demands of classrooms, individual diferences, and the
social basis and social integration necessary for learning. Initial
reading studies addressed assessments of reading comprehen-
sion, whereas current studies emphasize efective instructional
models of decoding and reading comprehension. In the United
States with the 2001 passage of the No Child Left Behind Act
(NCLB), federal educational policies for the frst time exerted a
profound infuence on the way that beginning reading is taught.
Te expectation was that explicit SBR instruction would guide
reading practices and curriculum development in phonemic
awareness, decoding, and fuency .
A broader array of language-related features is implicated in
more literate spoken and written language uses beyond phone-
mic awareness, however. Tese include the scope and density of
vocabulary knowledge, command of more advanced syntactic
constructions applied to diverse reading and writing purposes,
and familiarity with a variety of narrative and expository dia-
logue structures and their organization. Non-language factors
also contribute, such as working memory capacity for difer-
ent kinds of information-processing demands, the motivation to
learn, inferencing and integration, and metacognitive strategies
for the self-regulation of literacy learning. Te causal relation-
ship of these language and non-language-related variables and
Literacy
450
Cain , Kate , and Jane Oakhill , eds. 2007 . Childrens Comprehension
Problems in Oral and Written Language: A Cognitive Perspective . New
York : Guilford .
Caravolas , Markka , and Margaret Bruck . 1993 . Te efect of oral and
written language input on childrens phonological awareness: A cross-
linguistic study . Journal of Experimental Child Psychology 55 : 1 30.
Ho , Connie , David W. Chan , Kevin H. Chung , Suk-Han Lee , and Suk-Man
Tsang . 2007 . In search of subtypes of Chinese developmental dys-
lexia . Journal of Experimental Child Psychology 97 : 6183.
Mishra , Ranjita , and Rhona Stainthrope . 2007 . Te relationship between
phonological awareness and word reading accuracy in Oriya and
English: A study of Oriya-speaking ffth-graders . Journal of Research
in Reading 30 : 23 37.
Perfetti , Charles. A. 2003 . Te universal grammar of reading. Scientifc
Studies of Reading 7 : 3 24.
Ravid , Dorit ., and Liliana Tolchinsky . 2002 . Developing linguistic liter-
acy: A comprehensive model . Journal of Child Language 29 : 417 47.
Scarborough , Hollis S. 2005 . Developmental relationships between lan-
guage and reading . In Te Connections between Language and Reading
Disabilities , ed. H. Catts and A. Kamhi , 324. Mahwah, NJ : Lawrence
Erlbaum.
Silliman , Elaine R. , Louise C. Wilkinson , and Maria R. Brea-Spahn .
2004 . Policy and practice imperatives for language and literacy
learning: Who shall be left behind? In Handbook on Language and
Literacy: Development and disorders , ed. C. Stone , Elaine R. Silliman ,
B. Ehren , and K. Apel, 97129. New York : Guilford .
Vellutino , Frank R. , William E. Tunmer , James J. Jaccard , and RuSan
Chen . 2007 . Components of reading ability: Multivariate evidence for
a convergent skills model of reading development. Scientifc Studies
of Reading 11 : 3 32.
Wilkinson , L.C. , and E. R. Silliman 2000 . Classroom language and lit-
eracy learning. In Handbook of Reading Research. Vol. 3 Ed. M. Kamil ,
P. Mosenthal , P. Pearson , and R. Barr , 33760. Mahwah, NJ : Lawrence
Erlbaum .
LITERARINESS
Tis term refers to the perceived distinctive quality of the lan-
guage of literary, as opposed to nonliterary, texts. If the linguis-
tic study of literature attempts to understand how linguistic
form is adapted to literary purposes, then identifying a text as
literary based on its language is one of the central problems.
Tere is currently among scholars of literature, both linguists
and literary critics alike, little agreement about the status of lit-
erature as language. Despite this lack of consensus, or maybe
because of it, the investigation of literature using linguistic
models has become a productive feld within applied linguis-
tics. Te question of literariness is closely tied to the modern
concept of literature and its history, and so it seems best to pro-
ceed by examining frst the history of the concept of literature
before turning to central issues that the question of literariness
has raised.
Although most cultures consider verbal art a separate,
recognizable class of speech, its rendering into print has sub-
jected it to the transforming efects that mark the infuence
of print on every aspect of modern culture (see print cul-
ture ). Literature (from Latin littera letter) in its restric-
tive, modern English sense of imaginative writing in the main
genres of poetry, prose, and drama which has claim to con-
sideration on the ground of beauty of form or emotional efect
(Oxford English Dictionary), arose in the nineteenth century in
often using a response to intervention model to determine the
ef cacy of outcomes in alphabetic reading. While diferences exist
in the form of response to intervention designs, all involve a hier-
archical process of alphabetic instruction and ongoing reading-re-
lated assessments. Minimal responsiveness may mean that a child
requires special education support to be successful.
Te cognitive-experimentalist approach is not without criti-
cism, particularly as this research is refected in the NCLB goal
that all grade 3 children will read profciently by 2014. One cri-
tique pertains to individual diferences. Given the individual
diversity in neurobiological makeup and sociocultural experi-
ences, it is not possible to erase normal variation in the distri-
bution of reading ability. Instead, normal variation should be
treated as an asset to build upon and not as a liability (Berninger
and Richards 2002 ) .
In contrast, sociocultural approaches converge on the belief
that literacy learning does not consist exclusively of recruiting
neurobiological and cognitive events inside the learners head.
A signifcant question is how academic discourse serves as the
social mechanism for children to learn how to do literacy
and advance their language learning as members of the larger
school literacy culture. Study designs are typically descrip-
tive or quasi-experimental. A limitation of the sociocultural
tradition is that descriptive studies function best to generate
new hypotheses about causal mechanisms but cannot yield
broader generalizations unlike randomized controlled trials.
However, future literacy research may lead to mixed-methods
approaches that combine the tools of the cognitive and socio-
cultural sciences .
AN EXAMPLE OF INDIVIDUAL DIFFERENCES. Literacy research in
both the cognitive and sociocultural sciences examines group
diferences. However, classrooms consist of particularized difer-
ences. Te translation of group-diferences research into every-
day practices that meet the learning needs of individual children
is far from an easy task. In any third grade in the United States,
some children are still struggling with fuent decoding and spell-
ing; others have no problems with decoding but face signifcant
comprehension barriers when reading expository texts; and still
others may be exerting great efort to unravel the complexity of
academic discourse demands, which then impedes their lan-
guage and literacy learning. No uniform set of reasons accounts
for these individual patterns. Some patterns may be grounded
primarily in sociocultural experience, such as less familiarity
with academic discourse; other patterns may involve complex
interactions of neurobiological, cognitive, social, linguistic, and
communicative factors.
While much has been learned about literacy processes, two
signifcant educational challenges persist: understanding varia-
tions in individual profles and how to craft evidence-based prac-
tices that will assist individual children to become full members
of their larger literacy communities .
Elaine R. Silliman and Louise C. Wilkinson
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Berninger , Virginia W. , and Todd L. Richards , 2002 . Brain Literacy for
Educators and Psychologists . San Diego, CA : Academic Press .
Literacy Literariness
451
important element in the organicist metaphors of literary struc-
ture developed principally by Robert Penn Warren and other
so-called New Critics in the United States, though by no means
unique to them. Organicist approaches see the form of a text as
highly responsive to its meaning, and the New Critics especially
valued the ability of a text to reconcile within itself the various
strands of meaning that its language evokes. In semiotic terms,
an icon is a sign in which the signifer somehow resembles its
referent. Iconicity operates to a limited extent in language, for
example, in onomatopoeia, where a word sounds like the thing
it represents, as is often the case for linguistic representations
of animal sounds English cows moo and sheep baa (but
see Haiman 1985 for arguments that iconic signs are widely
exploited in language). In poetry, sounds sometimes mimic the
thing being described, as in these lines from Tennysons Te
Princess :
Te moan of doves in immemorial elms
And murmuring of innumerable bees (7.2212)
Te repeated nasal consonants suggest the hum of bees on a
summer afternoon.
Iconicity often contributes to thematization, the acting out
by a texts words or other speech elements of a particular theme
given semantically in the text. Te speaker of George Herberts
Deniall, for example, laments that his prayers are not being
heard by God:
When my devotions could not pierce
Ty silent eares;
Ten was my heart broken, as was my verse:
My breast was full of fears
And disorder:
Te fnal line of the stanza does not rhyme, nor does it match the
iambic rhythm established in the frst four lines. Te disorder
spoken of in the line is acted out by the lines lack of sonic ft,
implying in the process a parallel between the form of the prayer
and its failure to penetrate. In the fnal line of the fnal stanza the
prayer at last comes around:
O cheer and tune my heartlesse breast,
Deferre no time;
Tat so thy favours granting my request,
Tey and my minde may chime,
And mend my rime.
When the speakers mind is aligned with Gods wishes, his prayer
also becomes formally complete. Iconicity is one way that pat-
terns of linguistic elements can contribute to larger patterns
of meaning. Poetic iconicity is a local phenomenon, however,
which depends on the immediate environment of the signs
involved. Te phrase and disorder is not without rhythm, nor is
it unrhymable, but in the context Herbert has created, the phrase
stands out for its ill ft. Organicist approaches value the extent to
which the local patterns of signifcance can be reconciled to one
another, making each poem a coherent emotional and semantic
whole. In the fnal stanza of Deniall, the local pattern of end
rhyme acts out an additive semantic logic, so that the notions
of chime and time together constitute the meaning of the word
rime , and at the same time the words function sonically within
conjunction with the increasing availability of authorship as a
profession. Before the advent of movable type printing, to write
or copy a book required a great investment of time and energy,
and so only highly valued items would have been widely circu-
lated. Even after print technology began slowly to difuse into
the wider culture, it was never doubted that only important
works would ever be printed, distributed, and saved. Te rise of
industry and the middle class, with increased literacy rates and
an expanded market for writing, saw the normative concept of
literature emerge essentially as a means for diferentiating tra-
ditionally sanctioned texts from those of supposedly ephemeral
quality. Te reading and study of the superior texts were pro-
moted in secondary and university curricula as proftable for
cultural improvement and development of national identity
(see nationalism and language ).
Arguments for the suitability of literature for this project
depended for their success on demonstrating that the privileged
texts possessed certain inalienable qualities. Matthew Arnold,
an English school inspector and poet, argued in 1880 that the
superior character of truth and seriousness in the matter and
substance of the best poetry, is inseparable from the superiority
of diction and movement marking its style and manner ([1880]
1988, 416). For Arnold, and the liberal humanism he has come to
represent, the reading of literature functions to stabilize soci-
ety because it educates citizens in supposedly universal sociobe-
havioral norms.
Like Arnold, most infuential theorists in the frst half of the
twentieth century never questioned the status of literature as
an identifable and benefcial form of linguistic behavior. Tey
expected that the literary text could be diferentiated from non-
literary texts by some constellation of intrinsic linguistic char-
acteristics, the discovery of which, Roman Jakobson argued in
1921 when he coined the term literariness , should be the goal of
literary linguistics. But when formalist investigators failed over
time to adduce a convincing set of characteristics necessary and
suf cient for identifying literature, attention turned to the role of
extrinsic factors, such as audience and medium, in establishing
literary distinctiveness. Many literary critics have resolved the
problem by historicizing the concept of literature itself, arguing
that the search for intrinsic features determinate of literariness
cannot be successful because wherever or whenever the cate-
gory of literature arises, it does so from specifc sociocultural
forces that situate readers diferently with regard to the purposes
attributed to the literary texts within the cultural or theoretical
discourse that promotes the concept. Approaches grounded in
linguistics, however, prefer the term verbal art for their subject
because they also question the suf ciency of the traditional con-
cept of literature to account for the variety of genres developed
around the world, by principally oral cultures, that function in lit-
erary ways for the culture in question. Most contemporary study
can thus be characterized as interactionist , refusing to privilege
intrinsic or extrinsic characteristics but understanding textual
genre categorization as a complex process involving interaction
among text, reader, performance situation, and various sociocul-
tural practices .
Some intrinsic characteristics that have been recognized
as occurring across time and languages in texts that become
literary include iconicity and defamiliarization . Iconicity is an
Literariness
452
Strauss , he and his co-authors exhaustively catalog patterns of
linguistic or structural elements within a poem, often with no
attempt to discover motivations for individual patterns except to
observe their interaction as formal patterns. Tey aim to demon-
strate that the whole poem is essentially a tissue of many such
overlapping and interlocking patterns, a complex and indivis-
ible totality where a perpetual interplay of sound and mean-
ing establishes an analogy between the two facets, a relationship
either paronomastic and anagrammatic, or fgurative (occasion-
ally onomatopoeic) (1980, 23).
Objections to internalist theories of literariness include skep-
ticism about any readers being able to detect all or even many
of a poems linguistic patterns, as well as observations that some
passages of nonliterary prose contain as many patterns as poetry.
Indeed, the strongest reactions against internalist theories con-
cern the genre of prose. Internalist theories of literariness often
elevate the importance of poetry because as a genre, it is maxi-
mally distinct from everyday language. As a result, objections
to these theories are often concerned with accounting for the
characteristics of prose genres. Prose generally has observably
fewer sonic foregrounding devices and is more likely than poetry
to make use of devices that also appear in everyday language,
such as irony, metaphor, or repetition. Tere is some evidence
that these devices occur with greater frequency in literary than in
nonliterary prose, and that there are some linguistic forms, such
as free indirect speech, that tend to occur only in literary prose
(see Miall 2006 and narrative ). On the balance, however, liter-
ary prose tends to require some attention to elements of literary
interaction that are extrinsic to the text in order to account for
literariness .
One important class of extrinsic approaches, often termed
reader response theories , focus on the reader as the source of the
distinctiveness of literature. Perhaps the strongest version of this
extrinsic theory is ofered by Stanley Fish ( 1980 ), who describes
an impromptu experiment (later executed to the same ends with
readers from three continents) in which he told undergraduates
that a list of linguists on a classroom chalk board was a poem
and asked them to interpret it, which they had no trouble doing.
Tis demonstrated for Fish that literariness was wholly a func-
tion of prior reader commitment, rather than of anything within
the text. Readers learn interpretive practices from the communi-
ties of which they are members, and so what counts as literary is
what the community has determined to be so. Although Fishs
description of the interpretive community has not generally been
retained, his position on literariness became, until very recently,
the default standard within literary theory. Feminist and postco-
lonialist challenges to the hegemony of Western literary cultural
practice, demonstrating that texts created and valued by a domi-
nant class will be read quite diferently by less privileged classes,
helped establish that a precise mode or history of reception can-
not be inferred from the text itself (Harrison 2005 , 7). Evacuating
the text of any determining role in its own reading allows critics
to explore how the text participates in the various discourses of
power circulating at the time of its writing and reception .
Current theories of literariness that utilize linguistic tools for
analysis are unwilling to locate the determining characteristics
specifcally within the text or the reader, but generally see textual
genre categorization as a complex process involving interaction
the rhyme scheme to act out the metonymy whereby my rime
refers to the poem as a whole .
Te ideological content for the New Critical emphasis on
unity and coherence in literature is illuminated by comparison
to Viktor Shklovskys infuential theory of ostranenie defamiliar-
ization. Shklovsky was one of several scholars and writers who
met informally in Moscow and St. Petersburg. Often referred to
as the Russian Formalists, this group was the frst to mix literary
analysis with an increasing awareness of the status of the liter-
ary text as a linguistic object and, therefore, subject to descrip-
tion by specifcally linguistic tools (see Erlich 1955 ). Shklovsky
argued that the distinguishing characteristic of the literary text
was that it defamiliarized items or events the perception of which
had become automatized by the reader due to familiarity or
repeated exposure. Leo Tolstoy, as Shklovsky points out in sev-
eral examples, makes the familiar seem strange by not naming
the familiar object, like the description in War and Peace of an
opera scene as pieces of painted cardboard, or when the nar-
rator of Shame describes the sequence of actions in a fogging
without using the word ([1917] 1988, 21). Shklovsky called these
and similar techniques of creative distortion prim devices,
and Jakobson went so far as to claim in 1921 that if literary his-
tory wants to become a science, it must recognize the artistic
device as its only concern (quoted in Erlich 1955, 57; see also
foregrounding ).
Te young Jakobson, whose specifcally linguistic theory of
poetics eventually became widely infuential, believed that he
could justify his admiration for the futurist poetry of Velimir
Khlebnikov by explicating in linguistic terms the complex, sug-
gestive, phonemic, and morphemic patterning of poems heavy in
neologisms, like Kuznechik Te Grasshopper (Jakobson 1987,
252). Khlebnikovs poetry, and that of other so-called Russian
Futurists, so clearly eschewed traditional forms of poetic prac-
tice that there was no question of its being considered canoni-
cal. Like Arnold, the Futurists were interested in distinguishing
true literature from the products of the mass marketers, whom
they called traitors, and they also identifed a strong nation-
alistic purpose in reading and writing literature. Unlike Arnold,
however, they saw literature as having a revolutionary, rather
than a stabilizing, purpose (infuenced perhaps by the difer-
ing cultural conditions prevailing under the unstable czarist
regime in Russia and the imperative to govern as a global power
in England).
As Jakobsons poetics matured, he increasingly saw the role
of the device in terms of a Peircean semiotics. Te devices do not
function alone to interrupt the direct awareness of the iden-
tity between sign and object (1987, 378), but the whole text is
revealed as a system of systems of equivalences, that, through
similarities and contrasts at all levels of linguistic organization,
up to and including the arrangement of the entire text, display
the text as primarily interested in the linguistic medium, the
materiality of its linguistic signs. Tis approach resembles the
organicist metaphors of the New Critics in that linguistic form
is motivated by poetic meaning, and devices are valued less for
their interruption of usual relationships of signifcance than for
the surplus of meaning that iconic relationships create.
In Jakobsons best-known literary-critical essays, however,
such as Baudelaires Les Chats, written with Claude Levi-
Literariness
453
( 2001 ) have proposed the expansion of the notion of read-
ing to include the manipulation of poems to deform them
in ways that defamiliarize the poems and prompt new expe-
riences of them, by reversing them (so that they can be read
backwards), rearranging the words or lines, even removing or
replacing parts of speech. Such formally interactive processes
of reading, increasingly built into hypertext, video games, and
other forms of electronic textuality, continue to obscure the
traditional text/reader division that fgured so prominently in
early notions of literariness. Arguments for the literariness of
nontraditional (i.e., nonprint, nonverbal) formats usually rely
on some form of the strong sociocultural construction argu-
ment, yet there are also attempts to canonize some digitally
mediated works over others based on intrinsic characteristics,
suggesting that interactive defnitions of literariness continue
to prevail for the time being .
Claiborne Rice
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Arnold , Matthew . [1880] 1988 . Te study of poetry. In Te Critical
Tradition , ed. David H. Richter, 41116. Boston : Bedford .
Attridge , Derek . 2004 . Te Singularity of Literature . London : Routledge .
Bauman , Richard . 1984 . Verbal Art as Performance . Long Grove,
IL: Waveland .
Erlich , Victor . 1955 . Russian Formalism . Te Hague : Mouton .
Fabb , Nigel . 2002 . Language and Literary Structure . Cambridge :
Cambridge University Press .
Fish , Stanley . 1980 . Is Tere a Text in Tis Class? Te Authority of
Interpretive Communities . Cambridge : Harvard University Press .
Haiman , John . 1985 . Natural Syntax . Cambridge : Cambridge University
Press .
Harrison , Nicholas . 2005 . Who needs an idea of the literary? Paragraph
28 . 2 : 117.
Herbert , George . [1633] 1974 . Deniall. In Te English Poems of George
Herbert , ed. C. A. Patrides , 96. London : J. M. Dent & Sons .
Jakobson , Roman . 1980 . A postscript to the discussion on grammar of
poetry . Diacritics 10.1 : 21 35.
. 1987 . Language in Literature . Ed. Krystyna Pomorska and
Stephen Rudy . Cambridge : Harvard University Press . Tis text collects
Jakobsons writing on literary topics throughout his career.
McGann , Jerome J. , and Lisa Samuels . 2001 . Deformance and interpre-
tation. In Radiant Textuality , 10536. London : Palgrave Macmillan .
Miall , David S. 2006 . Literary Reading: Empirical and Teoretical Studies .
Frankfurt : Peter Lang .
Shklovsky , Viktor. [1917] 1988 . Art as technique. Trans. Lee T. Lemon
and Marion J. Reis . In Modern Criticism and Teory , ed. David Lodge,
1630. London : Longman .
Tompkins , Jane P. , ed. 1980 . Reader-Response Criticism: From Formalism
to Post-Structuralism . Baltimore : Johns Hopkins University Press .
Tough somewhat dated, this remains an excellent introduction to the
range of ideas usually identifed as reader-response.
Tsur , Reuven . 2003 . On the Shore of Nothingness: A Study in Cognitive
Poetics . Exeter : Imprint Academic .
LITERARY CHARACTER AND CHARACTER TYPES
Discussions of character types date back to Aristotle and appear
within numerous brands of literary criticism. Within the context
of the language sciences, however, character types have particu-
lar relevance to two strands of narratology .
among text, reader, and various sociocultural practices. Reuven
Tsurs sophisticated analysis of religious trance poetry, for exam-
ple, identifes not only textual devices that contribute to a hypnotic
reading but also diferent types of cognitive styles of individual
readers, who are psychologically predilicted to react to certain
environmental variables in diferent ways and so read texts dif-
ferently. Nigel Fabb, on the other hand, adopts a modularist
approach, identifying two types of structures in poetry, inherent
and communicated structure. Inherent structure is determinate
and is processed, rather than interpreted, by the appropriate
module (such as syntax or meter). Communicated structure is
implied by textual evidence in the context of the readers knowl-
edge, so that being a sonnet holds of a text by inference, rather
than as an independent fact about the text. Te important thing
for literariness is again that the text provides only some of the
cues necessary for categorizing it as literary. Communicated
structure is dependent on individual learning, so that responses
to any given textual characteristic might readily vary by indi-
vidual. Fabb also argues that literary texts communicate more
about their form than verbal behavior generally does, and so in
a Jakobsonian sense draw more attention to form. Form in this
sense takes on the characteristics of meaning (because implied);
therefore, form is more likely to be ambiguous, indeterminate,
metaphorical, or ironic in literary texts .
Emotional responses to literature have always been impor-
tant for the identifcation of literariness. Indian theories of
rasa , for example, categorized texts by the type of emotion they
prompted (see dhvani and rasa ). Fresh empirical approaches
to the issue of literariness have focused on the role of emotion in
literary response. Miall identifes literariness as a combination
of formal elements in the literary text and the array of responses
these initiate when read (2006, 144). Readers who encounter
foregrounding devices experience them as defamiliarizing, a
cognitive response that has an associated afective dimension.
Subsequent changes in understanding seem to be guided largely
by the feeling evoked by defamiliarization, especially the feeling
of self-modifcation that, Miall argues, accompanies the recon-
textualization of the defamiliarized concept (see literature,
empirical study of ).
In the last decade, literary critics interested in the ethical
dimensions of literary reading have begun to reconsider, to some
degree, the importance of the defamiliarizing efect of literature.
Derek Attridge ( 2004 ) has recuperated the concept of literariness
for literary studies by identifying it with the inventiveness that a
text manifests whenever it is enacted by a reader. For Attridge,
literariness inheres not in some fundamental unchanging core in
the work but in the inventiveness it shows over time, because it
remains open to change and porous to new contexts, continuing
to introduce what is unknown into the known. Readers experi-
ence that inventiveness anew when they perform the work, and
they desire to do justice to it by shifting the norms and habits
they use for dealing with the world .
As literature undergoes remediation into digital for-
mats (and it should be noted that the Russian Formalists were
applying Shklovskys concept of ostranenie to flm analysis
already before 1920), the notion of literariness is undergo-
ing concomitant refnement. Motivated in part by new media
forms of verbal art, Jerome J. McGann and Lisa Samuels
Literariness Literary Character and Character Types
454
or narrative universals , which is modeled in part on the
study of language universals, focuses more on the construction
than the reception of texts. Such approaches tend to treat char-
acter types as narrative components that carry out action neces-
sary for certain prototypical plot structures (see prototype ). At
the same time, writers adopting this approach (e.g., Hogan 2003 )
also consider the emotional efects elicited by diferent forms of
characterization in terms of empathy and identifcation. In addi-
tion, recent work by such writers as Toolan suggests that certain
contemporary grammars might provide better models of char-
acter types than the early generative grammar and trans-
formational grammar upon which the theories of classical
narratology relied. For example, Toolan examines characters in
terms of the meaning-oriented grammar detailed by Michael
Halliday , which considers the diferent types of participants
in actions. Tese participants include a medium (the afected
participant), agent (a participant acting intentionally), force
(an inanimate agent), instrument (participant controlled by the
agent), benefciary , and recipient (see also thematic roles ) .
Karen Renner
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bal , Mieke . [1985] 1997 . Narratology: Introduction to the Teory of
Narrative . Toronto : University of Toronto Press .
Culler , Jonathan . 1975 . Structuralist Poetics: Structuralism, Linguistics,
and the Study of Literature . Ithaca, NY : Cornell University Press .
Contains an excellent survey of structuralist approaches to narrative,
including a section on character, pp. 2308.
Gerrig , Richard J. , and David W. Allbritton . 1990 . Te construction of lit-
erary character: A view from cognitive psychology . Style 24 : 380 92.
Greimas , A. J. [1966] 1983 . Structural Semantics: An Attempt at a
Method . Trans. Daniele McDowell , Ronald Schleifer , and Alan Velie .
Lincoln : University of Nebraska Press .
Herman , David. 1997 . Scripts, sequences, and stories: Elements of a
postclassical narratology . PMLA 112 : 1046 59.
Hogan , Patrick. 2003 . Te Mind and Its Stories: Narrative Universals and
Human Emotion . Cambridge : Cambridge University Press .
Propp , Vladimir. [1928] 1968 . Morphology of the Folktale . Trans. Laurence
Scott . 2 d ed. Austin : University of Texas Press .
Schleifer , Ronald , and Alan Velie . 1987 . Genre and structure: Toward an
actantial typology of narrative genres and modes . Modern Language
Notes 102 : 1122 50.
Schneider , Ralf . 2001 . Toward a cognitive theory of literary charac-
ter: Te dynamics of mental-model construction . Style 35 : 607 40.
Toolan , Michael . [1988] 2001 . Narrative: A Critical Linguistic Introduction .
2d ed. New York : Routledge .
LITERARY UNIVERSALS
In parallel with language universals (see laws of language,
universal grammar, and universals, nongenetic,
among other entries), literary universals are, generally speaking,
patterns and structures exhibited widely by works of literature
across various familiar literary boundaries, whether national,
generic, or historical. Literary universals apply to various domains
besides world literature as a whole: individual regional and
national traditions; discrete literary forms and genres (poetry,
drama, narrative , etc.); separate literary histories and periods;
foundational literary concepts (plot, character [see literary
Te frst strand, which David Herman refers to as classical
narratology (see also generative poetics and narrative,
grammar and ), frst fourished in the 1960s. Tese linguistic
approaches to literature heavily infuenced by Vladimir Propps
Morphology of the Folktale (originally published in 1928 but not
available in translation until some 30 years later), the works of
Algirdas Julien Greimas , and structuralism in general treat
character types as functional units of meaning. Propps seven
classifcations of character types were determined by the roles
that characters commonly occupy in Russian folktales: the hero,
the villain, the helper, the donor (provider of magical agents),
the sought-for-person and her father, the dispatcher, and the
false hero. Building upon Propps model, Greimas claimed that
characters were signifcant for the actantial roles they perform
within a narrative. In Structural Semantics ([1966] 1983, 198),
he described actants as embodying a small number of roles
in the drama of discursive utterances. Greimass three pairs
of actants subject/object, sender/receiver, and helper/oppo-
nent were intended to correspond to grammatical concepts.
Subjects (characters who do the action) and objects (characters
who undergo the action) are clearly related to the equivalently
named sentence constituents, while helper and opponent can be
regarded, in Mieke Bals words, as adverbial adjuncts ([1985]
1997, 201). Te category of sender/receiver proved most prob-
lematic: While it attempted to supplement Propps dispatcher
with Roman Jakobsons distinction between the initiator of a
communication (e.g., a speaker) and the addressee of that com-
munication (the receiver), the relationship between those linguis-
tic concepts and the corresponding character types was unclear.
Critics have since expanded on the theoretical foundations laid
by Propp and Greimas, and such work has provided useful gener-
alizations about characters as structural units of meaning within
narratives (see Schleifer and Velie 1987, for instance, for a typol-
ogy of literary genres based upon Greimass receiver-actant).
Furthermore, Bals use of semantic axes and Michael Toolans
semantic feature analysis ofered ways to account for the less
generalizable details of characterization such as specifc physi-
cal and psychological qualities which were outside the scope of
story grammars .
Within the second strand of narratology, or what Herman
terms postclassical narratology , considerably more attention has
been devoted to narrative reception. Tis change in focus has
occurred in tandem with the rise of cognitive linguistics
in the 1970s and 1980s, which focused more attention on the
mental processes behind grammatical systems; it also undoubt-
edly was infuenced by reader response theories that considered
the impact of a readers experience and expertise, literary and
otherwise, upon their interpretations of a text (see interpre-
tive community and competence and performance,
literary ). Recently, literary characters have been discussed as
mental models or schemas (see story schemas, scripts,
and prototypes ) sets of expectations generated by exem-
plars , personal experience, stereotypes , and literary knowl-
edge which undergo continual modifcation as the reader
receives more information about a character (see Gerrig and
Allbritton 1990 ; Schneider 2001 ) .
Teorists continue to mine the connections between linguis-
tic models and narrative. Work done on literary universals
Literary Character and Character Types Literary Universals
455
three or more, and general or universal literature examined
all literatures as a whole (see Wellek and Warren 1977 , 4653).
Such distinctions correlated with the nineteenth-century evolu-
tionary paradigm: Comparative literature is coined by analogy
with the sciences of comparative zoology, comparative anat-
omy, comparative philology, and so on (see philology and
hermeneutics ) .
Te nature of literature as verbal art, and thus cognitively
founded in language, is not the sole spring of its universality
(see verbal art, evolution and , and verbal art, neu-
ropsychology of) . Te roots of literary universals also lie in
the common stock of anthropological development and social
behavior, or human universals (Brown 1991 ). Art is one such
human universal and, if the prehistoric impulse of making spe-
cial is the ultimate origin of all kinds of artistic production and
aesthetic appreciation (Dissanayake 1992 ), it follows that liter-
ature, as one of the arts, will also exhibit certain universal and
nontrivial patterns. To the extent that it stems from the mysteri-
ous biology of play, literature may re-present any or all of the four
fundamental types of human and biological play described by
Roger Caillois ( 1961 ): mimesis (dress-up and lets pretend/
mimicry and camoufage), alea (games of chance/random varia-
tion), agon (sports and contests/survival of the fttest), and ver-
tigo (swings and slides/fight and chasing). It could also be that
literature somehow recapitulates the Darwinian drama of sur-
vival (Meeker 1997 ), and the analysis of basic plots in world lit-
erature (e.g., Polti [1921] 1977 ) suggests that the same social and
sexual competitions of early human life are repeatedly replicated
in literature. On the other hand, given the dif culty that socio-
biology and evolutionary psychology have in explaining
why the human is so diferent from the rest of the natural world,
more promising ways of sourcing literary universals may lie in
the study of creativity as a universal human phenomenon. Only
the human seems to actualize the original sense of creature, or
the still-becoming-creation.
Terminology
Te basic vocabulary of literary universals stems from paral-
lels in the theory of language universals. Absolute universals
are those that apply to all literatures, past and present. As with
absolute universals of language, these may be few in number
and dif cult to substantiate completely since available informa-
tion about both the languages and literatures of the world is far
from complete. Nevertheless, some absolute literary universals
do appear to exist. One simple absolute universal is that liter-
ature (including oral literature or orature ) occurs in all known
cultures. Whether this is historically monogenetic , like language
in Homo sapiens sapiens may be, or polygenetic , like the inven-
tion of writing systems certainly was, remains an open ques-
tion since oral literature long precedes the written record (see
oral composition ). Another possible absolute is that all
literatures (eventually) develop fundamental generic diferen-
tiations, such as between poetry and prose (see poetic form,
universals of ). Another content-oriented absolute is the uni-
versality of myth (stories of creation, food, etc.) in the earliest
recorded traditions. Often, there are striking parallels between
very specifc elements among even the most areally and genet-
ically distinct myths, which may lend credence to the existence
character and character types ], etc.); and common liter-
ary devices (e.g., metaphor ), among others. Literary universals
may also be present as correlations across these various scalar
domains. Tus, Aristotles observation that tragedies have a
beginning, a middle, and an end applies to literary works more
broadly .
A specifc call for literary universals appeared in Hogan ( 1994 ;
as a subtype of aesthetic universals), but the general concept
can be traced to Goethe, if not Aristotle. Te Aristotelian bina-
ries of universal/particular and substantial/nonsubstantial,
when intersected, generate a four-category ontology (see Lowe
2006 ). Literary universals include both nonsubstantial univer-
sals (abstract properties and relationships) and substantial uni-
versals: literary kinds (genres) and literary morphology , very
generally considered. Examples of the latter include metrical
analysis, biblical form criticism, thematics of creation stories,
Freytags pyramid, Proppian functions, Bakhtinian speech
genres, and so on. Substantial universals implies something
very diferent from the oxymoron concrete universals , some-
times invoked by New Critics and others, to suggest the possibil-
ity that literature can entirely transcend the universal/particular
dichotomy. Even when substantial (i.e., measurable, able to
be cataloged), literary universals work outside of any particular
instantiation or touchstone.
Tis entry briefy discusses fve areas of literary universals: their
rationale and origins, some basic terminology, their relationship
to dominant strands of literary theory, various successful fnd-
ings and limitations, and possible lines of future investigation.
Te focus remains on universals as they apply directly to liter-
ary study, as opposed to universalist models conceived as more
purely linguistic or cognitive, and which merely invoke literary
terms (e.g., story grammar ) . In contrast, literary universals
seek to explore what unifes the incredible richness, beauty, and
diversity of global literatures, past and present.
Rationale and Origins
As with language universals, the key criterion for literary univer-
sals is not that they occur in all known literatures (though this
is possible for absolute universals; see absolute and statis-
tical universals ), but that they are represented, more often
than chance alone would suggest, by literatures that are areally
and genetically distinct (see areal distinctness and litera-
ture ), that is, free from the kinds of relations and infuences that
are to be expected when literatures are linked by literary history
or geography (Hogan 2003 , 1719). Tus, if epic can be called a
universal genre, it is not so because it was written both by Homer
and by Milton, who knew (i.e., read) Homer, but because it is a
literary form also recognizable in the Mahbhrata in India, the
Tale of the Heike in Japan, and the Popul Vuh in South America
all traditions that are areally and genetically unrelated.
One of the earlier precursors of literary universals is the idea
of Goethes world literature or Weltliteratur , by which Goethe
seems to have meant a broad cultural unity whose understand-
ing could also lead to global social progress. Weltliteratur was
one of the inspirations for the discipline of comparative litera-
ture. Refecting its scientifc ambitions, this feld was subdefned
so that comparative literature per se examined the relation-
ships of two national literatures, world literature compared
Literary Universals
456
concrete object of the work of art will elude our grasp (1977, 18).
In short, universal formulas have little purchase in any individ-
ual act of literary criticism. Yet, one page later, the fundamental
need for a universal theory reasserts itself: Like every human
being, each work of literature has its individual characteristics;
but it also shares traits with humanity. [Tus, the] characteriza-
tion [of its individuality] can be accomplished only in universal
terms, on the basis of literary theory (1977, 19). Te universal/
particular paradox of literary studies was already in place at the
felds birth around the turn of the twentieth century, when the
nomothetic/idiographic divisions of the German university were
dominant, and all disciplines were preconceived in the category
of Wissenschaften, sciences.
Linguistic approaches to literature, inaugurated by Ferdinand
de Saussure alongside the birth of modern structural lin-
guistics, also imply universals. Tough an odd literary digression
from his linguistic theory, the mysterious anagrams Saussure
culled from Latin texts might serve to show that literature
involves an equally systematic (and compensatorially non -ar-
bitrary?) selection of signs (even if authorially unintended; see
Starobinski 1979 ). In any case, the literary situation is an instance
of language, and thus the linguistic universals of the moment of
communication will apply and will also be available for artis-
tic exploitation. Roman Jakobson (1960) famously delineated
how each of six components of the communicative situation
(addresser, message, contact, etc.) could be exploited for difer-
ent linguistic and literary purposes (emotive, poetic, referential,
etc.), with diferent purposes predictably dominant in diferent
literary genres. Among other brilliant readings, Jakobsons per-
spicacious exploration of the neuropsychology of the metaphoric
and metonymic poles of linguistic competence in relation to
the divergent styles of various Russian novelists (1956) also sug-
gests that literary universals ( pace Wellek and Warren) can illu-
minate seemingly idiographic cases in literary criticism.
Te coming of various posts in late-twentieth-century lit-
erary theory (poststructuralism, postmodernism, postcolonial-
ism) has added new complexity and urgency to the search for
universals in literature (see Carroll 1995 ). Even while prioritiz-
ing irreducible diferences, polyvalent identities, and the local
particulars of each text and reader, current literary theory none-
theless proceeds by placing literature in the context of grand
sociopolitical, economic, or linguistic structures .
Findings and Limitations
Te domains with the most advanced treatment of literary uni-
versals thus far are poetics (i.e., prosody and meter ) and nar-
rative . Literary universals in the study of narrative are treated
thoroughly elsewhere (see narrative universals, narra-
tology, and story schemas ). One recent discovery by Patrick
Colm Hogan is that narratives seem to fall into three prototype
stories, a signifcant improvement over existing, more reductive
approaches (e.g., Joseph Campbells androcentric monomyth
of the heros journey).
Well before literary universals was coined, Paul Kiparsky pre-
sciently argued for a universal metrics: [A] theory of meter
cannot restrict itself to one poetic tradition, any more than a the-
ory of grammar can restrict itself to one language. We must make
our theory account for metrical systems of other languages, and
of a monogenetic mother literature. For instance, it is possible
to reconstruct a protoline of Indo-European epic poetry like he
killed the dragon (Watkins 1995 ), suggesting that heroic tales
are a common origin for global literatures .
Of course, common origin may imply prevalent rather than
across-the-board. If universals are not absolute, they are statis-
tical , that is, occurring more often than chance alone would pre-
dict. Te common distinction between poetry, drama, and fction
is a statistical universal because these forms are widely but not
universally distinguished in the literary traditions of the world.
Universals that correlate (in ways also not infuenced genet-
ically or areally) are typological universals (see typology) .
One typological universal may be that if a tradition has a cate-
gory for nonfction, it will also (n.b. the awkward non -) have
a category for fction, or that drama presupposes poetry.
In other words, such literary categories may function like basic
color terms (Berlin and Kay 1991 ), with traditions that difer-
entiate fewer kinds of literature, including the same few kinds.
Logical universals are typological universals that are logically
entailed by the nature of the given literary phenomena: Tus, a
narrative has only two options for recounting a plot sequence,
either temporally or out of that temporal order (e.g., a fashback).
Tis suggests, in turn, a less obvious but important statistical uni-
versal: very few plots are atemporal, far from the half expected by
random distribution.
Above all, literary universals are empirical they can be
(coldly) documented and are not the products of one cultural
point of view imposed upon another. Tus, any of the common
usages of universal(ity) in literary studies that imply norma-
tive, hegemonic, or totalizing judgments do not pertain to lit-
erary universals as discussed here but are instances of critical
contamination that parallel genetic and areal infuence. (For
further discussion of this issue, along with other terminology of
literary universals, see Hogan 2006.)
Universals and Literary Theory
One of the frst universalist schemes in Western literary theory is
presented in Platos Ion, which probes the inability of oral poets
to account for their activities. Socrates explains poetic divine
possession by imagining a magnetic chain beginning with the
Muse, the oracular lodestone, whose power suspends a descend-
ing series of iron rings: original composer, intermediate reciter,
and fnal audience. Tough meant in the Platonic scheme to
belittle all of art as a derivative and irrational form of knowledge,
this doctrine of inspiration has remained a fundamental myth of
literary theory through the twentieth century and beyond, as for
Muse has been substituted a virtual series of pervasively pow-
erful and likewise subconscious lodestones: psyche, economy,
ideology, identity, empire, and so on.
Modern literary theory, however, has taken an attitude
toward universals that is schizophrenic, tacitly assuming that
universalizing theoretical modes are possible while vociferously
denying that literary works are anything but contingent idio-
graphic particulars. Two passages from Ren Wellek and Austin
Warrens infuential Teory of Literature , just a page apart, are
symptomatic. Te frst distances itself from scientifc universal-
ism of any kind: [N]o general law can be assumed to achieve
the purpose of literary study: the more general the more the
Literary Universals
457
Future Directions
As the aforementioned paths are furthered, two broad new ave-
nues of investigation present themselves. Te frst has to do with
defning and integrating the fundamental disciplinary unit of
analysis of literature. In each discipline, the unit of analysis drives
the research paradigm, such as event and its time-context for
history, or the atom and its forces in physics. Literary studies are
blessed, or cursed, with an array of merely informative terms,
such as character , theme , genre, and reading, of which none by
itself provides the overarching anecdote for literary studies. In
practice, moreover, the feld is divided among such competing
intradepartmental interests as criticism, theory, creative writing,
rhetoric and composition, education, flm studies, and jour-
nalism. Tere may nonetheless be a grand theory of literariness
that unites these disparate subfelds; if so, it would also likely
reveal the fundamental interconnections among all the sister
arts (literature, visual art, performing art, and new media). Will
there be a new grammatike , the ancient word for literary study,
now a more limited purview of linguistics?
Te second little-explored territory for literary universals
(even since Carroll 1995 ) is the area of diachronic universals
(cf. language change, universals of ) . For instance, do all
literatures begin with poetry and then proceed to prose? Does
myth typically diverge into history (fact) and fction (imag-
ination)? Are there other universals of the rise and fall of various
genres over time? Te word tradition , often used in this entry as
synonymous with a literature, implies an entire literary history,
and the historical mode is one of the literary disciplines oldest
and largest strands. Much of this standing evidence might be
mined for diachronic literary universals .
Christopher M. Kuipers
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Berlin , Brent , and Paul Kay . 1991 . Basic Color Terms: Teir Universality
and Evolution . 2d ed. Berkeley : University of California Press .
Brown , Donald E. 1991 . Human Universals . New York : McGraw-Hill .
Burke , Kenneth. [1945] 1969 . A Grammar of Motives . Berkeley : University
of California Press .
Caillois , Roger. 1961 . Man, Play, and Games . Trans. Meyer Barash . New
York : Free Press .
Carroll , Joseph . 1995 . Evolution and Literary Teory . New York : Cambridge
University Press .
Dissanayake , Ellen . 1992 . Homo Aestheticus: Where Art Comes from and
Why . New York : Free Press .
Fabb , Nigel . 2002 . Language and Literary Structure: Te Linguistic
Analysis of Form in Verse and Narrative . Cambridge : Cambridge
University Press .
Hogan , Patrick Colm . 1994 . Te possibility of aesthetics . British Journal
of Aesthetics 34.4 : 337 50.
. 1997 . Literary universals . Poetics Today 18.2 : 223 49.
. 2003 . Te Mind and Its Stories: Narrative Universals and Human
Emotion . New York : Cambridge University Press .
. 2006 . What are literary universals? Literary Universals Project.
Available online at: <http://litup.unipa.it/docs/whatr.htm>.
Jakobson , Roman . 1956 . Two aspects of language and two types of apha-
sia disturbances. In Fundamentals of Language , ed. Roman Jakobson
and Morris Halle , 5582. Te Hague: Mouton .
. 1960 . Closing statement: Linguistics and poetics. In Style in
Language , ed. Tomas Sebeok , 35077. Cambridge, MA : MIT Press .
begin to construct a universal metrics (1981, 266). Metrical
schemes are by nature universal formulas for generating and
judging individual poems: [W]e can only begin to state invari-
ant facts about the iambic pentameter line if we state them in
terms of an abstract representation of the line, rather than by ref-
erence to any of the actual performances of the line (Fabb 2002 ,
6). Although the stanza has received some attention, the focus of
prosody is the verse line (or line group, e.g., the elegiac cou-
plet), afording a vast corpus for analysis.
Other suggestive work on literary universals has been done
on the level of fundamental literary concepts. Evolutionary psy-
chology has confated Darwinian survival with literary confict
(the primary problem or adversary to be overcome in a given
story), but this is only one of the bases of literature, being limited
primarily to narrative or dramatic works. In poetry, probably the
most ancient literary form, the dominant literary universals seem
to be imagery (regarding content) and meter (regarding form).
Some early formalists believed that imagistic language was the
root of all literature. However, literature is animated by both
imagistic tropes and also schemes (recalling the classical distinc-
tion within the rhetorical canon of style ), that is, the deliberate
patterning of existing linguistic components (i.e., multitudes of
parallelism: rhyme , chiasm, ascending cola, etc.). Schemes
are, in Jakobsons vocabulary, either poetic or phatic (i.e.,
at play with the message itself or its instantiation in a linguistic
code), rather than referential, iconic, or invoking deliberate
visualization.
Tropes have found other universalizing uses since Peter
Ramus, Giambattista Vico , and other Enlightenment rhetori-
cians elevated four of these traditional fgures of thought to the
status of master tropes : metaphor (see metaphor, universals
of ), metonymy, synecdoche, and irony . Of these, metaphor
has been most ascendant (see, e.g., Kvecses 2005 ). Te cogni-
tivist theory of conceptual metaphor even suggests that the
fundamental metaphoric principle, to understand one thing in
terms of another, applies to all concepts, since the mind is uni-
versally embodied through the constant real-world experience
of comparison (Lakof and Johnson 1999 ). Kenneth Burke took a
specifc form, the drama, as a metaphor of all social action ([1945]
1969). Going a step beyond both Aristotles triad and Jakobsons
six factors of communication, Burkes pentad of dramatism
(scene, act, agent, agency, purpose; with a sixth term, attitude,
supplemented later) was conceived synecdochically so that its
terms could pair into any of ten ratios (e.g., the pathetic fallacy
is an instance of the scene-act ratio) .
Even as previously literary things as metaphor and drama
show great power for comprehending human thought and soci-
ety, however, there remains the problem of the way that such
fndings feed back to illuminate literature per se. If we all use met-
aphor continually in language, what then sets apart the poetic
metaphor of a Shakespeare sonnet? Is all literature ultimately
dramatic? As the study of literary universals moves forward, it
must be remembered that ever since Saussurean structuralism
and Russian (and other Eastern European) formalisms, that the
fundamental theoretical agenda is literariness ( literaturn-
ost ), rather than the particular literary work (which remains the
realm of criticism) the langue of literature, as it were, rather
than the parole .
Literary Universals
458
sound ( metrical efects or alliteration ), syntax (such as
ellipsis or inversion), or semantics ( metaphor , hyperbole,
etc.). Te Russian Formalist critic Victor Shklovsky, commenting
on the purpose of literary devices, argued that literary art exists
to make one feel things; its purpose is to increase the dif culty
and length of perception ([1917] 1965, 12). Te immediate efect
of foregrounding is to make strange ( ostranenie ), to defamiliar-
ize. Tese ideas allow the empirical researcher to frame specifc
hypotheses addressing the impact that foregrounded passages
have on readers.
First, as studied by Willie van Peer ( 1986 ), readers should fnd
that passages high in foregrounding are striking when compared
with passages low in foregrounding. To test this hypothesis, van
Peer chose four short poems and carried out a comprehensive
analysis of the foregrounding in each line at the three diferent
levels (sound, syntax, semantics). Tis enumeration enabled
the lines of the poems to be rank-ordered for density of fore-
grounding. Readers were then asked to respond to the poems by
underlining all words and phrases they found striking. In all of
the poems, the frequency of readers underlinings was found to
correlate highly with the density of foregrounding.
A similar study by David S. Miall and Don Kuiken ( 1994 ) took
up the additional suggestions of Shklovsky that art makes us feel
and that perception is lengthened. Working with three modern-
ist short stories, they analyzed the presence of foregrounding in
each sentence. Te stories were then presented a sentence at a
time on computer; reading times per sentence were recorded
while readers undertook a frst reading at their normal reading
speed; they then read the story a second time while providing
a rating of each sentence. For all of the stories, after adjusting
for sentence length, the speed of reading correlated signifcantly
with foregrounding (highly foregrounded sentences took about
twice as long to read as sentences without foregrounding); and
readers ratings for strikingness and intensity of feeling also cor-
related with foregrounding. Te readers in both this study and
that by van Peer were university students from a wide range of
backgrounds, yet correlations with foregrounding were signif-
cant regardless of their expertise in literature; thus, response to
foregrounding appears to be independent of literary education.
Tese fndings suggest a theory of text processing: Te encoun-
ter with foregrounding is found striking by the reader, who then
slows down in order to gain a better apprehension of the unusual
textual features; the experience is defamiliarizing and arouses
feeling, and feeling may be the vehicle by which the reader elicits
an alternative framework for reconceptualizing the meaning of
the text at that moment. Te main fndings have been confrmed
in several later studies on the efects of foregrounding .
Literary reading is often said to be engaging in its power to
evoke ideas and feelings about the self. A method for investigat-
ing this idea, termed self-probed retrospection , was frst demon-
strated in a study by Ufe Seilman and Steen F. Larsen ( 1989 ).
Tey proposed that a literary text was more likely to evoke per-
sonal resonance than a nonliterary text. Readers were given either
a short story or an expository text (on population growth), both
of about 3,000 words. While reading, readers put a pencil mark
in the margin whenever they were reminded of something from
their own lives; otherwise, reading occurred at a normal pace.
Te two texts gave rise to the same number of remindings: about
Kiparsky , Paul . 1981 . Stress, syntax and meter. In Essays in Modern
Stylistics , ed. Donald C. Freeman , 22572. New York : Methuen .
Kvecses , Zoltn . 2005 . Metaphor in Culture: Universality and Variation .
Cambridge : Cambridge University Press .
Lakof , George , and Mark Johnson . 1999 . Philosophy in the Flesh: Te
Embodied Mind and Its Challenge to Western Tought . New York : Basic
Books .
Lowe , E. J. 2006 . Te Four Category Ontology: A Metaphysical Foundation
for Natural Science . Oxford : Clarendon Press .
Meeker , Joseph . 1997 . Te Comedy of Survival: Literary Ecology and a
Play Ethic . 3d ed. Tucson : University of Arizona Press .
Polti , Georges . [1921] 1977 . Te Tirty-Six Dramatic Situations . Trans.
Lucille Ray . Boston : Te Writer .
Starobinski , Jean . 1979 . Words upon Words: Te Anagrams of Ferdinand
de Saussure . Trans. Olivia Emmet . New Haven, CT : Yale University
Press .
Watkins , Calvert . 1995 . How to Kill a Dragon: Aspects of Indo-European
Poetics . New York : Oxford University Press .
Wellek , Ren , and Austin Warren . 1977 . Teory of Literature . 3d ed. San
Diego, CA : Harcourt Brace Jovanovich .
LITERATURE, EMPIRICAL STUDY OF
Te broad feld of empirical research over the last three decades
has included a range of topics and approaches. Sociologically
oriented researchers have taken up literary socialization, such as
the reputation of authors, and audience research; book histori-
ans have surveyed the experiences of readers, especially those in
the working class over the last two centuries, and the role of read-
ing clubs; writers interested in aesthetic response have studied
individual literary experience and compared literary response to
experiences of other media, such as flm, computer gaming, or
hypertext. At the heart of the empirical endeavor is the formation
of theories and narratives about the role and status of literature
based on actual data, either verbal or numeric: Tis may consist
of readers memoirs or statistics for library borrowings, the study
of questionnaires elicited from readers, or evidence of reading
processes gathered during carefully controlled laboratory exper-
iments. In this entry, the primary focus is on the formal aspects of
literary texts as refected in studies of literary reception.
Reception studies cover a wide spectrum of topics, includ-
ing style and narrative; readers feelings and the relation of liter-
ary understanding to the self; individual diferences in readers
preferences or the infuence on reading of personality traits;
cross-cultural diferences in reading; and the relation of literary
experiences to other media. Some empirical studies attempt to
clarify or improve on the models of reading developed by dis-
course psychologists; others may represent an attempt to test a
particular claim about reading proposed by literary theorists or
narratologists . Either explicitly or implicitly, a number of
studies have raised the question of literariness : whether lit-
erary texts involve response processes that are measurably dis-
tinctive in some way.
Experimental Examples
In this section I discuss three themes that have been pursued
empirically and provide examples of the ways that readers
responses have been studied.
Te term foregrounding refers to stylistically distinctive
aspects of literary texts. Tese may be apparent at the level of
Literature, Empirical Study of
459
reading. Te authors have examined the relationship with the
narrator both theoretically and empirically in a series of studies
involving aspects such as dialogue, plot, point of view, and char-
acterization. For instance, in studying the efects of free indirect
discourse (where the narrators voice represents the speech or
thought of a character without attribution), they found that this
style led readers to endow the narrator with the personality of the
character. Taking a story about a husband and wife, Rope, by
Katherine Ann Porter , that is related almost entirely in free indi-
rect discourse focused on the male character, they constructed
several other versions of the story in which the character roles
were reversed and the dialogue was rewritten as direct quoted
speech. After reading the story, readers were asked several ques-
tions about their impressions of the narrator and the characters.
It was found, for example, that the rationality of a character was
rated higher when it was associated with the narrator through
free indirect discourse. Judgments about the likely gender of the
narrator were also aligned with the gender of the character rep-
resented through free indirect discourse .
Prospects
Although studies of the kind just described are not yet well
known among literary scholars, they point to a basis for rethink-
ing the nature of literary studies and education. In contrast to
claims voiced by critics from I. A. Richards to E. D. Hirsch that
the responses of ordinary readers are ill-informed or whimsi-
cal, empirical studies demonstrate the presence of signifcant
regularities in readers responses, enabling signifcant conclu-
sions to be drawn about the efects of literary language and form.
Empirical studies also shift the focus away from the interpretative
issues that have largely preoccupied literary scholars onto the
experiential aspects of literary reading. In this respect, they invite
a reconsideration of the role of literature in human culture .
David S. Miall
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bortolussi , Marisa , and Peter Dixon . 2003 . Psychonarratology: Foundations
for the Empirical Study of Literary Response . Cambridge : Cambridge
University Press . An integrated approach to narrative theory and
the empirical study of reading, including exemplary studies by the
authors.
Kuiken , Don , David S. Miall , and Shelley Sikora . 2004 . Forms of self-im-
plication in literary reading . Poetics Today 25 : 171 203.
Larsen , Steen F. , and Jnos Lszl . 1990 . Cultural-historical knowl-
edge and personal experience in appreciation of literature . European
Journal of Social Psychology 20 : 425 40.
Lszl , Jnos , and Steen F. Larsen . 1991 . Cultural and text variables in
processing personal experiences while reading literature . Empirical
Studies of the Arts 9 : 23 34.
Martindale , Colin . 1990 . Te Clockwork Muse: Te Predictability of Artistic
Change . New York : Basic Books .
Miall , David S. 2006 . Literary Reading: Empirical and Teoretical Studies .
New York : Peter Lang . Chapter 3 in this book provides an introduction
to the methods of empirical study, while Chapter 7 surveys the princi-
pal research topics in empirical reception studies.
Miall , David S. , and Don Kuiken . 1994 . Foregrounding, defamiliariza-
tion, and afect: Response to literary stories . Poetics 22 : 389 407.
Oatley , Keith . 1999 . Meetings of minds: Dialogue, sympathy, and identi-
fcation, in reading fction . Poetics 26 : 439 54.
13 for each text. After reading, readers reviewed their marks and
completed a short questionnaire on each reminding: its type,
vividness, emotional quality, and the like. Te types of remind-
ing were found to distinguish the two texts: For the literary text,
twice as many remindings involved a memory of the self as an
actor. Te expository text, in contrast, invoked more memories of
things heard or read about. It was also noticed that remindings
in general were more frequent in the opening sections of both
texts (the downward trend was more marked in the literary text).
Tese fndings suggest that readers of a literary text situate them-
selves by recruiting specifc, self-related information, particularly
at the beginning of a text where an appropriate schema must be
developed, and that this information refers predominantly to the
active engagement of the reader in the world .
Several subsequent studies have built on the remindings
method. Larsen and Jnos Lszl (1990) studied the cultural
proximity of readers, working with Hungarian and Danish read-
ers of a Hungarian story. Of the remindings produced, they
found that markedly more event memories were produced by
the Hungarians and, of these, signifcantly more were of expe-
rienced rather than reported events. Lszl and Larsen ( 1991 )
then extended the method to look at the implications of point
of view in fction. Several passages of the Hungarian story
were rewritten to change point of view. As before, personal-
event remindings were signifcantly more frequent among the
Hungarian readers; in addition, shifting to the inside point of
view of a character increased the percentage of such remind-
ings from 55 percent to 75 percent for the Hungarians (but had
no efect on Danish readers). Tere was also some evidence that
inside point of view infuenced readers toward more emotional
remindings, regardless of their cultural background .
Another variant of the remindings method was developed by
Keith Oatley ( 1999 ) to examine gender and personality diferences
in readers. Instead of a simple mark, readers were instructed to
write a letter in the margin: an E for an emotion; M for an autobio-
graphical memory; and T for a train of thought. Among readers of
short stories by Alice Munro and Carson McCullers , female read-
ers produced overall signifcantly more emotions than the male
readers; in addition, male readers produced fewer emotions when
the protagonist was female, whereas gender of protagonist had
no infuence on female readers. In a second study, the method
was used to examine the aesthetic distance of readers from a
short story and how their responses mirrored their adult attach-
ment styles. Kuiken, Miall, and Shelley Sikora ( 2004 ) developed
the method of self-probed retrospection to elicit verbal commen-
taries by readers. Te reader is invited to read a text and make
marginal marks whenever a passage seems striking or evocative;
the reader later returns to the marked passages and chooses (say)
fve on which to provide a commentary. Readers are readily able
to recover the thoughts and feelings that occurred during reading,
giving access to at least some of the mental processes that appear
to make literary reading distinctive .
A third set of studies is focused on the role of the narrator in
fction. Marisa Bortolussi and Peter Dixon ( 2003 ) elaborate a
theoretical framework that accounts for the readers relation to
the narrator; in particular, they propose that readers treat the
narrator as a conversational participant and make inferences
about the narrators personality and values that infuence their
Literature, Empirical Study of
460
context, the latter sentences contain at least one logical connec-
tive . In standard approaches, the truth conditions for molecular
sentences are given recursively, piggy-backing, as it were, on
the truth conditions for atomics. (An example follows.)
For present purposes, a language is a precise syntax (involv-
ing, among other things, a precisely defned set of sentences,
some of which are atomics, some molecular) coupled with truth
conditions, which, as noted, provide truth-in-a-case conditions
for all sentences. So, in addition to specifying a syntax, ones
specifcation of a language involves specifying a class of cases in
terms of which all sentences, provided by ones specifed syntax,
enjoy truth-in-a-case conditions. (An example follows.)
Logical Consequence Qua Truth Preservation
Te consequence relation of a language (or fragment thereof)
is the chief concern of the feld of logic, broadly understood. In
efect, a consequence relation yields what follows from what,
what sentences of the language logically follow from what sen-
tences. Given a language, as understood here, we defne logical
consequence or validity (i.e., semantic validity) as follows, where
L is a language, and A and B sentences of L.
Val. B is a consequence of A in L if and only if there is no case in
which A is true but B untrue.
If B is a consequence of A in L, we say that the argument from
A to B is (semantically) valid in L, that B logically follows from A
in L, that A (semantically) implies B , with all such terminology
being equivalent (for present purposes).
(Val), in turn, may be generalized. We say that a set X of
L-sentences is verifed-in-a-case just if every member of X is
true-in-that-case. In turn, we say that the argument from X to A
is valid just if theres no case in which X is verifed but A untrue.
Similarly, we say that a sentence A is logically true in L just if there
is no case in which A is untrue .
Sample Language and Logic
Consider an example of the foregoing ideas, in particular, a so-
called classical propositional language . (Such languages are ter-
ribly simple; they have no quantifers. To simplify even more,
our propositional language will contain no names or predicates!)
One motivation for the language is that we seem to have so-called
truth-functional connectives in English (and natural languages,
generally), and one might be interested in clearly specifying the
logic of that (truth-functional) fragment of our language. For
example, there seems to be a truth-functional usage of and , one
in which and expresses conjunction , where a conjunction is true
in a given case if and only if both conjuncts are true in the given
case. Similarly, a truth-functional usage of negation in English is
evident, one in which, for example, negation does no more nor
less than toggle truth values.
As noted, we frst need to precisely specify a language . We
need to specify a vocabulary (in efect, the building blocks
of the language) and, in general, a full syntax, which contains a
(precisely defned) set of sentences; we then give our truth con-
ditions. We proceed to defne our language L as follows.
(1) Vocabulary: p, q, and r, with or without subscripts
(numerals for natural numbers), are our atomics . In addition
Seilman , Ufe , and Steen F. Larsen . 1989 . Personal resonance to litera-
ture . Poetics 18 : 165 77.
Shklovsky , Victor . [1917] 1965 . Art as technique. In Russian Formalist
Criticism: Four Essays , ed. and trans. L. T. Lemon and M. J. Reis , 324.
Lincoln : University of Nebraska Press .
Steen , Gerard , and Dick Schram , eds. 2001 . Te Psychology and Sociology
of Literature: In Honour of Elrud Ibsch . Amsterdam : John Benjamins .
A wide-ranging collection, mainly illustrating recent empirical studies
of literature.
van Peer , Willie . 1986 . Stylistics and Psychology: Investigations of
Foregrounding . London : Croom Helm .
van Peer , Willie , ed. 2007 . Foregrounding. Language and Literature 16 .2
(Special Issue). A recent collection of contributions to foregrounding,
including both theoretical and empirical studies.
van Peer , Willie, Frank Hakemulder , and Sonia Zyngier . 2007 . Muses
and Measures: Empirical Research Methods for the Humanities .
Cambridge : Cambridge Scholars Publishing .
Zwaan , Rolf . 1993 . Aspects of Literary Comprehension: A Cognitive
Approach . Amsterdam and Philadelphia : John Benjamins .
LOGIC AND LANGUAGE
Every language, suitably understood, has a logic, suitably under-
stood. Te suitable understanding is a common semantic con-
ception of logic and language. On this conception, the logic of
a language is the so-called consequence relation , which, on the
semantic conception, essentially involves truth preservation . Te
chief goal of this entry is to briefy convey the basic and very
common sense in which every language has a logic. (N.b.: for
space and simplicity reasons, this essay privileges the so-called
semantic, or model-theoretic, approach to logic. Moreover, this
essay again, for space reasons only aims to convey basic ideas;
it doesnt aim to be a history or even survey of the semantic con-
ception of logic and language.)
Languages and Truth Conditions
In specifying the logic of a language (or some fragment thereof),
one seeks precision. Much as physics idealizes away from the
messiness of physical reality, formal semanticists and logicians
(at least those concerned with natural languages) idealize away
from the messiness of linguistic reality. One such idealization is the
assumption that all (declarative) sentences of a language have so-
called truth conditions . (Another immediate idealization is that we
can easily, and precisely, specify the target declarative sentences,
the sentences that, in some sense, are used to make assertions . In
what follows, sentence will be short for declarative sentence.)
For present purposes, such truth conditions are best thought of as
truth-in-a-case conditions, that is, conditions that provide, for any
relevant case, what it takes for sentences to be true-in-that-case.
If one thinks of cases as possible worlds , then truth conditions
provide the conditions under which sentences are true-in-w , for
any possible world w . Similarly, if one thinks of cases as situa-
tions , then truth conditions provide the conditions under which
sentences are true-in-s , for any situation s . Moreover, if one thinks
of cases as Tarskian models, then truth conditions provide the con-
ditions under which sentences are true-in-M , for any model M .
In addition to the assumption of truth conditions, another
idealization is that sentences may be cleanly, precisely carved
into the atomics and molecular sentences, where, in the present
Logic and Language
461
consequence relation, which carries the information about what
follows from what in the given language .
Artifcial Versus Natural Languages
One might agree that every artifcial language, as understood
here, has a logic, that is, a consequence relation, specifed via
(Val). What, though, of natural languages?
Te question is a good one, but very complex. Natural lan-
guages appear to have arguments that are truth preserving
in the strict sense of (Val) arguments such that theres no
case in which the premises are true but the conclusion untrue.
(Consider the limit example: the argument from A to A .) Te
trouble, of course, concerns the relevant cases involved in nat-
ural languages truth conditions. Assuming (a not insignifcant
assumption) that all sentences of a natural language have truth
conditions in the relevant sense, it remains unclear what counts
as a relevant case in such truth conditions.
For present purposes, the right account of cases for natural
languages is not pressing. Te pressing issue is whether, in the
end, natural languages are suf ciently equipped with truth-in-a-
case conditions, whatever the cases may be. If so, the chief point
remains: Any such language, in virtue of (Val), enjoys a logic .
Logical Theories and Rivalry
A logical theory of a language (or fragment thereof) is a theory of
the consequence relation on that language (or fragment). One
way in which logical theories might disagree is on the choice
of logical connectives, but this is not necessary for disagree-
ment. Two logical theories might agree on the class of (rele-
vant) connectives in a language (fragment) while nonetheless
disagreeing about the logical behavior of such connectives a
disagreement that, in general, will show up in rival truth con-
ditions for the given connectives. (Such disagreement over
truth conditions often centers on what counts as a case in the
relevant truth conditions.) Suf ce to say that rivalry currently
reigns in the feld of logic, at least concerning the right logical
theory of natural language (or fragments thereof). Problems
of vagueness and consistency, truth , and paradox
are particularly fertile phenomena for contemporary rivalry
among logical theories .
J. C. Beall
SUGGESTIONS FOR FURTHER READING
Any textbook on classical logic (of which there are many ) will be
suitable further reading. From there, one might turn to textbooks
on nonclassical or so-called intensional logics (of which there
are many). As a frst step, one might proftably peruse entries
under logic in the Stanford Encyclopedia of Philosophy , edited
by Edward N. Zalta, available online at: http://plato.stanford.
edu/.
LOGICAL FORM
Te construction of systems in which valid inference can be char-
acterized has been the central concern of logic since its incep-
tion with the ancients. Beginning with Gottlob Freges seminal
insights of the late nineteenth century (Frege 1879 , 1893 ), it has
to the atomics, we have a set of punctuation marks, namely,
( and ). Furthermore, we have a set of connectives: & is a
binary connective (takes two sentences to make a sentence),
and ~ is a unary connective (takes one sentence to make a
sentence). Our three given sets of symbols are disjoint.
(2) Te set S of L-sentences is defned recursively as
follows.
a) All atomics are L-sentences.
b) If A and B are L-sentences, then so too are ~ A and ( A&B) .
c) Nothing is an L-sentence unless its being so follows from
(2a) and (2b).
(3) L- cases are (total) functions c from S into V = {1,0}, where
V is our set of semantic values.
(4) Truth conditions: an L-sentence A is true in a case c if
c ( A ) = 1.
a) An atomic sentence A is true in a case c if c ( A ) = 1. ( A is
false in c otherwise.)
b) A sentence of the form ~ A is true in a case c if c ( A ) = 0.
c) A sentence of the form ( A&B ) is true in case c if c ( A ) = 1
= c ( B ).
With our language L so given, we can now see the sense in which
every language at least given suitable idealizations has a logic.
Applying (Val), we immediately see that, for any L-argument, it is
either truth preserving , in which case valid in L, or not (in which
case, invalid in L).
EXAMPLE. Consider the L-argument from premise (p&q) to
conclusion q. According to (Val), this argument is valid in L just
if theres no case in which (p&q) is true but q untrue. In L,
our cases are functions from the L-sentences into {1,0}. Is there a
case in which (p&q) is true but q not true? No. To see this, just
consider the truth conditions on L-sentences. According to those
conditions, a sentence of the form ( A & B ) is true in a case just if
both A and B are true in the given case. So, for any case c , if c (p&q)
= 1, then c (p) = 1 and c (q) = 1, in which case c (q) = 1. Hence, there
is no case c in which (p&q) is true but q isnt true.
What one also notices or would notice, on refection about
L is familiar truth-functional behavior for negation. For exam-
ple, given the truth conditions in L, and given (Val), it is easy to
see that p is a consequence of ~~p, and vice versa. In other
words, the double negation of a sentence is logically equivalent
to the given sentence.
One may proceed (as an exercise) to record the other logical
forms that are valid in our artifcial language L. Once recorded,
one has a precise account of the logical behavior of & and ~
in L. In turn, one may evaluate whether such logical behavior
accurately captures the behavior of corresponding connectives
in ones natural language. In this respect, the artifcial L serves
the role of idealized models in natural sciences: It gives a clear
account of how the target phenomena (in this case, logical con-
nectives) behave. Rivalry among logical theories (more on which
follows) turns on the extent to which L is an accurate model.
Of course, L is but one example of a language (i.e., an ideal-
ized, artifcial language), and a very simple one at that. Still, it
is not hard to see that, provided languages come equipped with
precisely defned cases and each sentence enjoys truth-in-a-
case conditions, (Val) quickly yields a logic for the language a
Logic and Language
Logical Form
462
predecessors, Tarski was skeptical that natural languages are
such systems. Indeed, his recommendation was that we eschew
anything other than formal languages when engaging in scien-
tifc discourse. A more moderate view emerged, however, which
tried to show that natural languages, at least to a certain extent,
could be rendered in the logical idiom. Most closely associ-
ated with W. V. O. Quine ( 1950 ), the idea was that logic is to
be understood as schematic. On this view, there are no logical
propositions, per se, but only propositional schema; valid infer-
ence is characterized with respect to such schemata, and holds
for any instantiation of the schema. Natural languages can
then be regimented as instances of the schemata; a sentence
then has a certain logical form because it conforms to a prop-
ositional schema of a certain form. (A simple example: John
came and Mary left is an instance of the schema & , and
so has the logical form John came & Mary left, where & is
the logical symbol for conjunction.) Quines view nevertheless
is no departure from the tradition that distinguishes grammat-
ical from logical form; it does depart, however, in holding that
systematic associations can be established between grammat-
ical forms and logical forms, for signifcant aspects of natural
languages (1960) .
Accepting the traditional separation of grammatical and log-
ical form is not universal; rejection of it has been central within
linguistic theory since the mid-1970s. On this latter view, the
derivation of a sentences logical form is an aspect of its syntac-
tic derivation, hence, an aspect of its grammatical form . Again,
the central case is quantifcation; in pivotal work, Robert May
( 1977 ) showed that the scope of quantifers, including multi-
ple quantifers, can be represented, as in the manner noted,
by assuming that there is a transformational rule that moves
quantifer phrases, leaving a trace, interpreted as a variable
bound by the moved phrase. By hypothesis, the formulation
of Mays rule QR requires theoretical resources no greater
than those independently needed within linguistic theory to
otherwise express transformational mappings, for instance,
wh -movement. Te class of syntactic representations gener-
ated by transformational mappings, including those efected
by QR, is known as LF. Tus, sentences with multiple quanti-
fers, such as Everyone loves someone will have two distinct
LF-representations, roughly:
[everyone
i
[someone
j
[ t
i
loves t
j
]]]
and
[someone
j
[everyone
i
[ t
i
loves t
j
]]]
which can be defned as representing the difering scope
orderings of the quantifers, the traces of QR being interpreted
as variables bound by the quantifers, so that in this regard,
LF-representations constitute logical forms. Tat grammars
of natural languages have the rule QR is now a widely (if not
universally) accepted assumption within linguistics (Fiengo
and May 1994 ; Fox 2000 ; Hornstein 1984 , 1995 ; Larson and
Segal 1995 ; May 1985 ; Reinhart 1997 ); among the most well-
known independent arguments are those from weak crossover
(Chomsky 1976 ) and anaphoric binding more generally, and
antecedent contained deletion (May 1985). It has also become
a commonly accepted assumption within recent thinking in
been understood that accomplishing this goal in a manner suf-
fciently rigorous that the inferred proposition can be taken to
be proven requires strict attention to the structural properties
of propositions, to their logical form . What was crystalized by
Frege was that this form can be revealed only in a language that
difers in two key respects from the superfcial form of natural
languages: i) Te structure of propositions is function-argument,
not subject-predicate, and ii) expressions of generality, unlike
proper names, do not occur as arguments, but rather bind vari-
ables that do. Together, these two diferences aforded the frst
adequate account of multiple generalization; by distinguish-
ing xy( P ( x,y )) from y x ( P ( x,y )) in terms of the scopes of
the universal and existential quantifers, Frege was able to allay
one of the central problems that had plagued traditional logic.
(Cf. Kneale and Kneale 1962 , 483f.)
Freges insight, that grammatical form does not reliably reveal
logical form, was taken up by Bertrand Russell, most notably in
the theory of descriptions (Russell 1905 ). Russell proposed that
defnite descriptions, as in Te present King of France is bald,
is not to be understood in the manner of a proper name, that is,
standing as an argument, but rather as a complex term of gen-
erality. Tus, the proper logical form is not B ( k ), but rather
x ( K ( x ) & y ( K ( y ) x = y )) & B ( x ); that is, there is one and only
one present King of France and he is bald. By taking this to be
the proper logical form, Russell argued that a number of logical
issues could be directly addressed. For example, the ambiguity of
Te present King of France is not bald could be accounted for
by taking the negation as having scope either inside or outside
the existential quantifer; negation having broader scope brings
the case into conformance with the law of the excluded middle,
as Russell observed.
Both Frege and Russell realized not only that the insights
about logical form being surveyed clarifed the formal nature of
inference, but that these aspects of form also allowed for seman-
tic elucidation, being directly related to an account of the con-
ditions for the truth of propositions. For instance, for Russell,
a substantial virtue of the account of descriptions was that the
intuition that Te present King of France is bald is false can be
directly accommodated. But that we can proceed beyond intui-
tive elucidation to a formally and materially adequate defnition
of truth, based on the sort of conception of logical form pioneered
by Frege and Russell, is due to Alfred Tarski ([ 1936 ] 1956, 1944).
In the case at hand of quantifcation, Tarskis semantic clauses
run as follows: With respect to a universe of objects U , x ( P ( x ))
is true just in case every sequence of objects of U satisfes P ( x );
x ( P ( x )) is true if this is so for some sequence. (A sequence S
satisfes an open formula P ( x ) if there is an assignment of a
value a of S to the variable x such that a is P .) Because Tarskis
method iterates, it extends to multiple generalization, distin-
guishing the truth conditions of x y ( P ( x,y )) from those of
y x ( P ( x,y )), thus providing semantic foundation for the syn-
tactical insights of Frege.
Central to the importance of Tarskis formalization of seman-
tics is that it paved the way to metalogic, the study of the prop-
erties of logical systems, centrally their consistency, soundness,
and completeness. Tarski was clear to maintain that the def-
nition of truth could be reliably applied only to those systems
whose propositions have the requisite logical form; like his
Logical Form
463
Firstly, formal logic as it has been developed by Gottlob Frege ,
is seen both as an instrument for analysis and as an ideal language
wherein all scientifc knowledge is expressable. Many logical posi-
tivists also accepted Freges logicism, namely, the view that math-
ematics is reducible to logic. Hence, they endorsed the view that
mathematics is a language, not a science like, for example, physics.
Secondly, it follows that a clear distinction can be made
between analytic and synthetic sentences. Te former consist
of logical and mathematical tautologies, whereas the latter can
be either true or false and are therefore dependent on the way
things are; that is, they are empirical.
Tirdly, this leads to the principle of verifiability : If a sen-
tence is meaningful, then it should be possible to determine its
truth value .
Straightforward consequences of these principles are, frstly,
that metaphysical statements are neither analytic, as they are not
tautologies, nor synthetic, as they do not refer to the empirically
accessible world; hence, they are meaningless. And, secondly, if all
of the sciences can be expressed in one and the same language, that
is, the language of mathematics, all sciences can be unifed into a
single framework. Hence, the unity of science is a reachable goal.
It should be noted that although there are connections with
Ludwig Wittgensteins views, as expressed in the Tractatus
Logico-Philosophicus, and although Wittgenstein attended some
of the meetings of the Vienna Circle, it would not be correct to
label him a logical positivist.
On a more refned account, qualifcation is needed. It suf-
fces to look at the original manifesto of the Vienna Circle, the
Wissenschaftliche Weltaufassung (the Scientifc Worldview ), to
notice that the logical positivist program also included ethical-
societal views. In recent years, many authors have made a strong
case to have a second and historically more nuanced look at logi-
cal positivism (see, e.g., Michael Friedman 1999 ) .
It is generally accepted that both Karl R. Popper , the found-
ing father of falsifcationism , and Willard Van Orman Quine, the
founding father of naturalized epistemology, have been the most
important critics of logical positivism. Te former questioned
the verifability theory; the latter rejected the analytic-synthetic
distinction .
Jean Paul Van Bendegem
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ayer , Alfred . 1936 . Language, Truth and Logic . London : Victor Gollancz .
Friedman , Michael . 1999 . Reconsidering Logical Positivism .
Cambridge : Cambridge University Press .
Janik , Allan , and Stephen Toulmin . 1973 . Wittgensteins Vienna . New
York : Simon and Schuster . Tis book outlines the cultural setting
wherein logical positivism could arise.
M
MAPPING
In what follows, mapping is used in the general mathematical
sense of a partial or total correspondence between elements,
relations, and/or structures in two sets.
philosophy of language (King 2001 ; Neale 1990 ; Stanley 2000 ;
Ludlow 2002 ) .
Robert May
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , Noam . 1976 . Conditions on rules of grammar . Linguistic
Analysis 2 : 303 51.
Fiengo , Robert , and Robert May . 1994 . Indices and Identity . Cambridge,
MA : MIT Press .
Fox , Danny. 2000 . Economy and Semantic Interpretation . Cambridge,
MA : MIT Press .
Frege , Gottlob. [ 1879] 1967. Begrifsschrift: A Formula Language Modeled
Upon Tat of Aithmetic, for Pure Tought. Trans. Stefan Bauer-
Mengelberg . In From Frege to Gdel , ed. Jean van Heijenoort , 582.
Cambridge : Harvard University Press .
. [ 1893] 1964. Te Basic Laws of Arithmetic . Trans. Montgomery
Furth . Berkeley and Los Angeles : University of California Press .
Hornstein , Norbert. 1984 . Logic as Grammar . Cambridge, MA : MIT
Press .
. 1995 . Logical Form: From GB to Minimalism . Oxford : Blackwell .
King , Jefrey C. 2001 . Complex Demonstratives: A Quantifcational
Account . Cambridge, MA : MIT Press .
Kneale , William , and Martha Kneale . 1962 . Te Development of Logic .
Oxford : Oxford University Press .
Larson , Richard , and Gabriel Segal . 1995 . Knowledge of Meaning .
Cambridge, MA : MIT Press .
Ludlow , Peter. 2002 . LF and natural logic, In Logical Form and Language ,
ed. Gerhard Preyer and Georg Peter . Oxford : Oxford University Press .
May , Robert . 1977 . Te Grammar of Quantifcation . Ph.D. diss.,
Massachusetts Institute of Technology .
. 1985 . Logical Form: Its Structure and Derivation . Cambridge,
MA : MIT Press .
. 1999 . Logical form in linguistics. In Te MIT Encyclopedia of
the Cognitive Sciences , ed Robert A. Wilson and Frank C. Keil , 4867.
Cambridge, MA : MIT Press .
Neale , Stephen. 1990 . Descriptions . Cambridge, MA : MIT Press
Quine , W. V. O. 1950 . Methods of Logic . New York : Henry Holt.
. 1960 . Word and Object . Cambridge, MA : Technology Press .
Reinhart , Tanya. 1997 . Quantifer scope: How labor is divided between
QR and choice functions . Linguistics and Philosophy 20 : 399 467.
Russell , Bertrand. 1905 . On denoting , Mind 14 : 479 93.
Stanley , Jason. 2000 . Context and logical form . Linguistics and
Philosophy 23 : 391 434.
Tarski , Alfred. [ 1936] 1956 . Te concept of truth in formalized lan-
guges. In Logic, Semantics, Metamathematics, trans. J. H. Woodger .
Oxford : Oxford University Press .
. 1944 . Te semantic conception of truth . Philosophy and
Phenomenological Research 4 : 341 75.
LOGICAL POSITIVISM
Also known as logical empiricism, logical positivism was an
important philosophical movement in the frst half of the twenti-
eth century that reached its peak in the interbellum period and is
associated with the Vienna Circle (Wiener Kreis) and the Berlin
Circle (Berliner Kreis) . Te most prominent members of the
former were Moritz Schlick , Rudolf Carnap , Otto Neurath , Hans
Hahn , and Friedrich Waismann ; of the latter, Hans Reichenbach ,
Kurt Grelling , Carl Gustav Hempel, and Richard von Mises .
On the standard account (see, e.g., Alfred Ayer 1936 ), logical
positivism is committed to the following principles:
Logical Positivism Mapping
464
Creative metaphors are often elaborations of conventional ones,
as in the following typical literary example:
Perhaps time is fowing faster up there in the attic. Perhaps the
accumulated mass of the past gathered there is pulling time out
of the future faster, like a weight on a line. (McDonald 1992 ,
8283 )
Tought and language are embodied . Conceptual structure
arises from our sensorimotor experience and the neural struc-
tures that give rise to it. Te properties of grammars are the prop-
erties of humanly embodied neural systems. Inference inherently
built into a source domain will be transferred by projection to an
abstract domain. For example, the conventional metaphors of
SEEING as TOUCHING (e.g., I couldnt take my eyes of her ) and
KNOWING as SEEING (e.g., I see what youre saying ) combine
with one schema for the English preposition over to motivate
overlook : Te line of sight travels over (i.e., above) the object;
hence, there is no contact; hence, it is not seen; hence, it is not
noticed or taken into account. In contrast, look over ( she looked
over the draft ) uses a related but diferent schema for over , a path
covering much of a surface, as in she wandered over the entire
feld . Tis sense combines with the same mappings to produce
a very diferent abstract meaning the object this time is seen
and noticed.
Metonymic mappings link two relevant domains, which may
be set up locally. Tey typically correspond to two categories of
entities, which are mapped onto each other by a pragmatic func-
tion. For example, authors are matched with the books they write,
or hospital patients are matched with the illnesses for which they
are being treated. Metonymic mappings allow an entity to be
identifed in terms of its counterpart in the projection. So, when
a nurse says Te gastric ulcer in room 12 would like some cofee ,
he/she uses the illness (the gastric ulcer) to identify the patient
who has it. Metonymy allows information to be compressed. If
Jack is the patient and if the nurse is addressing a physician, his/
her statement simultaneously conveys that Jack wants cofee and
that he has a gastric ulcer, which could be further intended as a
question to ask if cofee is permitted under the circumstances.
Im in the phone book uses a metonymic mapping from people
to names. It says not only that my name is written in the phone
book but also that the number linked to my name is indeed my
phone number. So it really says something about me, not just
about my name: how to reach me, that I dont mind making my
number publicly available, and so on .
Metonymic and metaphoric mappings can combine to pro-
vide even greater compression, as in Martina is three points away
from the airport , said by a sports announcer of the tennis star
Martina Navratilova , who was about to lose a tournament match.
Te points stand metonymically for the events of losing a point.
Tree such events would lead to defeat. Te events are on a met-
aphorical spatial scale to which the tennis player gets mapped.
On that scale, the player is metaphorically at a spatial distance
of three points from the end of the match which would mean
defeat. A metonymic chain takes us from the end of the match to
defeat, then to exclusion from the rest of the tournament, then
to returning home. Te airport (a place) stands metonymically
for an event (fying home) that starts in that place. Trough the
Much of the theoretical thinking in modern linguistics has
been strongly linked to the development since the 1950s of cog-
nitive science, artifcial intelligence, and neuroscience. Te frst
wave of cognitive science looked upon the brain as a sophisti-
cated symbol-processing digital computer, and linguistic models
in the ffties and sixties took a largely algorithmic approach, with
a strong focus on syntax and logic.
In the 1970s and 1980s, there was a sharply diferent second
wave of thinking that launched a rigorous, empirically based
study of conceptual mappings: analogy, frames, meta-
phor, metonymy, grammatical constructions, and men-
tal space projections. Tis original and ambitious research
program revisited from a modern point of view some funda-
mental issues that have been known since antiquity. It drew
on a powerful multidisciplinary mix of psychology, linguistics,
computational modeling, and philosophy. Names associated
with pioneering eforts in the new feld of conceptual map-
pings include Douglas Hofstadter , Melanie Mitchell , Dedre
Gentner , and Keith Holyoak , for analogy; George Lakof , Mark
Johnson , and Mark Turner , for metaphor and image schemas ;
Erving Gofman and Charles Fillmore , for frames and frame
semantics ; Ronald Langacker , Charles Fillmore, and Adele
Goldberg , for cognitive and construction grammars ; Gilles
Fauconnier , Eve Sweetser , and John Dinsmore , for mental space
projections; and Geofrey Nunberg for metonymic mappings
( pragmatic functions ).
In the 1990s and up to the time of writing, there was substan-
tial further evolution of our thinking on these issues. Te creative
dimension of conceptual mappings was explored through the
study of conceptual blending and compression (Fauconnier
and Turner 2002 ; Coulson 2001 ) and through the modeling of
emergent structure in analogy (Hofstadter 1995 ; Hummel and
Holyoak 1997 ). Te role of primary metaphors was discovered by
Joe Grady ( 1997 ); constraints on mappings were proposed within
metaphor theory and within blending theory .
Metaphor was once commonly viewed as literary, fgurative,
poetic something exotic that we add to ordinary language to
make it more colorful, vivid, and emotional. But since the incep-
tion of conceptual metaphor theory, it is widely acknowl-
edged that metaphor is, in fact, central to thought and language
and necessary for human language in its many forms. In order
to talk and think about some domains (target domains), we use
the structure of other domains (source domains) and the corre-
sponding vocabulary (see source and target ). Some of these
mappings are used by all members of a culture, for instance, in
English, TIME as SPACE. We use structure from our everyday con-
ception of space and motion to organize our everyday conception
of time, as when we say Christmas is approaching , Te weeks go
by , Summer is around the corner , Te long day stretched out with
no end in sight . Rather remarkably, although the vocabulary often
makes the mapping transparent, we are typically not conscious of
the mapping during use unless it is pointed out to us. Tough cog-
nitively active, such mappings are opaque: Te projection of one
domain onto another is automatic. Metaphoric mappings may
also be set up locally, in context, in which case they are typically
perceived not to belong to the language but rather to be creative
and part of the ongoing reasoning and discourse construction.
Mapping
465
multiple modalities (Alac 2006) and anchored by human cul-
tural artifacts as part of socially distributed cognition
(Hutchins 1995).
Biologically, it is currently widely assumed that mappings are
efected by means of neural binding (Shastri 1996 ). Computational
models of such binding have been proposed within the neural
theory of language (Feldman 2006 ). Experimental techniques to
show the psychological reality of various mappings have been
devised by Lera Boroditsky ( 2000 ), Ray Gibbs (Gibbs et al. 1997 ),
and Seana Coulson ( 2001 ), among others
Gilles Fauconnier
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Alac , Morana . 2006 . How brain images reveal cognition: An ethno-
graphic study of meaning-making in brain mapping practice. Ph.D.
diss., University of California, San Diego.
Boroditsky , L . 2000 . Metaphoric structuring: Understanding time
through spatial metaphors. Cognition 75 . 1 : 1 28.
Coulson , Seana . 2001 . Semantic Leaps . Cambridge : Cambridge University
Press .
Fauconnier , Gilles . [1985] 1994. Mental Spaces . Cambridge : Cambridge
University Press .
. 1997 . Mappings in Tought and Language . Cambridge : Cambridge
University Press .
Fauconnier , Gilles , and Mark Turner . 2002 . Te Way We Tink . New
York : Basic Books .
. 2008 . Rethinking metaphor. In Cambridge Handbook of
Metaphor and Tought , ed. Ray Gibbs , 5366 . Cambridge : Cambridge
University Press .
Feldman , Jerome . 2006 . From Molecule to Metaphor . Cambridge,
MA : MIT Press .
Gentner , Dedre . 1983 . Structure-mapping: A theoretical framework for
analogy. Cognitive Science 7 : 155 70.
Gentner , Dedre , Keith Holyoak , and Boicho Kokinov , eds. 2001. Te
Analogical Mind: Perspectives from Cognitive Science . Cambridge,
MA : MIT Press .
Gibbs , R ., J. Bogdonovich , J. Sykes , and D. Barr . 1997 . Metaphor in idiom
comprehension. Journal of Memory and Language 37 : 141 54.
Gofman , E . 1974 . Frame Analysis . New York : Harper and Row .
Grady , J . 1997 . Foundations of meaning: Primary metaphor and primary
scenes . Ph.D. diss., University of California, Berkeley.
Hofstadter , Douglas . 1995 . Fluid Concepts and Creative Analogies .
New York: Basic Books.
Hummel , J ., and K. Holyoak . 1997 . Distributed representations of struc-
ture: A theory of analogical access and mapping. Psychological Review
104 : 42766.
Hutchins , Edwin . 1995 . Cognition in the Wild . Cambridge, MA: MIT
Press.
Lakof , George , and Mark Johnson . 1980 . Metaphors We Live By .
Chicago : University of Chicago Press .
Lakof , George , and Rafael Nez . 2000 . Where Mathematics Comes
From: How the Embodied Mind Brings Mathematics into Being .
New York : Basic Books .
Liddell , Scott K . 2003 . Grammar, Gesture, and Meaning in American Sign
Language . Cambridge : Cambridge University Press .
McDonald , Ian . 1992 . King of Morning, Queen of Day. New York: Bantam
Books.
Mitchell , M . 1993 . Analogy-Making as Perception . Cambridge, MA : MIT
Press .
Nunberg , G . 1978 . Te Pragmatics of Reference . Bloomington : Indiana
University Linguistics Club .
metonymic chaining, fying home links to leaving the tourna-
ment, which links in turn to losing the match, itself caused by
the three lost points. Strikingly, very little of this is indicated
by the linguistic structure itself. It is constructed by means of the
cognitive models that we have for games, tennis, tournaments,
and travel and by applying to them the appropriate mappings.
Te same sentence can take on completely diferent meanings if
we bring in diferent cognitive models.
Mental space projections link elements and relations in con-
nected mental spaces. For instance, in saying Liz thinks her hus-
band is tired , we build a mental space for Lizs reported beliefs,
with a counterpart for her husband and properties within that
space (tired) that may or may not be satisfed in connected
spaces: Liz thinks her husband is tired, but actually hes in great
shape. In saying Last year, Lizs husband was tired , we build a
mental space for last year, and in saying Liz thinks that last
year, her husband was tired , we build a space for last year embed-
ded in a belief space, itself embedded in a base space. presup-
positions (such as Lizs having a husband) can spread across
spaces: In the last example, we infer that Liz has a husband, that
she thinks she has a husband, and that last year, she also had
this husband. But any of these presuppositions can be prevented
from projecting by an explicit overriding entailment.
In mental space projection, the access principle allows a
description of an element to identify its counterpart in another
mental space. For example, if Liz got married to Bob yesterday,
we can say Last year, Lizs husband was tired , identifying Bob in
the mental space last year by means of his counterpart ( Lizs
husband ) in the mental space now.
Conceptual blending generalizes the notion of conceptual
mapping to arrays of multiple mental spaces with the creation
of new blended spaces and the emergence of novel structure.
Such arrays of connected spaces are called integration networks .
Partial mappings link the mental spaces in such networks, and
selective projection maps the spaces onto novel blended spaces.
Te mappings are supported by a small number of vital rela-
tions , such as analogy, change, identity, rolevalue, causeefect.
Compression is systematic in integration networks: A vital rela-
tion in one part of the network can be compressed into a dif-
ferent (or a scaled-down) vital relation in another part of the
network. Take, for example, My tax bill gets longer every year . Te
inputs are the mental spaces corresponding to diferent years. In
each one, there is a tax bill. Tese input spaces are linked by the
vital relation of analogy: Each one is structured by the frame of
paying taxes in a particular year, and each tax-paying situation
is analogous to the others. Te inputs are also linked by disanal-
ogy : Each tax bill is diferent (longer than the previous one).
Te analogous input spaces are integrated into a single blended
space, in which all the tax bills are fused into one: Analogy is
compressed into identity . Disanalogy is compressed into change .
In the blended mental space, there is a single tax bill that changes
over time.
Metaphors typically result from double-scope integration
networks, whereas metonymy turns out to be the compression of
one vital relation into another .
Conceptual mappings are not prompted only by spoken or
signed language . Tey are part of human thought, communi-
cation, and interaction quite generally; they are signaled through
Mapping
466
and leveling toward the unmarked term in pidgins/creoles,
dialects, informal speech, and so forth. A further outgrowth has
to do with markedness hierarchies that are scalar in nature, such
as the noun phrase accessibility hierarchy for relativiza-
tion (e.g., from less marked to more marked role of the relative
pronoun in the relative clause, subject > direct object > indirect
object > prepositional object) .
Te formal ( generative ) approaches, especially univer-
sal grammar (UG), the principles and parameters plus
minimalist approach to syntax (Chomsky 1995), and optimal-
ity theory in phonology (Prince and Smolensky, 2004 ), focus
on competence and reject criteria for markedness related to
use ( performance ). Using cross-linguistic details of grammar,
UG has been posited; UG determines a set of possible core gram-
mars for languages by setting parameters , so that systems that
fall within a core grammar constitute the unmarked phenomena
and more marked elements are found in the periphery (see core
and periphery ) (Chomsky 1981 ). More recently, work has also
focused, for example, on expression of markedness relations by
constraints , on an explanation of markedness asymmetries
through constraint interaction, and the use of constraint forms
to express markedness hierarchies.
In both of these approaches, the unmarked category has also
at times been assimilated with the concept of naturalness, as in
natural phonology and natural morphology , as well as in optimal-
ity theory; some see it as overlapping with normality , regularity ,
generality , and productivity ; it has also been used, for example,
in studies of word order to defne the basic , dominant , or pre-
ferred WORD ORDER (e.g., subject-verb-object in English); and
it has certain elements in common with the notion of proto-
type . While not everyone uses the term markedness and some
linguists think that it is an unwieldy cover term with too wide a
range of application and no central defnition, others in both tra-
ditions see it as a major conceptual and explanatory tool that will
continue to be of interest and utility for understanding various
phenomena of language .
Linda Waugh
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Battistella , Edwin. 1990 . Markedness: Te Evaluative Superstructure of
Language . Albany : State University of New York Press . Battistellas two
books are the most accessible long treatments of the topic.
. 1996 . Te Logic of Markedness . New York : Oxford University Press .
Battistellas two books are the most accessible long treatments of the
topic.
Chomsky , Noam . 1981 . Markedness and core grammar. In Teory of
Markedness in Core Grammar , ed. A. Belletti , L. Branmdi , and I. Riozzi ,
12346. Pisa: Scuola Normale Superiore di Pisa.
. 1995 . Te Minimalist Program . Cambridge, MA : MIT Press .
Croft , William . 2003 . Typology and Universals . 2d ed.
Cambridge : Cambridge University Press .
Givon , Talmy . 1990 . Syntax: A Functional-Typological Introduction , 2 .
Amsterdam : John Benjamins .
Jakobson , Roman. 1990 . On Language . Ed., with an introduction by,
L. Waugh and M. Monville-Burston . Cambridge : Harvard University
Press .
Prince , Alan , and Paul Smolensky . 2004 . Optimality Teory: Constraint
Interaction in Generative Grammar . Malden, MA : Blackwell .
Nez , Rafael . 2005 . Creating mathematical infnities: Metaphor, blend-
ing, and the beauty of transfnite cardinals. Journal of Pragmatics
37 : 1717 41.
Nez , Rafael , and Eve Sweetser . 2006 . Looking ahead to the
past: Convergent evidence from Aymara language and gesture in the
crosslinguistic comparison of spatial construals of time. Cognitive
Science 30 : 401 50.
Shastri , Lokendra . 1996 . Temporal synchrony, dynamic bindings, and
SHRUTI a representational but non-classical model of refexive rea-
soning. Behavioral and Brain Sciences 19.2 : 331 7.
Sweetser , Eve . 1996 . Reasoning, mappings, and meta-metaphorical
conditionals. In Essays in Semantics and Pragmatics , ed. Masayoshi
Shibatani and Sandra Tompson , 22134 . Amsterdam : John
Benjamins .
Turner , Mark . 1991 . Reading Minds . Princeton, NJ : Princeton University
Press .
Williams , Robert . 2005 . Material anchors and conceptual blends in time-
telling . Ph.D. diss., University of California, San Diego.
MARKEDNESS
Te original insight of markedness was that many linguistic phe-
nomena consist of polar opposed pairs for example, the phono-
logical feature unvoicedvoiced and the grammatical relation
activepassive and that typically there is an asymmetry, such that
one term is more general and thus unmarked (given frst in the
examples) and the other is more constrained and thus marked.
Markedness was frst developed in phonology as an explanation
for asymmetries in phonological systems based on cross-linguistic
comparisons, with evidence from typology and universals for
example, more (unmarked) oral consonants than marked nasal
ones: Unmarked consonants occur in places of nonconditioned
neutralization (e.g., only unmarked voiceless consonants in word-
fnal position in Russian). Later, markedness was used to study
grammatical semantics (where the unmarked term has a larger
semantic range than the marked term), to explain the order of pho-
nological acquisition in child language (unmarked terms learned
before marked terms, e.g., stops before fricatives) and apha-
sia (marked terms lost before unmarked ones), and to identify
implicational universals , in which the presence of a marked
element implies the presence of the corresponding unmarked
element, but not vice versa (all of these in Jakobson 1990 ). Since
then, it has developed into an important (though controversial)
concept in other areas of linguistics, such as morphology, syn-
tax, lexical semantics, historical linguistics, second
language acquisition, stylistics , and so on.
Since about the 1960s, two substantially diferent approaches
to markedness have developed and with them diferent types of
evidence, and explanations, for markedness. Te FUNCTIONAL-
(-typological ) approaches (e.g., Givon 1990 ; Croft 2003 ), based
on earlier work in typology universals, depend on diagnostic
criteria, not only from linguistic systems but also from language
use , and these are related to functional criteria, communica-
tive needs, processing ef ciency, learnability, memory, and so
on. Criteria include zero or simple expression for the unmarked
term ( isomorphism ) unmarked singular with zero expression as
in cat versus marked plural with the marker -s in cats ; text (token)
frequency (unmarked term more frequent); contextual distribu-
tion (unmarked category has greater freedom of occurrence);
Markedness
467
language forms. Moreover, this mismatch is itself recognized by
individuals, albeit implicitly or unconsciously, who understand
the symbolic value of language. Language is, therefore, another
form of cultural capital in that it is symbolic in the way it both
values and is valued in terms ultimately related to the structure
of the feld. For Bourdieu, the most predominant feld structures
are those of social class, which also express the distribution of
power in society.
Tere are relations of linguistic production and authorized
language within the linguistic market. Moreover, everyone
enters the market in order to compete as a way of gaining and
sanctioning social prestige, and, consequently, status and posi-
tion, through the acknowledgment of others in the market. Value
is ascribed to individuals; it is not within their own capacity to
give it to themselves. Tere is a kind of anticipation and actu-
alization of profts, much in the same way as in any market (see
Bourdieu [1982] 1991 , 76 f). Bourdieu refers to many examples
where the power relation between two or more individuals is
expressed in the language they use with respect to one another.
For example, in the postcolonial context (see colonialism
and language ), those in a position of dominance sometimes
abdicate their position of authority by linguistically reaching
down to the interlocutor. However, he sees this as simply a
strategy of condescension aimed at reasserting their domina-
tion. Normally, it is the opposite that applies: Tose dominated
are forced to adopt the language of the dominant. Bourdieu also
contrasts the broken English of the black American vernacu-
lar with the air of naturalness of the English (1992, 143). For
Chomskyan linguistics, both are natural and unbroken since
they follow the same complex principles (e.g., binary merge
and wh- movement ). Te point is not only that power relations
are expressed in such linguistic exchanges but that the linguis-
tic market also defnes what is and is not linguistically valued by
rewarding and sanctioning specifc forms of language. In the-
ory, everything is available to all in the market. However, some
already hold specifc forms of linguistic capital, which they have
obtained from family background, education, and professional
trajectory. Moreover, such symbolic value is not only expressed
in language forms but also structurally homologous to other
forms of cultural capital; indeed, it can be found in physical body
gestures, as well as other forms of self-presentation. For those
without this capital, it is almost impossible to catch up .
Ultimately, such relations are expressed in political relations,
where certain individuals and representatives are endowed with
the power to sanction. For Bourdieu, these are acts of quasi-
magic as, through this endowment, power is literally invested
in someone by a formal acknowledge of status a form of social
consecration. A most obvious form of this phenomenon is when a
title is bestowed on an individual: Head of Department, for exam-
ple. Some who write similarly of the linguistic variation between
individuals conclude with a defcit model of language, whereby
lack of language competence is addressed through complemen-
tary education. Ultimately, this leads to a form of linguistic
communism where all are linguistically equal (see Bourdieu
and Boltanski 1975 ). However, the logic of the linguistic market
is that such compensatory measures will always give rise to dis-
appointing results in terms of social inclusion since, ultimately,
they go against the logic of practice constituting the feld the
Waugh , Linda , and Barbara Laford . 2000 . Markedness. In
Morphology: An International Handbook on Infection and Word-
Formation , I: 27281. Berlin : Walter de Gruyter . An accessible treat-
ment of the topic.
MARKET, LINGUISTIC
Pierre Bourdieu defnes a linguistic market as a system of rela-
tions of force which determine the price of linguistic products and
thus helps fashion linguistic production (1989, 47). If linguistic
habitus is the subjective element of habitus connected with
language use, linguistic market represents the objective field
relations. As always with Bourdieu, the two are in a constant state
of dynamic interrelationship, as well as evolving dynamically as
a part of the transformation of social structures .
In positing a concept such as linguistic market, Bourdieu is
targeting traditional linguistics. His quarrel is with all linguis-
tics going back to the work of Ferdinand de Saussure, which he
sees as treating language as an object of study rather than as a
practice. Te concept thus constitutes language as logos rather
than praxis. Bourdieus critique extends to Noam Chomsky and
Chomskyan linguistics, with its discovery a of semibiological
language acquisition device , deep syntactical structure
(see underlying structure and surface structure ),
and universal grammar . Bourdieu cannot accept the
Chomskyan precepts that linguistics should be concerned with
an ideal speaker-listener, a homogeneous speech community,
and perfect grammatical competence. Bourdieus alternative
can be summed up as follows:
In place of grammaticalness it puts the notion of acceptability ,
or, to put it another way, in place of the language ( langue ), the
notion of legitimate language. In place of relations of communica-
tion (or symbolic interaction) it puts relations of symbolic power ,
and it replaces the meaning of speech with the question of the
value and power of speech. Lastly, in place of specifcally linguis-
tic competence, it puts symbolic capital , which is inseparable
from the speakers position in the social structure. ( 1977 , 646;
italics in original)
In other words, Bourdieu is seeking to socialize, or at least soci-
ologicalize, all the major principles of traditional linguistics.
Te linguistic market is, therefore, essentially an expression
of linguistic relations. However, like all markets, not everyone
to be found within it is equal, and linguistic knowledge is never
perfect. In reality, some are found to have greater practical mas-
tery ( connaissance ). Tis knowledge is itself defned not simply
in terms of use but as an expression of legitimate language. In
most social contexts, there is a dominant language form. Tis
is most evident at a national level where there is received pro-
nunciation and other standard language forms. However, it can
extend to sublevels and categories and feld microcosms . In each
case, there is a right way of using language. Tis rightness is
defned by social common assent or common acknowledg-
ment. Te particularity of language is that while orthodox lan-
guage forms are maintained by this consensus and recognized
as such reconnaissance not all can use them. Tere can be a
mismatch between any one individuals connaissance and recon-
naissance , resulting from upbringing and proximity to legitimate
Market, Linguistic
468
like consciousness, only arises from the need, the necessity, of
intercourse with other men (Marx and Engels 1964 , 42) .
Te stress on language as central to human activity, or praxis,
indicates the important role that Marx and Engels gave it in their
account of the distinctiveness of human life. Language forms an
essential part of the evolving process by which human beings in
social relationships create historical reality through the negotia-
tion of material needs and the requirement for self-reproduction.
It is important to note, however, that language was not viewed as
either primary or derivative; it was not the faculty that enabled
human beings to become social in the frst place, nor was it the
means by which they could express themselves once they had
been socialized. Instead, it was an aspect of the social, material
activity labor in its general, technical sense by which human
beings were constituted qua human beings and by which they
acted upon nature and other human beings in order to create
history.
Within the Marxist tradition, the stress on the constitutive
aspect of language as a form of labor material practice was
almost lost, as the term labor itself became narrowly conceived
simply to mean certain types of work. As a result, more atten-
tion was paid to other statements by Marx and less to his original
focus on language as social activity. Tese comments included
his reference to the existence of a bourgeois form of language
(Marx and Engels 1964 , 249), his assertion that ideas do not
exist separately from language ( 1973 , 163), and his declaration
that the ideas of the ruling class are in every epoch the ruling
ideas ( 1964 , 60). Marxs remarks, which amount essentially to
the observations that the language in use is afected by the class
relations that hold in a given social formation and that ideology
is disseminated in language, were again rather narrowly inter-
preted within orthodox Marxism.
In the Soviet Union, in particular, a whole set of somewhat
fruitless debates ensued as to whether language belonged to the
base or superstructure of society. For N. S. Marr , for example,
languages were stratifed in such a way that between communi-
ties employing distinct languages, the speech of the same class
would be closer than the speech of diferent classes using the
same language. In this account, language belongs to the social
superstructure of society, which is simply determined by class;
the idea that the unity of a group not based on class (such as the
nation) could be explained by the idea of a common language
was dismissed. Marrs infuence, which was widespread in the
1930s and 1940s, was ended by Stalins equally dogmatic dec-
laration in Marxism and the problems of linguistics [ 1950 ]
1974) that languages did not have a class character but rather
a national character and were thus not part of the superstruc-
ture. Despite the title of Stalins piece, and though it was an
important correction to the misleading efect of Marrs theories,
it did not represent any sort of breakthrough in the Marxist treat-
ment of language .
In fact, precisely such an advance had been heralded in the
writings of a number of linguists in the Soviet Union primarily in
Vitebsk and Leningrad which, in efect, amounted to a school of
Marxist linguistics. Because of the terror exercised by Stalinism,
the exact membership of this group is unknown and the names
used for publishing may or may not be those of the authors of
market in the frst place. Just as communist alternatives to capi-
talism eventually collapsed, leading to an embracing of liberal
economics and free-market principles, so linguistic communism
cannot work since it runs counter to the raison dtre of the lin-
guistic market, which, in terms of substantive cause and efect,
is social diferentiation. However, this should not be seen as a
form of poststructuralist nihilism; rather, Bourdieu is ofering a
metanoia , a new gaze or way of looking at the world through
his epistemological thinking tool (see Grenfell 2004 , Chapter 7) .
Michael Grenfell
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bourdieu , Pierre. 1977 . Te economics of linguistic exchanges. Social
Science Information 16.6 : 645 68.
. [1982] 1991. Language and Symbolic Power. Trans. G. Raymond
and M. Adamson . Oxford: Polity Press .
, with Loc Wacquant. 1989. Towards a refexive sociology: A work-
shop with Pierre Bourdieu. Sociological Teory 7.1 : 26 63.
, with Loc Wacquant. 1992. An Invitation to Refexive Sociology .
Trans. L. Wacquant. Oxford : Polity Press .
Bourdieu , Pierre , and Luc Boltanski . 1975 Le ftichisme de la langue.
Actes de la recherch en sciences sociales 2 : 95 107.
Fehlen , Fernand . 2004 Pre-eminent role of linguistic capital in
the reproduction of the social space in Luxembourg. In Pierre
Bourdieu: Language, Culture and Education , ed. M. Grenfell and M.
Kelly , 6172. Bern : Peter Lang .
Grenfell , Michael . 1993 . Te linguistic market of Orlans. In France:
Nation and Regions , ed. M. Kelly and R. Bock , 7299. Southampton, UK :
ASM & CF.
. 2004 . Agent Provocateur: Pierre Bourdieu . London : Continuum .
Snook , Ivan . 1990 . Language, truth and power: Bourdieus Ministerium.
In An Introduction to the Work of Pierre Bourdieu , ed. R. Harker ,
C. Mahar , and C. Wilkes , 16079. Basingstoke : Macmillan .
MARXISM AND LANGUAGE
Te aim of Marxism is to understand history and society accord-
ing to the precepts frst outlined in the works of Karl Marx and
Friedrich Engels, later developed by other thinkers in this tra-
dition, in order to efect revolutionary social change. Given the
fact that Marxism is in part a description of the determinants of
everyday life as a way of explaining the social order, it is some-
what surprising, therefore, to note that the Marxist contribution
to thinking about language has been limited. Tis omission has
been unfortunate both for Marxism and for those nonformal-
ist accounts of language that stress its historicity in a general
sense and its specifc and variable links to particular social
formations.
In Te German Ideology , as part of their attack on philosophi-
cal idealism, Marx and Engels provide a sketch of their material-
ist conception of history. With regard to the nature and function
of language, they assert:
From the start the spirit [mind] is af icted with the curse of
being burdened with matter, which here makes its appear-
ance in the form of agitated layers of air, sounds, in short, of
language. Language is as old as consciousness, language is prac-
tical consciousness that exists also for other men, and for that
reason alone it really exists for me personally as well; language,
Marxism and Language
469
theorists concerned themselves directly with language, and
when they did, as in the case of Walter Benjamin or Jean-Paul
Sartre, it is dif cult to see how the work qualifes as Marxist in
any recognizable sense. Yet a number of Marxist theorists, such
as Ferrucio Rossi-Landi ( 1983 ), Terry Eagleton ( 1982 ), and Jean-
Jacques Lecercle ( 2006 ), have produced interesting work based
on Voloinovs text. More signifcantly, it was the inspiration for
much of the later work of Raymond Williams, the major British
socialist critic of the twentieth century.
Williamss chapter on language in Marxism and Literature
( 1977 ) stressed the importance of Voloinovs theory of signif-
cation, both in general and for his own original work on histori-
cal semiotics in Keywords (1976) . Beginning with Voloinovs
argument that signs are neither expressive nor systematic in any
simple sense but, rather, communicative media deployed in the
social process of making history, Williams stressed that signs are
shaped by past use but are engaged at the same time in the cre-
ative making of the present (and are thus of necessity open to
the future). Tis idea of the historical variability of signs, which
Voloinov calls their mulitaccentuality, formed the basis of
Williamss investigation of the vocabulary of a number of discur-
sive felds, centrally those that involved discussion of culture and
society. In essence, what he provides in Keywords and Marxism
and Literature is a retrospective theoretical account of his work
in Culture and Society (1958) , a text that efectively began the
debates that led to the appearance of cultural studies as an aca-
demic discipline. Tough rarely acknowledged as such, it was an
historical materialist approach to language that lay at the base of
this important intellectual development.
Marxs comments on the existence of bourgeois language
and Voloinovs assertion that the sign becomes an arena of
the class struggle ([1929] 1973 , 23) point to another feld of
research in which Marxist thought has been signifcant: the
politics of language, with particular regard to the historical
construction of national languages, the class-based hier-
archy of language within education, and the role of language
in imperialism and colonialism . Important work in this area
was conducted by Antonio Gramsci , the Italian Communist
Party intellectual and leader, who drew attention to the class
perspective in his discussion of the merits and demerits of the
use of dialect versus a national form of language in political
struggles in Italy. Other examples include Rene Balibars his-
torical research on the emergence of a standard language in
France in Les franais fctifs ( 1974 ) and Linstitution du franais
( 1985 ) , and Tony Crowleys related work in the British con-
text in Te Politics of Discourse ( 1989 ) . Writing from the post-
colonial conjuncture, the Kenyan writer Ngg Wa Tiongo
used a Marxist approach to denounce the colonial linguistic
legacy in his Decolonising the Mind ( 1986 ) . And in educa-
tional debates, Basil Bernsteins theory of restricted and
elaborated codes attempted to explain the diferential aca-
demic achievement of children from diferent social classes. In
Reproduction in Education, Society and Culture , written with
Jean-Claude Passeron (1977), and Language and Symbolic
Power ( 1992 ), Pierre Bourdieu used a neo-Marxist framework
to account for the same phenomenon .
Tony Crowley
the works. Nevertheless, the principal texts are recognized as
V. N. Voloinovs Marxism and the Philosophy of Language , pub-
lished in 1929 and translated in 1973; Mikhail Bakhtins Problems
in Dostoyevskys Poetics , published frst in 1929 and translated
from the second (1964) edition in 1984; and P. N. Medvedevs Te
Formal Method in Literary Scholarship: A Critical Introduction
to Sociological Poetics , published in 1928 and translated in 1978.
Despite the fact that the work of Bakhtin is the best known to
readers in the West, the most signifcant contribution to a strictly
Marxist treatment of language was provided by Voloinovs pio-
neering text.
Te radical thrust of Voloinovs work came in his opposition
to two key tendencies that he identifed in thinking about lan-
guage: individualistic subjectivism and abstract objectivism.
Te frst, traced by Voloinov to the German idealist tradition and
articulated most clearly in the work of Wilhelm von Humboldt ,
takes the individual human psyche as the most important site
of linguistic production and focuses on the individual creative
act of speech. Regarding speech as a type of aesthetic cre-
ativity , this approach rejects language, understood as a
fxed system, as simply the product of the abstract methods of
linguistics. Te second tendency, abstract objectivism, is the
binary opposite of the frst and is typifed in the model proposed
by Ferdinand de Saussure and developed by structuralism .
In this approach, the static and apparently immutable linguistic
system is divorced from history, is distinguished rigorously from
individual instances of language use, and is considered to be
composed of nothing other than the normatively identical forms
of lexis, grammar, and phonetics. If the frst focuses on the
unceasing process ( energeia ) of individual linguistic creativity,
then the second treats language as a fnished product ( ergon ),
open to the objective gaze of the science of linguistics .
For Voloinov, the concentration on individual conscious-
ness as the basis of an explanation of linguistic signifcation
is a mistake. Te individual consciousness cannot serve as the
foundation of linguistic analysis because it is itself in need of
explication from a social point of view: [C]onsciousness takes
shape and being in the material of signs created by an organized
group in the process of its social intercourse nurtured on signs,
it derives its growth from them; it refects their logic and laws
(Voloinov [1929] 1973 , 13). Tis does not, however, mean that
the individual consciousness is formed by and in the normatively
identical signs of the abstract objectivist system. On the contrary,
Voloinovs point is that signs themselves, as dynamic com-
plexes of form and meaning, are not simply presented as given,
fxed elements of a system but are open products of the activ-
ity the material practice of language making between socially
organized individuals. Language, in this sense, is not the middle
term that unites the individual and the social, nor is it a medium
that refects a preexistent reality. Instead, it is an aspect of the
constitutive social activity labor, in Marxs original sense that
allows for the very possibility of the individual, the social,
and reality itself.
Despite the importance of semantic indeterminacy to post-
structuralist literary theory, and the stress on context in linguis-
tic pragmatics, the radical challenge of Voloinovs work has not
been taken up widely in twentieth-century thinking on language.
Even in the tradition of Western Marxism, few of the major
Marxism and Language
470
sum, meaning and belief are diferent (the distinctness compo-
nent of the traditional view), and meaning allows the articulation
of belief (the expressive component) .
In the last half-century or so, this picture of the relation
between meaning and belief has been challenged from a num-
ber of perspectives. One important challenge concerns the
interaction of meaning and belief, addressing such questions as
whether meaning is a relatively neutral vehicle for expressing
belief or something that may afect belief. Tis is the challenge of
linguistic relativism . Another concerns the validity of the
division between meaning and belief. Tis is the critique of ana-
lyticity . Te two challenges point in somewhat contradictory
directions . (Tese are not by any means the only ways in which
meaning and belief have been discussed in recent years. For
example, Akeel Bilgrami [1992] addresses the issue of how to rec-
oncile meaning externalism with certain subjective aspects
of belief. Unfortunately, a short entry can only point to a couple of
key issues that have arisen in connection with this broad topic.)
As to linguistic relativism, a number of writers (most famously
Edward Sapir and Benjamin Lee Whorf) have argued that mean-
ing is not simply a means for articulating belief, but a means
of shaping belief (as well as emotion, action, even perception).
A popular version of this view is developed in George Orwells
novel 1984 , where the government seeks to control peoples
ideas by changing their language. Te idea of any conceptual
scheme relativism is dif cult to sustain in global terms, as writers
such as Donald Davidson ( 1984 ; see Chapter 13) have pointed
out. However, it is clear that we do not have at least some beliefs
before we have some categories, and commonly those categor-
ies go along with words and meanings. For example, small chil-
dren do not have beliefs about gravitation or terrorism because
they do not have the relevant concepts , and the concepts are
presumably something they acquire by learning the words and
their meanings. More importantly, it seems likely that peoples
beliefs about particular events are afected by the concepts (thus,
meanings) available to and salient for them. Tus, without the
concept of terrorism , perhaps Americans would have under-
stood the events of September 11, 2001, as crimes. Tis would
have changed their beliefs about the nature of the event, proper
responses to the event (e.g., police investigation, extradition,
etc., rather than war), and so on.
Tis challenge to the expressive component of the common-
sense view seems to preserve the distinctness component. In
other words, it seems to rely on a presumption that meaning and
belief are diferent. After all, if meaning and belief are not dis-
tinct, then it is dif cult to tell exactly how meaning could afect
belief. Tis division between meaning and belief is precisely what
is challenged by the critique of analyticity.
Tere are two clear ways in which the meaning/belief divi-
sion may be criticized. Tey relate to two obvious ways in which
the division itself may be formulated. One way concerns revis-
ability. We might say that the belief component of our ideas
is revisable by reference to experience or facts. In contrast, the
meaning component is steady in the face of new experiences
or facts. However, W. V. Quine has argued famously that there
is no way of putting some truths into empirical quarantine and
judging the remainder free of infection. Tus meaning and
sensory evidence are inextricably intertwined (1976, 139; see
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bakhtin , M. M. [1929] 1984. Problems in Dostoyevskys Poetics. Ed. and
trans. Caryl Emerson . Manchester, UK : Manchester University Press .
Balibar , Rene . 1974 . Les franais fctifs . Paris: Hachette.
. 1985 . LInstitution du Franais . Paris : Presses Universitaires de
France .
Balibar , Rene , and Dominique Laporte . 1974 . Le Franais National .
Paris : Hachette .
Bernstein , Basil . 1971 . Class, Codes and Control . Vol. 1. London : Paladin.
Bourdieu , Pierre . 1992 . Language and Symbolic Power . Ed. John
Tompson , trans. Gino Raymond and Matthew Adamson. Cambridge,
UK: Polity.
Bourdieu , Pierre , and Jean-Claude Passeron . 1977 . Reproduction in
Education, Society and Culture . London : Sage .
Crowley , Tony . 1989 . Te Politics of Discourse: Te Standard Language
Question in British Cultural Debates. Houndmills : Macmillan.
Eagleton , Terry . 1982 . Wittgensteins Friends. New Left Review
1.135 : 64 90.
Gramsci , Antonio . 1985 . Selections from Cultural Writings . Ed. David
Forgacs and William Nowell-Smith , trans. William Boelhower .
London : Lawrence and Wishart.
Lecercle , Jean-Jacques . 2006 . A Marxist Philosophy of Language .
Boston : Brill .
Marx , Karl , and Friedrich Engels . 1964 . Te German Ideology .
Moscow : Progress Publishers .
. 1973 . Grundrisse . Trans. Martin Nicolaus . Harmondsworth :
Penguin .
Medvedev , P. N. [1928] 1978. Te Formal Method in Literary Scholarship: A
Critical Introduction to Sociological Poetics . Trans. Albert J. Wherle .
London : Johns Hopkins University Press.
Ngg Wa Tiongo . 1986 . Decolonising the Mind: Te Politics of Language
in African Literature . London : James Currey .
Rossi-Landi , Ferruccio . 1983 . Language as Work and Trade . Trans.
Martha Adams . South Hadley, MA : Bergin and Garvey.
Stalin , Joseph. [1950] 1974. Marxism and the problems of linguistics.
In A Primer of Linguistics , ed. Anne Fremantle , 20318. New York :
St. Martins.
Voloinov , V. N. [1929] 1973. Marxism and the Philosophy of Language .
Trans. L. Matejka and I. R. Titunik. London : Seminar Press.
Williams , Raymond. 1958 . Culture and Society 17801950 . London : Chatto
and Windus .
. 1976 . Keywords: A Vocabulary of Culture and Society . London :
Fontana.
. 1977 . Marxism and Literature . London : Oxford University Press .
MEANING AND BELIEF
Tere is a commonsense view of the relation between meaning
and belief that has been tacitly presupposed in many philosophi-
cal, linguistic, and other treatments of these topics. It runs some-
thing like this. words refer to objects and have defnitions. Te
defnitions represent properties of the objects, provide criteria
for using the words to refer to objects, and allow understanding
of such uses. Tus man means, roughly, adult, human, male.
Tat meaning picks out properties of some objects, isolating men
(rather than women, children, apes, and so on). In keeping with
this, the meaning allows us to use the term to refer to certain sets
of objects (men) or particular members of that set (individual
men) and to understand what other people refer to when they use
the term. Additionally, meanings allow us to express our beliefs
about members of sets of objects (generally or individually). In
Meaning and Belief
471
other words, understanding father involves various schemas
that cluster information into relations. Tese schemas have
default values (such as father is mothers husband), perhaps
along with specifed alternatives to defaults (such as father is
divorced from mother). Tis information is hierarchized in that
we generally consider items higher in the hierarchy to be more
criterial for application of the term than items lower in the hier-
archy. Put very simply, if we fnd out that Peter is Sallys progeni-
tor but is not the breadwinner, we are more likely to count him
as Sallys father than if we fnd out that he is the breadwinner but
not the progenitor. On the other hand, hierarchy efects are not
absolute. We may be more inclined to apply father to the bread-
winning, afectionate, live-in husband of Sallys mother than to
an unknown progenitor. (Te last point, if developed further,
would lead us to the place of prototypes in lexical seman-
tics. However, the inclusion of prototypes or for that matter,
exemplars would not afect the main argument as it bears on
meaning and belief.)
Insofar as this model of meaning is accurate, it suggests,
frst of all, that there is no sharp meaning/belief division. Tere
does not seem to be any point at which the information associ-
ated with a given heading stops being semantic and starts being
empirical. On the other hand, it also suggests that the meaning/
belief division is not wholly pointless in that there does appear
to be a continuum from more defnition-like information to
more observation-like information. But this, too, is not all.
Te hierarchical continuum is not determinative. We may think
of the hierarchy as a series of weighted properties and/or rela-
tions. Although those higher in the hierarchy are more heavily
weighted, they may be outweighed by a large enough number of
lower-level properties/relations. Alternatively, in connectionist
terms, a large number of weak connections may reach some acti-
vation threshold that is not reached by a small number of strong
connections. Tis last point suggests that despite the hierarchy,
all information associated with lexical items is in some ways
more akin to belief than to meaning (though perhaps neither
term is truly adequate here) .
As the preceding reference to connectionism suggests, the
same conclusions hold for accounts of meaning that rely on neu-
ral networks. For example, neural accounts treat meaning as a
complex of circuits linking confgurations of neurons in difer-
ent areas of the brain insofar as these bear on the sound of the
relevant word, the appearance of the referent, our own actions
as they might bear on the referent, and so on (see, for example,
Chapter 4 of Pulvermller 2002 ). Tese circuits are presumably
not fully fxed and identical across all uses. Rather, the precise
confguration activated at any given moment will vary, depend-
ing on what other neural circuits are simultaneously activated.
For example, suppose I say squeeze. Tat activates circuits that
include neuron populations that govern closing together the fn-
gers of the dominant hand. Suppose I then say ball. Te, so to
speak, resting circuit for ball includes a range of neuron popu-
lations, some of which bear on closing together the fngers of the
dominant hand. Since some part of the latter population was just
activated by squeeze, it should be more fully activated by ball. Te
prior activation due to squeeze will slightly alter the circuit acti-
vated by ball, perhaps enough to make one think of hand-sized,
rubber balls .
also 1981, 67, 712). Tis suggests a number of things about the
relation between meaning and belief. Perhaps most obviously,
it indicates that sentences are not true by their meanings alone.
More importantly for our purposes, it suggests that what we con-
sider meanings are open to empirical revision, precisely in the
manner of beliefs. For instance, a hundred years ago, one might
have thought that My father is a man was true analytically.
However, sex-change operations have shown us that the mean-
ing component male is revisable due to empirical information
about the objects to which father refers. If some idea about an
object or set of objects is revisable due to empirical information
about referents, then it would seem to count as a belief, not as a
meaning.
Tere are undoubtedly cases where we would fnd it dif cult
to imagine such revision. For example, I have no good idea for
how I could possibly revise my view that If Jones is currently a
man, then Jones is currently a male, adult, human. (Obviously,
there are scenarios where the meaning of man could change, but
that is not at issue.) However, it might be argued that this tells us
something about my imagination, not about the facts. Perhaps
it was impossible for people to imagine sex-change operations a
century ago. It may be that what we consider to be meaning is a
function of what we can imagine changing. But our imagination
could always be mistaken.
On the other hand, perhaps the obvious cases of revision are
not so obvious as they initially appear. For example, people did
imagine men changing into women and women changing into
men well before sex-change operations. If Tiresias had a child
as a man, then was transformed into a woman, his child could
truthfully say, My father is a woman. So perhaps male was
always only more limitedly part of the meaning of father , closer
to a belief than we recognize. In this way, it may be that the revis-
ability argument is not defnitive .
A more productive approach may still be Quinean in orien-
tation naturalizing our treatment of the topic, as Quine
often urged ( 1969 ; see Chapter 3) by turning to the natural sci-
ences. Here, we might consider two sorts of cognitive architec-
ture that are common in discussions of meaning today. (Tis,
of course, is not Quinean as it is mentalistic.) Te frst is inten-
tional/representational; the second relies on neural networks,
either artifcial (see connectionism, language science,
and meaning ) or natural (see semantics, neurobiology
of ) .
A standard intentional/representational account of lexical
semantics involves headings, some sort of meaning units con-
nected with the headings, and connections across headings.
Te connections across headings establish lexical relations
of various sorts, including semantic fields . Te semantic
units themselves are structured into complexes of relations
with default values and are typically hierarchized, such that
some units are more important than others. Consider, for
example, man . Tis entry is linked to woman for one domain
(adult human), to boy for another domain (male human),
and so on. It includes a range of information, comprising not
only defnitional components, but empirical components as
well. For example, 50 years ago, father included not only male,
adult, human, and progenitor of ego but probably hus-
band of mother and breadwinner of the family or the like. In
Meaning and Belief
472
of items to which the noun refers (see intension and exten-
sion ). One can only adjudicate the defnition of a term by refer-
ence to an extension. For example, consider a defnition of U.S.
state that involves the criterion of continuous land. One can
reject this defnition by pointing to Hawaii, which is part of the
extension of U.S. state . But one can only adjudicate an extension
by reference to a defnition. In other words, we rely on a defni-
tion of U.S. state to judge that Hawaii is a U. S. state. Tus, one
cannot adjudicate a defnition and an extension simultaneously.
One of the two has to be established arbitrarily. By this argument,
there is no such thing as the real meaning of any term, includ-
ing meaning . Meaning may be social, intentional, or whatever,
as we choose in particular contexts. Tus, whenever we engage
in an interpretive task, the type of meaning at issue should be
stipulated.
Tis argument disposes of one problem what meaning
really is. But it leads us to three other concerns.
Te frst is ontological just what sorts of meaning actually
exist. We may, for example, stipulate Platonic meaning as our
object of hermeneutic interest (see philology and herme-
neutics ). But we cannot actually interpret for Platonic mean-
ings if they do not exist.
Te second concern takes up the demarcation of our stipula-
tive categories. Tese need to be adequately precise. For exam-
ple, we might stipulate that we are concerned with intentional
meaning. But there are numerous sorts of intentional meaning
that should often be distinguished in the case of legal inter-
pretation, the self-conscious intent of the author who drafted a
piece of legislation, the intents of the legislators who passed it,
the intents of the judges who gave opinions on its constitutional-
ity, and so on .
Te fnal concern bears on the particular purposes for which
we are interpreting. For example, for any given term in a law,
there may be variable social meanings. Ordinary people may use
a term with one meaning; scientists may use it with a slightly dif-
ferent meaning. In particular cases of interpretation, the meaning
associated with one or the other group may be more signifcant.
Note that in these cases, we are not trying to determine the real
meaning of the law. Rather, we are acknowledging that there are
many sorts of meaning and we are trying to determine which is
the most important in the case at hand .
Patrick Colm Hogan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bundgaard , Peer, and Frederik Stjernfelt. 2009. Patrick Hogan
[Interview]. In Signs and Meaning: 5 Questions , 7185. Copenhagen:
Automatic Press/VIP.
Hogan , Patrick Colm . [1996] 2008. On Interpretation: Meaning
and Inference in Law, Psychoanalysis, and Literature . 2d ed.
Athens : University of Georgia Press .
Levinson , Sanford , and Steven Mailloux . 1988 . Interpreting Law and
Literature: A Hermeneutic Reader. Evanston, IL : Northwestern
University Press .
MEANING EXTERNALISM AND INTERNALISM
Hilary Putnam ( 1975 ) argued for a view now known as mean-
ing (or semantic) externalism the view that there are terms
Of course, not everything is equally variable. Tere are some
connections in these networks that are stronger than others.
Tese diferences in connection strengths should correspond
roughly with the hierarchy of properties/relations in the inten-
tional/representational account. Here too, then, we have reason
to believe that there is some sort of continuum. Not all of our
ideas about a set of objects are equally salient, expected, and so
forth. However, none seems precisely to qualify as a meaning, to
be distinguished from a belief and, once again, a greater degree
of activation bearing on initially weaker connections may have
greater efects than a weaker activation bearing on initially strong
connections. Here too, then, any correlates we may posit for the
neuronal circuits seem more like beliefs than like meanings.
In conclusion, we might return briefy to linguistic relativ-
ism. If the preceding discussion of meaning/belief (non)distinct-
ness is accurate, then it seems that we cannot reasonably say
that meanings guide beliefs. We can only say that some beliefs
afect other beliefs. On the other hand, we also cannot say that
meanings simply allow us to express beliefs. Our ideas about the
world and our production and reception of language are, rather,
dynamic (neurocognitive) processes. Tese processes do not trap
us in a prison house of language (as some writers have put it).
But they also do not allow us some simple freedom to describe
and evaluate the world in abstract removal from the perception,
memory, and other circuits that are already in place when we
come to formulate our descriptions and make our evaluations .
Patrick Colm Hogan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bilgrami , Akeel. 1992 . Belief and Meaning: Te Unity and Locality of
Mental Content . Cambridge, MA : Blackwell .
Davidson , Donald. 1984 . Inquiries into Truth and Interpretation .
Oxford : Clarendon Press .
Pulvermller , Friedemann . 2002 . Te Neuroscience of Language: On Brain
Circuits of Words and Serial Order . Cambridge : Cambridge University
Press .
Quine , W. V. 1969 . Ontological Relativity and Other Essays . New
York : Columbia University Press .
. 1976 . Te Ways of Paradox and Other Essays . Rev. ed. Cambridge :
Harvard University Press .
. 1981 . Teories and Tings . Cambridge : Harvard University Press .
MEANING AND STIPULATION
Meaning is commonly understood as social, mentalistic, or
abstract. A social account views meaning as existing in social
groups. A mentalistic account places meaning in individual
minds. An abstract account locates meaning in a Platonic realm.
Discussions of meaning often involve debates about which
of these gives the real meaning of a term or utterance. For
example, in legal interpretation , there have been debates
between writers who view the Constitution as an ongoing prod-
uct of social developments and those who see it as fxed by the
Framers intent .
A stipulative account of meaning argues that such debates are
pointless. Tey are, in efect, debates over the real meaning of
the word meaning . Formally, the meaning of any common noun
(such as meaning ) involves a defnition and an extension or set
Meaning and Stipulation Meaning Externalism and Internalism
473
precisely the same conditions. So they do not mean the same
thing by their utterances, and this diference in meaning appears
traceable to the term water . Water in English has a diferent
extension from water in Twin Earth English. Hence, since mean-
ings are (at least) intensions, and intensions determine exten-
sions, water means something diferent in Oscar
1
s mouth than
it does in Oscar
2
s. Externalism vindicated .
As Colin McGinn ( 1977 ) and Tyler Burge ( 1979 ) have pointed
out, the same thought experiment can be used to challenge the
view that beliefs and other propositional attitudes are
narrow psychological states. When Oscar
1
says Water is odor-
less, he is expressing one of his beliefs, but this belief is difer-
ent from the belief Oscar
2
expresses via the same sentence. (Te
two beliefs are true under diferent conditions; that, according to
most theorists, suf ces to distinguish them.) Similar points could
be made about the other propositional attitudes. If we continue
to assume that there are narrow psychological states and that the
narrow psychologies of Oscar
1
and Oscar
2
could be identical, it
will follow that at least some beliefs and other propositional atti-
tudes are not narrow.
Under the sway of Putnam ( 1975 ), McGinn ( 1977 ), and Burge
( 1979 ), however, many theorists are now skeptical that there
are any narrow psychological states. Even sensations and other
phenomenally conscious psychological states narrow psycho-
logical states if any such there be have recently been argued
to be examples of wide psychological states. (See Dretske 1996 ;
Lycan 2001 ; and Tye 1995 .) Putnam himself (1996) avows skep-
ticism about narrow psychological states. What becomes of the
thesis of meaning externalism if we suppose that narrow psy-
chological state is an empty term? If an ordinary human subject,
S, lacks narrow psychological states, then S's narrow psychol-
ogy determines the meanings of S's terms is true, but vacuously.
To avoid this hollow victory for internalism, it is perhaps best to
recast the distinction between meaning externalism and inter-
nalism in terms of a distinction between intrinsic and extrinsic
properties: An intrinsic property is one that an object possesses
independently of its relations to other objects, whereas an e xtrin-
sic property is one that an object possesses in virtue of its rela-
tions to other objects .
Given the intrinsic/extrinsic properties distinction, we can
reformulate meaning externalism as the view that there are
terms whose meanings are not determined by their users intrin-
sic properties (regardless of whether there are any intrinsic psy-
chological properties) .
Max Deutsch
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Burge , Tyler . 1979 . Individualism and the mental. Midwest Studies in
Philosophy 4 : 73 122.
Dretske , Fred . 1996 . Phenomenal externalism. In Philosophical Issues .
Vol. 7. Ed. Enrique Villenueva , 14358 . Atascadero, CA : Ridgeview.
Lycan , William. 2001 . Te case for phenomenal externalism.
Philosophical Perspectives 15 : 17 35.
McGinn , Colin . 1977 . Charity, interpretation, and belief. Journal of
Philosophy 74 : 52135.
Putnam , H . 1975 . Te meaning of meaning. In Mind, Language,
and Reality: Philosophical Papers , II: 21571. Cambridge : Cambridge
University Press .
whose meanings are not determined by their users psycholog-
ical states. Meaning internalism is simply the denial of meaning
externalism .
Putnam qualifes his meaning externalism by explaining that
he intends psychological state to be understood in the nar-
row sense, according to which a psychological state implies the
existence of nothing but its possessor (1975, 21922). Another
equally signifcant qualifcation is that the argument will show
that meanings just aint in the head, as Putnam memorably
puts it, only if a terms meaning is taken to be (at least) an inten-
sion, that is, a function from possible circumstances (or worlds)
to its extension, or the set of objects to which the term applies
(Putnam 1975 , 227; see intension and extension, refer-
ence and extension , and possible worlds semantics ) .
Although there is a good deal of contemporary skepticism
about the existence of narrow psychological states (see the fol-
lowing), Putnam assumes that at least some psychological states
are narrow; beliefs, thoughts, feelings, and interior monologue
are all given as examples ( 1975 , 224).
Putnam argues for meaning externalism with his famous
Twin Earth thought experiment (1975, 2237). Suppose that
somewhere in the galaxy there is a planet, Twin Earth, which is
just like Earth save one detail: Te liquid that fows from Twin
Earthian faucets, falls from Twin Earthian skies, and flls Twin
Earthian oceans is not water. It is macroscopically identical to
water, but, unlike water, it is not the chemical compound H
2
O.
Instead, it is some other complicated chemical compound
that can be abbreviated XYZ. Twin Earthians speak a dialect of
English, and Earthians and Twin Earthians both use the term
water , but the extension of water in their respective dialects is
diferent. In English, water applies to all and only samples of
H
2
O. In Twin Earth English, water applies to all and only sam-
ples of XYZ.
Now consider two subjects, Oscar
1
, an Earthian, and Oscar
2
,
a Twin Earthian, both of whom have interacted with, and have
beliefs and other psychological attitudes concerning, the water-
like liquid native to their respective planets. Suppose both to be
living in 1750, before anyone on their planets knew anything
about the underlying chemistry of the liquids found thereupon.
Putnam claims that it is possible for Oscar
1
and Oscar
2
to be in
the same narrow psychological state (1975, 224). Since both are
chemically unsophisticated, neither has beliefs characterizable
with H
2
O or XYZ that could potentially distinguish their nar-
row psychologies. Furthermore, given the macroscopic identity
between H
2
O and XYZ, it seems plausible to suppose that all of
Oscar
1
s attitudes, feelings, and sensations about the liquid that
is in fact H
2
O on his planet could be matched by exactly simi-
lar attitudes, and so on, of Oscar
2
s toward the liquid that is in
fact XYZ on his planet. Indeed, as Putnam suggests, Oscar
1
and
Oscar
2
could well be molecule for molecule Doppelgngers ,
thus, it would seem, guaranteeing their narrow psychological
identity (1975, 227).
When Oscar
1
says Water is odorless, however, does he
mean what Oscar
2
means when he says Water is odorless? It
seems not. For what Oscar
1
says is true if and only if H
2
O is odor-
less, while what Oscar
2
says is true if and only if XYZ is odorless.
If, however, Oscar
1
and Oscar
2
mean precisely the same thing
by their utterances, then those utterances would be true under
Meaning Externalism and Internalism
474
be lengthy, involving several stages and many revisions. And
writing especially in printed or other permanent forms may
be received in quite diferent contexts from those in which it was
produced. Te writer must anticipate how the efects of a dis-
placed or unknown context might guide interpretation or lead
to misinterpretation. And readers, of course, must typically rely
on the written text alone in arriving at its sense. Writing, conse-
quently, is forced to be less reliant on its immediate context for
its meaning.
Speech is often treated as the primary medium of communica-
tion and this for various reasons. In human evolutionary terms, it
is broadly universal and involves specifc biological adaptation
unlike writing, which emerges as the product of particular his-
torical societies and is not universal either within or across them.
Speech is acquired during a critical language-learning period
(see critical periods ) very early in life. Writing is acquired
later and usually as the focus of explicit instruction. Nonetheless,
with the advent of a range of alternatives to speech and writing
as media of communication, it is dif cult to insist upon a simple
dichotomy between oral, situated, face-to-face communication,
one the one hand, and visual, decontextualized, noninteractive
communication, on the other especially when technological
developments in communication media are considered .
We may distinguish broadly among three overlapping phases
in the development of alternatives to speech as media of commu-
nication: mechanical (writing, print), electrical (telegraphy and
wireless telegraphy, radio, and television) and digital (World
Wide Web and the Internet, cellular phones, and the convergence
or interaction between these and previous media of communica-
tion). Developments in communication at a distance for military
and commercial purposes using semaphore and other fag sig-
naling systems are particularly evident in Europe in the late eigh-
teenth and early nineteenth century. Tese were forerunners of
the electric telegraph initially designed by Samuel Morse in the
1830s. Te use of electrical impulses to make possible commu-
nication at a distance then underpins the development of the
telephone in the 1870s, and forms of wireless telegraphy in the
1890s, to be followed by radio and television broadcasting in
the frst and second half of the twentieth century. In most cases,
each technological development may be seen to favor particular
linguistic selections over others. Te telegraph and subsequent
telegram, because of the cost of transmission and the premium
placed on time, tended to favor certain kinds of abbreviation
principally the deletion of grammatical function words, such as
articles, determiners, and verb auxiliaries.
Te early electrical media of communication at a distance tend
to be dyadic and reciprocal, rather than one-way and noninterac-
tive. But many of the subsequent and most far-reaching develop-
ments in communication at a distance in the twentieth century
tend to be one-to-many rather than two-way. Broadcasting is
perhaps the best term for these developments which include,
preeminently, radio and television; and in one form or another
these have become ubiquitous forms of communication in the
modern era .
Despite the ubiquity of radio and television, it is dif cult
to characterize the language of broadcasting in any distinctly
homogenous fashion. Instead, it is best understood as a medley
of distinct genres, including news interviews and reports, comic
. 1996 . Introduction. In Te Twin Earth Chronicles: Twenty Years
of Refection on Hilary Putnams Te Meaning of Meaning, ed.
Andrew Pessin and Sanford Goldberg , xvxxii. London: M. E. Sharpe.
Tye , Michael . 1995 . Ten Problems of Consciousness. Cambridge,
MA : Bradford Books, MIT Press.
MEDIA OF COMMUNICATION
Tis term refers to the means by which communication
takes place or the choice of substance for realizing a commu-
nicative act.
It has long been recognized that choosing between alterna-
tive means or media of communication can have consequences
for the linguistic form of the message. Tus, the choice of writ-
ing instead of speech as a medium of communication may entail
particular grammatical, syntactic , or lexical preferences over
others. Indeed some work suggests that certain kinds of linguistic
patterning may be distinctive to particular media (Biber 1988 ).
From Ferdinand de Saussure (1912) onward, much work
has been devoted to specifying the diferences between speech
and writing. (See also oral composition; oral culture;
writing systems; writing, origin and history of ).
One approach deals with speech and writing as alternative
expressions of the same underlying language system, realized
in difering ways depending on the medium adopted. Tus,
the written medium is associated with greater lexical density,
a wider range of grammatical structures, a greater degree of
embedding, and more varied forms of connectivity between sen-
tences. Conversely, the spoken medium is associated with lexi-
cal repetition, low lexical density, vague or indefnite expressions
(thingymajig), high incidence of coordinated clauses linked by
common conjunctions ( and , but ), and selection of the active
rather than the passive voice. Te character of these diferences
has led one author to characterize speech as a process and writ-
ing as a product (Halliday 1985 ).
A more radical view suggests that sentence grammars gener-
ally have been implicitly biased toward the study of writing, and
grammars would be quite diferent if they were formulated from
the outset to take account of speech phenomena. As David Brazil
observes, if any part of the outcome [of a grammar of speech]
looks like a sentence, this comes about as an interesting by-prod-
uct of the processes we are interested in, not as the planned out-
come to which these processes owe their defnition (1995, 39).
Certainly there is widespread agreement that the commu-
nicative potentialities of writing and speech are very diferent.
Speech typically takes place between interlocutors who are in
some way copresent to each other, and this enables them to
adjust their utterance in the light of the apparent reactions of
the other. Te process of composing and planning speech goes
hand in hand with the act of speaking, and speaking, in turn,
goes hand in hand with the process of interpretation that must
keep pace with it. Tere is no time lag between production and
reception. Instead, speech is temporally bound, transient, and
dynamic, rooted in an unfolding context, with paralinguistic
behavior providing an important supplementary layer to com-
munication (see paralanguage ). Conversely, writing as a
semipermanent product enables a gap across time and space to
open up between participants. Te process of composition may
Media of Communication
475
from reading a script to speaking ex tempore, from address to
the absent audience to address to a copresent interlocutor, from
script written to be read aloud as if unscripted to reading aloud
an e-mail from the audience .
In the movement from one phase to another in the devel-
opment of technologies of communication, there are shifts of
emphasis between one-to-one and one-to-many. Te emergence
of writing and print allows communication of the one to the
many. Te emergence of wireless telegraphy allows one-to-one
but over extreme distances. Broadcasting prioritizes one-to-
many and further collapses both temporal and spatial distances.
Te recently launched digital phase that has followed in the wake
of broadcasting has allowed the most radical innovations regard-
ing the confgurations in time and space of participants to com-
munication: Instantaneity over distance is possible, and extreme
forms of both one-to-one and one-to-many communication can
become blended in a single message .
Text messaging (SMS) and e-mail, for instance, can be one-
to-one or forwarded to a larger audience; are primarily asyn-
chronous, but single messages can develop into an extended
dialogue; often assume a fast response, but this may be delayed
if the recipient is of ine; and seem transient or ephemeral but
may be archived (sometimes in hard copy, in the case of e-mail)
for later use. Te language style of such communication may
well include extreme abbreviation, slang, contractions, phonetic
spelling , erratic punctuation, and short forms, and it seems
to operate in an unstable and fuctuating zone between speech
on the one hand and writing on the other. Tis might only be a
matter of linguistic curiosity except that variation between styles
of communication interact with questions of formality and the
quality of the social relationship. Many commentators have
pointed to growing informality in communication in the modern
era, using such terms to describe the shift as informalization
(Elias 1996 ), the democratisation (or conversationalisation) of
discourse (Fairclough 1992 ), intimacy at a distance and para-
social interaction (Horton and Wohl 1956 ), synthetic personal-
isation (Fairclough 2001 ), and broadcast sociability (Scannell
1996 ). Te emphasis in these accounts varies between attention
to forms of the message and attention to forms of the relation-
ship aforded by the message, but what generally seems to be
at stake is a changing sense of what counts as public space and
what counts as the appropriate linguistic and social demeanor
for it . While larger processes of social change may well underpin
these shifts, the changing media of communication have clearly
contributed to them .
Martin Montgomery
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aitchison , Jean , and D. Lewis . eds. 2003. New Media Language .
London : Routledge.
Biber , Douglas . 1988 . Variation across Speech and Writing .
Cambridge : Cambridge University Press.
Brazil , David. 1995 . A Grammar of Speech . Oxford : Oxford University
Press .
Crystal , David . 2001 . Language and the Internet . Cambridge : Cambridge
University Press .
Elias , Norbert . 1996 . Te Germans . New York : Columbia University
Press .
monologue, soap opera, various kinds of reality programing,
commercials, commentary on public events including sporting
occasions, argument, drama, talk shows, and phone-ins (several
of which have begun to attract systematic study; see Hutchby
2005 ; Tolson 2005 ). Although there may be generic antecedents
to these in the world of real-time, face-to-face communication,
certain properties seem to set broadcast genres apart from every-
day nonmediated communication. For one thing, the idealized
speaker and hearer of the canonical speech situation, recipro-
cally exchanging roles and utterances, no longer easily applies
except in grossly simplifed ways.
Instead, as Erving Gofman ( 1981 ) observes, broadcast com-
munication takes shape from complex production formats and
participation frameworks in which the discourse is sometimes
scripted, sometimes relatively spontaneous, sometimes spoken,
sometimes written, sometimes written to be spoken, sometimes
single authored, sometimes multiply authored, sometimes dia-
logue, and sometimes monologue. Indeed, Gofman suggests
replacing the term speaker with notions of author, animator,
and principal. Te author is the one who has selected the sen-
timents that are being expressed and the words in which they are
encoded (Gofman 1981 , 144). Te animator is the one who gives
voice to the words that have been selected, sometimes by some-
one else. Te principal is whoever is potentially held to account
for the sentiments expressed. In many situations, the three roles
coalesce, but in broadcast communication in news programs,
for instance the presenter who reads the news from the autocue
may merely be animating a script authored elsewhere, by the
editorial team, and the ultimately accountable source for the dis-
course the principal may be the organization itself. Tus, in
the case of a BBC news bulletin, it may be the director general
or members of the board of trustees who resign their positions
should an item be called into question, not necessarily the news
editor, and certainly not the news presenter .
Just as various alignments are possible in terms of the produc-
tion of broadcast communication, important distinctions apply
in its reception, where the potential participation framework
is equally complex. As Gofman again observes, an utterance
does not carve up the world beyond the speaker into precisely
two parts, recipients and non-recipients, but rather opens up an
array of structurally diferentiated possibilities, establishing the
participation framework in which the speaker will be guiding
his delivery (1981, 137). Broadcast communication is quite fre-
quently oriented to two kinds of recipient. In studio interviews,
for instance, in chat shows or news programs, there is the imme-
diate recipient of the talk the interviewee or the interviewer
but beyond them is the overhearing audience numbering in size
from thousands to millions. In this way, in posing a question to
an interviewee, the discourse of the interviewer is bidirectional. It
is oriented in the frst instance to the interviewee, but the design
of the question will also be shaped by the assumed concerns of
the broadcast audience beyond. Talk for an overhearing mass
audience in this way assumes characteristics distinctive to the
medium that are diferent from ordinary talk or conversation .
It should be noted also that in the broadcast media, the
foundational distinction in considering media of communica-
tion between speech and writing becomes confounded. Within
a continuous stretch of discourse, a language user may switch
Media of Communication
476
In linguistics, the plausibility of meme-based approaches is
supported by the increasing productivity of Darwinian thought
in language studies, which has inspired new eforts to explain
the evolution of language (see Hurford 2006 ) and to understand
the historical development of languages in Darwinian terms
(e.g., Croft 2001 ). Although the Darwinian algorithm depends on
the existence of replicating units, the potential of meme-based
approaches to language remains largely to be explored .
Memetic theories of language need to address at least three
fundamental questions. First, linguistic memes, or replicat-
ing constituents of linguistic competence , need to be plausi-
bly conceptualized as material patterns with determinable and
empirically detectable structures . Second, the mechanics by
which memes are replicated need to be determined. Tird, the
factors that determine the diferential replication of meme vari-
ants need to be identifed .
A study that adheres strictly to Dawkinss original pro-
posal is that of Nikolaus Ritt 2004 . Following connectionist
approaches to competence modeling, he sees language memes
as patterns acquired by neural networks during language
acquisition. Tus, a meme representing a phoneme contains
a) links to confgurations underlying articulatory gestures and
b) links to areas that are excited by specifc auditory impres-
sions, as well as c) links to representations of morphemes for
whose distinction the phone-meme is relevant. Terefore,
phone-memes have both determinable internal structures (i.e.,
the links between auditory and articulatory confgurations) and
determinable positions within the larger networks that imple-
ment linguistic competence . Memetic constituents coding for
phonotactic regularities, rhythmic confgurations, morphs, or
syntactic categories and constructions are construed in sim-
ilar terms.
Te replication of language memes involves communica-
tion, acquisition, and accommodation. Since a speakers com-
municative behavior is constrained by his or her competence,
utterances will automatically express the memetic constituents
by whose activation they are caused. Ten, the mind-brains of
recipients and those of children in particular will attempt to
assume organizations by which they can emulate the utterance
behavior they are exposed to. Tereby, copies of memes that are
expressed in utterances get created.
Among possible factors determining the diferential repli-
cation of meme variants, three types are distinguished. First,
meme replication must be constrained by physiological proper-
ties of their hosts. Tus, meme variants that are easy to express in
articulation and whose expression is easy to perceive will be uni-
versally ftter than more costly and less easily perceivable com-
petitors. Second, memes will be sensitive to such social factors
as power relations within and across groups. Te more powerful
and prestigious that individuals or groups are perceived to be,
the more often will their behavior be imitated. Tird, the repli-
cation of any meme will depend on other memes in the system.
Since utterances always express many memetic constituents
simultaneously, stable languages will contain mutually coad-
apted memes, which co-express with minimal distortion of one
anothers expressions .
While the predictions derived from physiological and social
constraints on meme selection seem to mirror those of speaker-
Fairclough , Norman .1992. Discourse and Social Change . Cambridge,
UK : Polity .
. 1995 . Media Discourse . London : Arnold.
. 2001 . Language and Power . 2d ed. London : Pearson Education .
Gofman, Erving. 1981. Forms of Talk . Oxford : Basil Blackwell .
Halliday , Michael Alexander Kirkwood . 1985 . Speech and Writing .
Oxford : Oxford University Press.
Horton , Donald , and R. Richard Wohl . 1956 . Mass communication
and para-social interaction: Observations on intimacy at a distance.
Psychiatry 19 : 215 29.
Hutchby , Ian . 2005 . Media Talk: Conversation Analysis and the Study of
Broadcasting . Buckingham : Open University Press .
Saussure , Ferdinand de . 1916 . Cours de linguistique gnrale . Ed. C. Bally
and A. Sechehaye , with the collaboration of A. Riedlinger. Lausanne
and Paris : Payot.
Scannell , Paddy. 1996 . Radio, Television and Modern Life . Oxford : Basil
Blackwell .
Tolson , Andrew . 2005 . Media Talk: Spoken Discourse on TV and Radio .
Edinburgh : Edinburgh University Press
MEMES AND LANGUAGE
Memes are information patterns that are culturally transmittable
and undergo Darwinian evolution: Variation among meme types
is created when patterns are altered, recombined, or transmit-
ted imperfectly, and selection takes place when more stable or
more easily transmittable meme variants come to oust competi-
tors that are less ft, that is, less stable or transmittable. As far as
their material implementation is concerned, memes are gener-
ally thought to exist in brains as constituents of human knowl-
edge. Whether human behavior and artifacts should be regarded
as external expressions of memes or as alternative ways in which
memetic information can be implemented is still disputed,
although the former view seems to be gaining ground .
Te term meme was coined by the evolutionary biologist
Richard Dawkins ( 1976 , 192) to denote cultural counterparts
of genes. Dawkins introduced the concept to support the argu-
ment that Darwinian evolution is not limited to the biological
domain but represents an algorithmic process that will afect
any patterns that are suf ciently stable and copied in suf cient
numbers with suf cient fdelity. While in the evolution of species
those patterns are genes, the historical development of human
cultures might refect the evolution of memes.
Te concept of memes is linked to the idea that human cul-
ture is a Darwinian system that can be understood best on the
level of the elements on which selection operates. A memetic
view of culture regards humans as physiologically complex, yet
relatively passive meme hosts. Teir instinctive inclination to
imitate one another turns them into meme vehicles with lim-
ited control over the memes they acquire, express in behavior,
or pass on to other humans. Of course, meme replication will
depend, to a considerable extent, on the physiological makeup,
the well-being, and the needs of their hosts, and memes that
infict obvious harm on their human carriers are unlikely to
thrive. However, the ultimate reason why any Darwinian rep-
licator exists is its capacity to get itself transmitted before it
disintegrates. Tus, a memetic approach to human behavior
challenges hermeneutic views (see philology and herme-
neutics ), which derive it from the irreducibly subjective per-
spectives of intentional agents.
Memes and Language
477
represents the occurrence of specifc events, for example, when
and where you last wrote a letter .
Despite the pervasiveness of memory in cognitive functions
such as language, historically memory systems have been inves-
tigated and theorized as separate and distinct from other cogni-
tive systems. Tis approach, however, can be dif cult to sustain
because of the centrality of memory to cognitive functions,
especially language. Some current models postulate that rep-
resentations and processes used for memory and language
are inseparable and part of the same system at both a behavioral
and neural level (MacKay et al. 2007 ; see hippocampus ). Here,
we review the relation between memory and language in behav-
ior and consider evidence relevant to whether they are distinct
systems.
Semantic Encoding in Memory and Language
A long-standing feature of models of language processing is that
perception of a word activates its lexical representation and that
activation automatically spreads to associated conceptual infor-
mation, including semantic properties of the word that consti-
tute its meaning (e.g., Rapp and Goldrick 2000 ; see spreading
activation ). A clear demonstration of this feature is seen in the
Stroop task in which participants are instructed to ignore a word
and simply name its ink color. Despite instructions to ignore the
word, color naming latency is faster when the base word is the
same as the ink color than when it is a diferent color, for exam-
ple, the word blue written in red. Tis diference in latency can
occur only if the meaning of the base word is accessed, despite
instructions to ignore the word. Tis automatic semantic encod-
ing of a word is clearly a process that is an essential part of a pri-
mary language function, namely, language comprehension, as it
is essential to understanding the meaning of a word , sentence
and discourse (see discourse analysis [linguistic]) .
Tis language process is also part of encoding in episodic
memory. For example, in a variant of the Stroop task, color nam-
ing latency was slower for taboo base words (e.g., whore ) than
for neutral base words (e.g., wrist ). In a subsequent surprise
recall test, memory was better for taboo than neutral base words
(MacKay et al. 2004 ; see emotion and language and emo-
tion words ). Tis diference in the efects of taboo and neutral
base words demonstrates that the automatic semantic activation
that occurred during perception of the base word was the basis
not only for lexical comprehension (which slowed color nam-
ing latency for taboo base words) but also for the representation
involved in subsequent episodic memory recall. Te strong infu-
ence of meaning on memory indicates an overlap of comprehen-
sion and memory representations .
Te degree of semantic activation during encoding also has
a strong efect on how well verbal material is remembered.
Participants remember more words in a surprise episodic
memory test after making judgments about word meaning
compared to judgments about phonology or physical form
(Craik and Tulving 1975 ). Te idea that semantic processing is
a deeper level of processing that improves memory has been
criticized for being a circular explanation. Nevertheless, partici-
pants had no advance knowledge that memory would be tested
and thus they did not engage in mnemonic strategies, and so
the fndings demonstrate that semantic processes involved in
based theories that derive the properties of languages from the
needs of their users, the co-adaptive pressures among memes
promise new explanations of long-term conspiracies in language
change, or the existence of typological classes. Tus, most Old
and Middle English sound changes that altered the metrical
weight or the syllabic structure of lexemes produced outputs that
were more trochee-like than their inputs. From a memetic per-
spective, they can be explained as morphotactic adaptations of
lexemes to rules coding for foot isochrony.
Strictly memetic approaches to language are still a minor-
ity program. While adherents regard memes as essential to any
truly Darwinian theory of language, even some of the linguists
who pursue explicitly Darwinian agendas (e.g., Croft 2001 ) prefer
to think of selection as being performed on utterance constitu-
ents and to attribute more active roles to speakers as agents of
change. Skeptics (e.g., Aunger 2001 ) also emphasize the need to
formalize memetics, the missing evidence of neural replicators,
and the paucity of empirical studies demonstrating the explana-
tory potential of the approach .
Nikolaus Ritt
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aunger , Robert . 2001 . Conclusion. In Darwinizing Culture: Te Status
of Memetics as a Science , ed. Robert Aunger, 20533. Oxford : Oxford
University Press .
Blackmore , Susan . 1999 . Te Meme Machine . Oxford : Oxford University
Press .
Croft , William . 2001 . Explaining Language Change . London : Longman .
Dawkins , Richard . 1976 . Te Selfsh Gene . Oxford : Oxford University
Press .
Dennett , Daniel . 1995 . Darwins Dangerous Idea: Evolution and the
Meanings of Life . New York : Simon and Schuster .
Hurford , James R . 2006 . Recent developments in the evolution of lan-
guage. Cognitive Systems 7 (November): 23 32.
Ritt , Nikolaus . 2004 . Selfsh Sounds and Linguistic Evolution .
Cambridge : Cambridge University Press .
MEMORY AND LANGUAGE
Te study of memory is concerned with the way that informa-
tion is represented and stored over time in the mind and how
it is retrieved and infuences behavior. Memory is essential for
all cognitive functions, including those that are intentional and
under conscious control and those that occur automatically
without conscious control. Te dominant theoretical approach
postulates several separate memory systems that vary in the
nature of their encoding and retrieval, their duration, and their
neural substrates, as well as how they are afected by diferent
variables, such as the age of the person or the level of process-
ing during encoding (e.g., Tulving and Schacter 1990 ). For
example, semantic memory stores long-term conceptual
knowledge including linguistic knowledge, such as words used
to express concepts ; procedural memory represents learned
skills and perceptual-motor routines that can be enacted with
little attentional control, for example, reading; working
memory represents the current content of consciousness and
enables manipulation of this information as when, for exam-
ple, a reader develops the meaning of a text; episodic memory
Memory and Language
478
target word, especially in sound, sometimes persistently come to
mind, but these alternate words are a consequence, not a cause,
of the gap produced by the TOT (Burke et al. 1991 ).
TOTs are caused by a retrieval failure at the phonological
level of the representation of the word while semantic informa-
tion is available for retrieval. Low-frequency words are more vul-
nerable to TOTs than high-frequency words, and recent use of a
word makes it less vulnerable to TOT. Clearly, TOTs represent a
memory retrieval failure and demonstrate the interdependence
of memory and language processes. TOTs can be explained in
terms of impaired spreading activation from lexical to phono-
logical representations because of weak connections between
these representations, caused by disuse . Consistent with this
explanation, a TOT can be resolved by pronouncing phonolog-
ical segments of the word, which increases phonological activa-
tion (James and Burke 2000 ). Memory and language processes
are indistinct here because identical representations and pro-
cesses (spreading activation) are crucial for both memory and
language .
Language processes are also closely linked to working mem-
ory. Teories of working memory include a storage component
and a controlled attention or central executive component that
maintains or computes information that is the focus of attention.
Working memory has a limited capacity that constrains the abil-
ity to perform complex mental computations, including seman-
tic and syntactic computations necessary for constructing
linguistic representations that are the basis for language compre-
hension and production (Just and Carpenter 1992 ; Caplan and
Waters 1999 ). For example, limited capacity causes reading time
to increase at points in a sentence where dif cult syntactic com-
putations are required. Moreover, a persons language ability is
correlated with the capacity of their working memory: Language
comprehension and production are better for people with large
rather than small working memory spans.
Te theory that working memory is a separate construct from
linguistic processes that constrains language functions has been
challenged recently by a connectionist approach to language.
Tis approach postulates that computational ef ciency in lan-
guage processing is determined by the state of the network rep-
resenting linguistic knowledge, not by the capacity of a separate
working memory system (MacDonald and Christiansen 2002 ).
Knowledge and experience infuence the state of the representa-
tional network. For example, repetition strengthens connections
among representations so that they pass activation more quickly,
increasing processing ef ciency. Consistent with this idea, lan-
guage that is more frequent at either a lexical or syntactic level
is easier to process. Within this approach, complex syntax slows
reading not because it requires more working memory capacity
but because such syntax is infrequent, which weakens connec-
tions among relevant representations. Similarly, because linguis-
tic experience increases the processing ef ciency of the language
network, individuals with greater language experience are pre-
dicted to have larger verbal working memory spans than individ-
uals with less language experience. Tat is, the observed relation
between language ability and working memory span is attributed
to a common cause: increased ef ciency of the language network
because of increased language experience. Tis approach elimi-
nates the architectural and computational distinction between
understanding language form the basis for representing specifc
occurrences of words in memory .
Semantic Activation and Memory Errors
Semantic activation during language comprehension is also the
basis for memory errors, especially constructive memory errors,
which occur when there is false memory for material that is con-
ceptually related to presented material but was not actually pre-
sented. For example, implications of sentences are commonly
remembered as having been presented when they were not. Te
target sentence, Te hungry python caught the mouse is likely to
be remembered as Te hungry python ate the mouse (Harris and
Monaco 1978 ). Te implication is encoded in memory as part
of the presented sentence because it is activated during com-
prehension. What is remembered is what is computed by com-
prehension processes, not what was actually presented. Tis
integration of language comprehension and memory makes it
extremely dif cult for people to remember language verbatim
and makes memory for what people said or wrote notoriously
unreliable.
False memories based on semantic activation processes have
also been demonstrated in the Deese/Roediger-McDermott
(DRM) experimental paradigm (e.g., Roediger and McDermott
1995 ). In the DRM paradigm, participants are asked to remem-
ber a list of words (e.g., snooze, wake, dream, blanket , etc.) that
are associated with an unpresented critical word (e.g., sleep ).
Participants falsely remember the unpresented critical word at
rates equivalent to the presented items; their confdence in their
memory accuracy is as high for critical words as for presented
words. Te high rate of false memory for a critical word in this
paradigm has generally been explained by semantic activation
of the list words during their presentation that spread to and
summated at the representation for the critical word. Te high
level of activation of the critical word at the test produces a feel-
ing of familiarity that leads to the false recognition. Consistent
with this explanation, increasing the number of related words on
the studied list increases the likelihood that the critical word is
falsely remembered (Robinson and Roediger 1997 ). Tere is also
evidence that semantic activation of the critical word may have
decayed before the test, but the critical word is reactivated at the
test because of its association with the list (Meade et al. 2007 ) .
Memory Processes and Language
Frequency and recency of occurrence have strong efects on
memory. Classic forgetting curves show that the more recent the
presentation of material, the better the memory for it. Frequency
or repetition improves both episodic and procedural memory.
Parallel efects of frequency and recency are seen in language.
Words that are repeated frequently or more recently in natural
language are easier to perceive and to produce. Te efect of
recency and frequency is demonstrated in a dramatic language
production failure known as the tip-of-the-tongue state (TOT) in
which a person is temporarily unable to produce a well-known
word. In the throes of a TOT, a person can produce semantic
information about the TOT target and sometimes partial infor-
mation about the phonology of the word, such as number of
syllables or frst phoneme , but the complete word remains
maddeningly out of reach. Alternate words related to the TOT
Memory and Language
479
logician C.S. Peirce formulated an account of verbal reasoning
based on diagrams that were models of assertions (Peirce
193158, vol. 4). In the 1970s, cognitive scientists converged
again on the idea of mental models. Tey argued that when we
understand discourse, we use sentence meaning and our
general knowledge in order to construct a mental model of the
situation under description (also known as a situation model ).
Such a model is as iconic as possible; that is, its structure corre-
sponds to the structure of the situation it represents. Hence, the
model represents each referent with a single mental token, the
properties of referents with properties of the tokens, and the rela-
tions among referents with relations among the tokens (Johnson-
Laird 1983 ). Te model captures what is common to the diferent
ways in which a possibility might occur, and so the theory is
analogous to possible worlds semantics and to discourse
representation theory (Kamp and Reyle 1993). However, these
approaches postulate that representations are logically correct,
whereas mental models have inbuilt shortcomings as a result of
the constraints of the human mind (see the following) .
As an example of a model, consider a simple spatial descrip-
tion (see Byrne and Johnson-Laird 1989):
Te of ce door is on the left of the elevator. Te exit door is on
the right of the elevator.
We can construct a mental model of the spatial layout, which is
analogous to this diagram:
of ce-door elevator exit-door
Te diagram is iconic in that its layout corresponds to a plan of
the three entities, but a mental model of the layout is likely to
represent the doors rather than to use verbal labels, which occur
in the diagram for simplicity.
Suppose that the discourse continues:
A man is standing in front of the of ce door. A woman is
standing in front of the exit door.
We incorporate this information in our model:
of ce-door elevator exit-door
man woman
It follows that the man is on the left of the elevator and the
woman is on the right of the elevator. No alternative model of
the discourse is a counterexample to this conclusion, and so it
is logically valid; that is, it must be true, given the truth of the
premises.
Mental models govern our memory for discourse. Suppose
that this discourse continues:
Te man standing in front of the of ce door was using a cell phone.
Later it states:
Te man using the cell phone was wearing a suit.
Both assertions can be used to update our model. In an unex-
pected memory test, as Alan Garnham (1987) showed, we are not
likely to recall which of the following sentences occurred in the
discourse:
working memory and the language system. It views working
memory limitations as emerging from the architecture of the
language network, rather than from a fxed capacity .
Elizabeth R. Graham and Deborah M. Burke
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Burke , Deborah M ., Donald G . MacKay , Joanna S . Worthley , and
Elizabeth Wade . 1991 . On the tip of the tongue: What causes word
fnding failures in young and older adults? Journal of Memory and
Language 30 : 542 79.
Caplan , David , and Gloria S. Waters . 1999 . Verbal working memory and
sentence comprehension . Behavioral and Brain Sciences 22 : 77126.
Craik , Fergus I. M ., and Endel Tulving . 1975 . Depth of processing and
the retention of words in episodic memory. Journal of Experimental
Psychology: General 104 : 268 94.
Harris , Richard J ., and Gregory E. Monaco . 1978 . Psychology of prag-
matic implication: Information processing between the lines. Journal
of Experimental Psychology: General 107 : 1 22.
James , Lori E ., and Deborah M. Burke . 2000 . Phonological priming
efects on word retrieval and tip-of-the-tongue experiences in young
and older adults . Journal of Experimental Psychology: Learning,
Memory and Cognition 26: 1378 91.
Just , Marcel A ., and Patricia A . Carpenter . 1992 . A capacity theory of com-
prehension: Individual diferences in working memory. Psychological
Review 99 : 122 49.
MacDonald , Maryellen C ., and Morten H. Christiansen . 2002 . Reassessing
working memory: Comment on Just and Carpenter (1992) and Waters
and Caplan (1996). Psychological Review 109 : 35 54.
MacKay , Donald G., Lori E. James , Jennifer K. Taylor , and Diane E.
Marian . 2007 . Amnesic H. M. exhibits parallel defcits and sparing
in language and memory: Systems versus binding theory accounts.
Language and Cognitive Processes 22: 377 452.
MacKay , Donald G ., Meredith Shaf to , Jennifer K. Taylor , Diane E. Marian ,
Lise Abrams , and Jennifer R. Dyer . 2004 . Relations between emotion,
memory, and attention: Evidence from taboo Stroop, lexical decision,
and immediate memory tasks. Memory and Cognition 32 : 474 88.
Meade , Michelle L ., Jason M . Watson , David A. Balota , and Henry L.
Roediger , III. 2007. Te roles of spreading activation and retrieval
mode in producing false recognition in the DRM paradigm. Journal of
Memory and Language 56 : 305 20.
Rapp , Brenda , and Matthew Goldri ck. 2000. Discreteness and interac-
tivity in spoken word production. Psychological Review 107 : 460 99.
Robinson , Kerry J ., and Henry L. Roediger , III. 1997 . Associative pro-
cesses in false recall and false recognition. Psychological Science
8 : 231 7.
Roediger , Henry L., III , and Kathleen B. McDermott . 1995 . Creating
false memories: Remembering words not presented in lists. Journal
of Experimental Psychology: Learning, Memory, and Cognition
21 : 803 14.
Tulving , Endel , and Daniel L. Schacter . 1990 . Priming and human mem-
ory systems. Science 247 : 301 6.
MENTAL MODELS AND LANGUAGE
How do we represent discourse, and how do we reason from
its contents? One answer to both questions is that we rely on
mental models of the situations that discourse describes. Te
Scottish psychologist Kenneth Craik (1943, 61) wrote that if
we construct a small-scale model of the world, we can use it
to make sensible decisions about our actions. Several thinkers
anticipated him, and the nineteenth-century American
Mental Models and Language
480
Te mental models of the frst assertion are:
king ace
not(king) ace
where not represents negation. Te second assertion elimi-
nates the second of these possibilities, and so it seems that there
is an ace in the hand. However, the connective or else means that
at the very least, one of the propositions that it connects may be
false. Given, say, the falsity of if theres a king in the hand then
theres an ace in the hand , we realize that its possible that theres
a king in the hand without an ace. So, even though the second
assertion tells us that there is a king in the hand, no guarantee
exists that theres an ace, too. Te inference is fallacious. Tis
analysis relies only on two well-attested facts about our under-
standing: 1) or else allows that one of the clauses it connects is
false, and 2) the falsity of a conditional allows that its if-clause
can be true and its then-clause false. A computer program imple-
menting the principle of truth led to the discovery of a variety of
illusions, and subsequent studies have corroborated their occur-
rence (Johnson-Laird 2006) .
When you read the earlier description of the man and woman
standing in front of the doors, you might have formed a visual
image of the layout. Spatial relations are usually easy to visualize.
You might, therefore, assume that mental models are nothing
more than visual images. Tis assumption is wrong. Some rela-
tions, such as Pam is better than Viv , are impossible to visual-
ize. You can imagine, say, Pam as further up a ladder than Viv is,
but nothing in your image or in any possible image can make
explicit the meaning of better than . Not all relations are rooted in
a sensory modality or have a spatial interpretation. Some rela-
tions, such as the cat is cleaner than the dog , are easy to visualize
but do not invoke a spatial representation. Reasoning with these
visual relations, which elicit images rather than spatial models,
takes longer than reasoning with other sorts of relation, and it
activates a region of the brain underlying vision.
Te hypothesis that we represent discourse in mental models is
uncontroversial in psycholinguistics , though not all accounts
stress the iconicity of models (cf. Kintsch 1988; and Gernsbacher,
1990). We make a dynamic representation of entities, their prop-
erties, and the relations among them. Te heart of the problem
in building a mental model is to recover the appropriate referent
for each expression. Speakers refer back to entities that they have
already introduced in the discourse, and they can use diferent
noun phrases, demonstratives, or pronouns to do so . Te correct
interpretation of such anaphora depends on many factors. Given
a sentence like Te man confessed to the priest because he wanted
absolution, we understand that he refers to the man rather than
the priest. We make this attribution because we know the purpose
of confession, because we have a preference for locating the ante-
cedents of pronouns in the subjects of clauses, and because we also
have a preference for a parallel grammatical role of antecedent and
anaphor (Stevenson, Nelson, and Stenning 1995).
In computational linguistics, centering theory shows how the
focus on a local segment of discourse determines the anteced-
ents of anaphora, especially pronouns (e.g., Grosz, Joshi, and
Weinstein 1995; Webber et al. 2003) Another factor is the seman-
tic diference between antecedent and anaphor (Almor 1999).
Te information load on interpretation increases when, unlike
Te man standing in front of the of ce door was wearing a suit.
Te man using a cell phone was wearing a suit.
We forget the sentences and recall only the situation represented
in our model. Hence, given an assertion that follows at once from
our model, we are also prone to suppose that it too occurred in
the discourse (Bransford, Barclay, and Franks 1972).
Models of discourse can be abstract. Tey can combine both
iconic elements and symbolic elements, such as negation. We
might translate negation into a set of alternative af rmative pos-
sibilities (Schroyens, Schaeken, and dYdewalle 2001). We repre-
sent the proposition that the man isnt in front of the exit door as
a set of af rmative possibilities: He is in front of the elevator, or
he is in front of the of ce door, or But this representation calls
for a procedure that interprets a set of models as alternatives. As
Peirce realized, this machinery is not iconic but symbolic. And,
often, there are too many af rmative possibilities to represent
negation in this way .
If we have a dynamic model of what happens in a story, then
changes in location should afect our ease of accessing referents.
Experiments have shown that if, say, the protagonist in a story
walks through a door into another room carrying an object, then
it is easier for us to access this object and the entities in the new
room than those in the room the protagonist has left. It takes us
longer to respond to questions or to a probe word; and these
efects occur for stories (e.g., Glenberg, Mayer, and Lindem
1987; Rinck and Bower 1995), movies (e.g., Magliano, Miller,
and Zwaan 2001), and virtual reality on a computer screen
(Radvansky and Copeland 2006). We therefore maintain a model
of discourse that has perceptual and spatial features that parallel
those in models that we construct from witnessing events, and
the model may rely on many of the same brain areas underlying
perception .
Te principle of truth is an interpretative assumption gov-
erning mental models (Johnson-Laird 2006). It postulates that
they represent only what is true according to the discourse. As a
consequence, an assertion, such as
Te man is wearing a suit or else the woman is wearing a suit,
but not both.
is represented in separate mental models of the two possibilities,
depicted here on separate lines:
man wears suit
woman wears suit
Again, these sentences stand in for mental models. What the
models do not represent, at least explicitly, is the falsity of the
woman is wearing a suit in the frst possibility and the falsity of
the man is wearing a suit in the second possibility. Te principle
of truth reduces the load on our memory, and it seems innoc-
uous. Yet, it can lead us into the illusion that we understand a
description that, in fact, is beyond us.
A striking illusion of this sort occurs with the description:
If theres a king in the hand then theres an ace in the hand
or else theres an ace in the hand if there isnt a king in the hand.
Tere is a king in the hand.
Mental Models and Language
481
Rinck , Mike , and G. Bower . 1995 . Anaphor resolution and the focus of
attention in situation models. Memory & Language 34 : 1 10 31.
Schroyens , Walter, W . Schaeken , and G. dYdewalle . 2001 . Te process-
ing of negations in conditional reasoning: A meta-analytic case study
in mental model and/or mental logic theory. Tinking & Reasoning
7 : 121 72.
Stevenson , Rosemary J., A. W. R. Nelson , and K. Stenning . 1995 . Te role
of parallelism in strategies of pronoun comprehension. Language and
Speech 38 : 393 418.
Webber , Bonnie, M. Stone, A . Joshi , and A Knott . 2003 . Anaphora and
discourse structure. Computational Linguistics 29 : 545 87.
MENTAL SPACE
What Is a Mental Space?
Mental spaces are partial assemblies constructed as we think
and talk, for purposes of local understanding and action. Tey
contain elements and are structured by frames and cognitive
models. Mental spaces are connected to long-term schematic
knowledge, such as the frame for walking along a path, and to
long-term specifc knowledge, such as a memory of the time you
climbed Mount Rainier in 2001. Te mental space that includes
you, Mount Rainier, and your climbing the mountain can be acti-
vated in many diferent ways and for many diferent purposes.
You climbed Mount Rainier in 2001 sets up the mental space in
order to report a past event. If you had climbed Mount Rainier
sets up the same mental space in order to examine a counter-
factual situation and its consequences. Max believes that you
climbed Mount Rainier sets it up again, but now for the purpose
of stating what Max believes.
Mental spaces are constructed and modifed as thought and
discourse unfold and are connected to each other by mappings ,
such as identity and analogy . It has been hypothesized that
at the neural level, mental spaces are sets of activated neuronal
assemblies and that the connections between elements cor-
respond to coactivation bindings. On this view, mental spaces
operate in working memory but are built up partly by activat-
ing structures available from long-term memory. Connections
link elements across spaces without implying that they have the
same features or properties. When I was six, I weighed ffty pounds
prompts us to build an identity connector between the adult and
the six-year-old despite the manifest and pervasive diferences.
Mental spaces are built up dynamically in working memory but
can become entrenched in long-term memory .
An expression that names or describes an element in one
mental space can be used to access a counterpart of that element
in another mental space ( access principle ).
Exploring Mental Spaces
In the 1970s, it became clear that grammatical and semantic
structure provide evidence for general features of human con-
ceptual systems and operations.
Logical phenomena, such as quantifer scope, anaphora,
opacity, and presupposition had been largely the province of
analytic philosophy. Bypassing the mind/brain, semantics was
framed in terms of an external theory of truth and reference.
cognitive linguistics embarked on a diferent course, plac-
ing mental constructs at the forefront of the study of language.
Te initial motivation for mental space theory (Fauconnier
normal cases, the anaphor is more specifc than its antecedent,
for example, He had a beer, and the Guinness tasted good.
Within the framework of mental models, the most comprehen-
sive account of anaphora is that of Garnham and his colleagues
(cf. Cowles and Garnham 2005). Tis theory takes into account
all the preceding factors, but also postulates that a crucial factor
is the number of potential antecedents for an anaphor. In looking
backward, an anaphor should have enough content to pinpoint its
antecedent among the candidates. But the choice of a particular
anaphor also signals the future direction of the discourse subse-
quent content may provide the information needed to pinpoint
the antecedent. And the content in the anaphor may also signal
a shift in theme. So, the theory is Janus-faced, looking both back-
ward for antecedents and forward for thematic shifts. No current
theory, however, has led to a computer program that constructs
models for anything more than a fragment of the language .
P. N. Johnson-Laird
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Almor , Amit. 1999 . Noun-phrase anaphora and focus: Te informational
load hypothesis. Psychological Review 106 : 748 65.
Bransford , John D., J. R. Barclay, and J. J. Franks . 1972 . Sentence
memory: A constructive versus an interpretive approach. Cognitive
Psychology 3 : 193 209.
Byrne , Ruth M. J ., and P. N. Johnson-Laird . 1989 . Spatial reasoning.
Journal of Memory and Language 28 : 564 75.
Cowles , Wind , and A. Garnham . 2005 . Antecedent focus and conceptual
distance efects in category noun-phrase anaphora. Language and
Cognitive Processes 20 : 725 50.
Craik , Kenneth . 1943 . Te Nature of Explanation . Cambridge : Cambridge
University Press .
Garnham , Alan . 1987 . Mental Models as Representations of Discourse and
Text . Chichester : Ellis Horwood .
. 2001 . Mental Models and the Interpretation of Anaphora . Hove,
East Sussex : Psychology Press . A major statement of the theory of men-
tal models for discourse.
Gernsbacher , Morton A . 1990 . Language Comprehension as Structure
Building . Hillsdale, NJ : Erlbaum .
Glenberg , Arthur M., M . Meyer , and K. Lindem . 1987 . Mental models
contribute to foregrounding during text comprehension. Memory &
Language 26 : 69 83.
Grosz , Barbara, A. Joshi , and S. Weinstein . 1995 . Centering: A frame-
work for modelling the local coherence of discourse. Computational
Linguistics 21 : 203 26.
Johnson-Laird , Philip N . 1983 . Mental Models . Cambridge : Harvard
University Press .
. 2006 . How We Reason . Oxford : Oxford University Press .
Kamp , Hans , and U. Reyle . 1993 . From Discourse to Logic . Dordrecht, the
Netherlands : Kluwer .
Kintsch , Walter . 1988 . Te role of knowledge in discourse compre-
hension: A construction-integration model. Psychological Review
95 : 163 82.
Magliano , Joseph P., J. Miller , and R. A. Zwaan . 2001 . Indexing space and
time in flm understanding. Applied Cognitive Psychology 15 : 533 45.
Peirce , Charles S . 1931 58. Collected Papers of Charles Sanders Peirce .
8 vols. Ed. C. Hartshorne , P. Weiss , and A. Burks . Cambridge : Harvard
University Press .
Radvansky , Gabriel A ., and D. E. Copeland . 2006 . Walking through
doorways causes forgetting: Situation models and experienced space.
Memory & Cognition 34 : 1150 6.
Mental Models and Language Mental Space
482
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Coulson , Seana . 2001 . Semantic leaps . Cambridge, UK : Cambridge
University Press .
Cutrer , M . 1994 . Time and tense in narratives and everyday language.
Ph.D. diss., University of California, San Diego .
Dancygier , Barbara . 1998 . Conditionals and Prediction . Cambridge :
Cambridge University Pres s.
Dancygier , Barbara, and Eve Sweetser . 2005 . Mental Spaces in
Grammar: Conditional Constructions . Cambridge : Cambridge
University Press .
Dinsmore , J . 1991 . Partitioned Representations . Dordrecht, the
Netherlands : Kluwer .
Epstein , Richard . 2001 Te defnite article, accessibility, and the con-
struction of discourse referents. Cognitive Linguistics 12 : 33378.
Fauconnier , Gilles . [1985] 1994. Mental Spaces . Cambridge : Cambridge
University Press .
. 1997 . Mappings in Tought and Language . Cambridge : Cambridge
University Press .
Fauconnier , Gilles, and Eve Sweetser . 1996 . Spaces, Worlds, and
Grammar . Chicago : University of Chicago Press.
Fauconnier , Gilles, and Mark Turner . 2002 . Te Way we Tink . New
York : Basic Books .
Huumo , Tuomas . 1996 . A scoping hierarchy of locatives . Cognitive
Linguistics 7 : 26599.
Liddell , Scott K . 2003 . Grammar, Gesture, and Meaning in American Sign
Language . Cambridge : Cambridge University Press .
Van Hoek , Karen . 1997 . Anaphora and Conceptual Structure . Chicago :
University of Chicago Press .
MERGE
Merge is the primitive combinatorial operation in the most recent
version of transformational grammar known as minimal-
ism . In its most austere variety, merge is a generalized transforma-
tion that simply turns its input elements into a set with the input
elements as members ( set-merge ). Unlike the earlier government
and binding model of the principles and parameters the-
ory , Noam Chomskys (1995) minimalist model does not assume
a deep structure (see underlying structure ) representation
as a starting point of the derivation; instead, syntactic computation
starts out from individual words. Merge combines words as well as
syntactic objects it has already formed in a recursive manner (see
recursion , iteration , and metarepresentation ), gener-
ating an infnite array of discrete expressions with a hierarchical
constituent structure . In principle, merge can freely apply
to elements available to it, but its application is constrained by
principles of computational ef ciency and by output conditions
imposed by external systems of sound/gesture and meaning .
movement is construed as merge of a syntactic object with a
syntactic object contained in it , whence the term internal merge
(vs. external merge ; see Chomsky 2004). Consider, for instance,
the derivation of the passive sentence (1a), where the underly-
ing object is moved to the subject position. Here the expression
(1b), constructed by recursive applications of merge, undergoes
merge with its subset { a , house }, yielding (1c). (2) is a tree dia-
gram representation of (1c).
(1) a. A house will be built
b. { will , { be , { built , { a , house }}}}
c. {{ a , house }, { will , { be , { built , { a , house }}}}}
[1985] 1994, 1997) was that it provided simple, elegant, and gen-
eral solutions to problems such as referential opacity or pre-
supposition projection that had baf ed logicians and formal
linguists. Opacity results from the application of the access prin-
ciple across mental spaces as discourse unfolds. What emerged
was a unifed cognitively based approach to anaphora, presup-
position, conditionals, and counterfactuals. Additionally, the
gestural modality of signed languages revealed other ways in
which mental spaces could be set up and operated on cognitively
and physically .
Shortly thereafter, J. Dinsmore (1991) developed a powerful
approach to tense and aspect phenomena, based on mental
space connections. Te approach was pursued and extended
in fundamental ways by M. Cutrer (1994), who made it possible
to understand the role of grammatical markers as prompts to
deploy vast networks of connected mental spaces. Further gen-
eralizations were achieved in areas exemplifed by the diverse
contributions to Spaces, Worlds, and Grammar (Fauconnier and
Sweetser 1996). Sophisticated research continues to be done in
all of the areas where mental space theory was frst applied, in
particular on conditionals (see Dancygier 1998; Dancygier and
Sweetser 1996, 2005), scoping phenomena on locative and tem-
poral domains (see Huumo 1996), grammar of sign languages
(see Liddell 2003), discourse (see Epstein 2001), and frame shift-
ing (see Coulson 2001). But at the same time, there has been an
explosion of research triggered by the discovery of wide-ranging
phenomena whereby mental spaces are assembled, connected,
and constructed within networks of conceptual integration (see
conceptual blending ). Tis area of research links linguis-
tic and nonlinguistic phenomena in systematic ways that begin
to explain how and why there can be imaginative emergent
structure in human thought in its everyday manifestations,
as well as in its most original and singular sparks of creativity .
Mental Spaces in Discourse: A Simple Example
Suppose the current president of our country is Nick, and that
someone says:
Tirty years ago, the president was a baby.
Te base mental space, B, corresponds to the time at which the
statement is made and contains an element a which flls the
role president in a political frame and has the name Nick . Te
space-builder thirty years ago sets up a new space M relative to
the base (30 years before now); a in B has a counterpart a in
the new space M; the president identifes a in B, and can therefore
access its counterpart a in M. Te property baby is assigned
to a in M. Te sentence is interpreted as saying that Nick was a
baby 30 years ago.
Te expression the president , however, can equally well be
construed as directly identifying an element b in M: It flls the
role president in the political frame for M. Te property baby
is now assigned to b . Te sentence is now interpreted as saying
that a baby was president 30 years ago.
It is an empirical fact that the example sentence does
indeed have the two interpretations, and this fact, like many
others, follows from the accessing principles of mental space
confgurations .
Gilles Fauconnier
Mental Space Merge
483
. 2004 . Beyond explanatory adequacy. In Structures and
Beyond: Te Cartography of Syntactic Structures , ed. Belletti, 10431.
Oxford : Oxford University Press .
METALANGUAGE
Languages may be used to talk about languages. In the frst sen-
tence of this entry, English is used to mention (or talk about) lan-
guages. (See use and mention .) In general, a language under
discussion, on a given occasion, is called the object language ,
and the language being used in the discussion (to talk about the
object language) is the metalanguage .
Alfred Tarski proved that no classical language (language with
classical logic) can express its own truth predicate. Accordingly,
the semantics of a classical language (with standard syntax)
can only be done in a richer metalanguage .
J. C. Beall
SUGGESTION FOR FURTHER READING
Tarski , Alfred . 1990 . Logic, Semantics, Metamathematics: Papers from
19231938 . 2d ed. Ed. J. Corcoran . Indianapolis, IN : Hackett .
METAPHOR
Tis term derives from the Greek metapherein , indicating a
transfer of meaning from one linguistic expression to a second,
semantically diferent expression. According to standard dic-
tionary defnitions, a metaphor is a fgure of speech in which a
word or phrase that ordinarily designates one thing is used to
designate another to which it is not literally applicable . Tus,
in metaphor, there is an implicit comparison between unlike
things, as in the Shakespearean metaphor Juliet is the sun.
Employing the terminology introduced by I. A. Richards (1936),
the target concept (Juliet) is labeled the topic or tenor , the met-
aphoric source (sun) is the vehicle and the emergent meaning,
the ground (see also source and target ). Despite this seem-
ingly simple defnition, the task of actually identifying metaphor
has proven to be a dif cult enterprise because metaphors can be
implied (there is a vehicle but no specifed topic), dead (a usage
so conventionalized that the metaphoric transference is no lon-
ger actively recognized), mixed (the confation of two distinct
metaphors), submerged (in which the vehicle is implied but not
stated), or extended (suggested throughout a text). Tis dif culty
has been especially apparent in recent attempts with computer
applications, such those based on the identifcation of metaphor
in text corpora or in computer systems capable of generating
representations of metaphorical meaning.
Metaphor as a Linguistic Phenomenon
THE INFLUENCE OF ARISTOTLE. Te classic approach, as presented
in the dictionary defnition, treats metaphor as a rhetorical
trope in which language is used in a way other than what may be
considered normal or literal. One can see in this approach the
infuence of Aristotle. As Umberto Eco put it, [O]f the thousands
and thousands of pages written about the metaphor, few added
anything of substance to the frst two or three fundamental con-
cepts stated by Aristotle (1984, 88). Whether or not one agrees
(2)
Te syntactic object { a , house } has two copies or occurrences , but
it is realized phonetically only as a member of the (largest) set in
(1c). Note that the two occurrences resulting from movement are
not distinct syntactic objects. Rather, the same syntactic object
is a member of two sets, where one set is properly contained in
the other .
While there is currently no agreement regarding restrictions
on merge, or how to deduce them, the mainstream view holds
that merge is binary, always taking exactly two input elements
(entailing strict binary branching in syntactic trees), and it can-
not alter set-membership relations that it has established before.
It is unresolved whether the output of merge should be enriched
to encode any asymmetry between its operands. On Chomskys
(1995) original defnition, merge forms a set with the following
two members: the set of the input elements and the word func-
tioning as the head of the constituent (also known as label ),
thereby representing the asymmetry in the choice of the input
element that projects (see x-bar theory ). Following this for-
mulation, (1c) can be rewritten as (3). Te structure in (3) is rep-
resented by the labeled tree diagram in (4).
(3) {will, {{a, { a , house }}, {will, { will , {be, { be , {built, { built , {a, { a ,
house }}}}}}}}}}
(4)
Another asymmetry is that between an adjunct (e.g., an adver-
bial) and the host it is adjoined to (e.g., a verb phrase). Chomsky
(2004) suggests that when an adjunct and its host undergo
merge, the result is an ordered pair Adjunct, Host (also known
as pair-merge ) .
Balzs Surnyi
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chametzky , Robert A . 2000 . Phrase Structure: From GB to Minimalism .
Oxford : Blackwell .
Chomsky , Noam . 1995 . Te Minimalist Program . Cambridge, MA : MIT
Press .
will
be
built
house a
a house
will
a
a
a
a house
house
will
will be
be built
built
Merge Metaphor
484
which the novel meaning of a metaphor is not based on identi-
fying a shared set of (possibly marginal) meanings of the words
being compared. Meaning, he argues, is generated by the inter-
action between a principal subject (the more literal usage of the
word, similar to the topic) and the complex of associations con-
nected with a subsidiary subject (analogous to the vehicle). Te
process is interactive inasmuch as reciprocal action between the
principal and subsidiary subject selects, emphasizes, suppresses
and organizes features of the principal subject by implying state-
ments about it that normally apply to the subsidiary subject
(Black 1962, 46). Te outcome is the creation of novel meaning
formed by a parallel implicational complex in which the topic
can be viewed in a radically diferent light and in which novel
emergent meanings can be created between words. Despite the
popularity of this general approach, it, too, has been subject to
various criticisms, notably regarding the ambiguity in defning
the theoretical terms employed and in determining which of the
terms is the principal and which is the subsidiary subject .
Subsequent psycholinguistic theories have attempted
to describe cognitive mechanisms that are consistent with
the interactive approach. Salience imbalance theory is a vari-
ant of traditional comparison models aimed at describing why
some statements are seen as literal and others as metaphoric by
assuming that the features shared by topic and vehicle difer in
relative level of salience: Literal statements are those in which
the shared features are salient to both terms, whereas with meta-
phor, they are salient to the vehicle but not the topic . Domain
interaction theory is an extension of a computational model of
analogy and assumes that metaphor involves the fnding of
similarity both within and between the conceptual domains
evoked by words. Tus, a metaphor such as George Bush is a
hawk would be comprehended by fnding a spot in semantic
space that would be consistent with the analogy George Bush
is to world leaders as hawks are to birds. In this model, ease
of comprehension is a function of the ease of fnding a shared
similarity (i.e., in determining ways in which Bush is similar to
a hawk), whereas a sense of metaphor aptness increases as
the distance between conceptual domains, such as leaders and
birds, becomes greater . Finally, structure-mapping theory, also
emerging from computational work in analogy, is based on iden-
tifying a system of shared relations between the target and source
domains and not by merely identifying a feature shared by topic
and vehicle. Although there are psycholinguistic studies that
support each theory, each is based ultimately on fnding similar-
ity between the words presented as topic and vehicle and, conse-
quently, heir to all of the criticisms of such models (reviewed, for
instance, in Glucksberg 2001) .
Sam Glucksberg proposes a novel solution that rejects simi-
larity as the basis for metaphor by arguing that in metaphor, one
does not look for a similarity between topic and vehicle (i.e., by
treating the comparison as an implicit simile). He avers, rather,
that metaphor should be understood as a class inclusion state-
ment analogous in how we treat such statements as my dog is a
collie. He argues, and has presented convincing evidence, that
with metaphor, the vehicle has dual reference (both as the lit-
eral object and as indicative of higher-order categories) and that,
in comprehension, one classifes the topic to the category sug-
gested by the vehicle. Tat is, in a metaphor such as my lawyer is
with Eco, it is clear that interpretations of Aristotelean thoughts
have infuenced much of the thinking on the topic for the past
several millennia, most notably a presumed distinction between
literal and nonliteral language and in framing the basic theo-
retical issues that have guided much of the subsequent schol-
arship: understanding the cognitive process that permits the
stretching of meaning from the literal to the metaphoric and in
determining the pragmatic reasons why metaphor would be
employed when literal counterparts could have been used.
In his Poetics and in Rhetoric, Aristotle situated metaphor
in the realm of language, a position that has been the basis for
subsequent theories but has been contested since 1980 by theo-
rists working within cognitive linguistics (described later).
Aristotles basic premise is that with metaphor, one word (or
expression) is substituted for another. He described several cat-
egories of substitution, though the forms most studied are what
we would call today nominal and predicative metaphor, the
former in which one noun is substituted for another and the lat-
ter in which the substitution is of verbs. Aristotle provided some
explanation of the process involved in metaphor comprehen-
sion, namely, an innate tendency to see likeness in objects and
events that are, on the face of it, dissimilar (or in which the sim-
ilarity is not transparent). Moreover, he provided some reasons
why metaphor might be employed, primarily, to serve a stylistic
and aesthetic function wherein the listener, forced to decode the
message, experiences a pleasurable reaction; and secondarily,
to serve the creative cognitive function of providing a name to
things that do not have proper names of their own. Tere has
been considerably elaboration of the seminal ideas of Aristotle in
the twentieth century, despite what some would consider a fatal
faw in the logic of substitution as the basis for metaphor: If, for
example, with the nominal Shakespearean metaphor Juliet is
the sun the vehicle, sun is a substitution for another word that
falls within the same genus= as the topic, Juliet, what could
that word be?
TWENTIETH-CENTURY ELABORATIONS. Two interpretations con-
sistent with Aristotle have been most infuential. According to
the substitution position, the transfer from vehicle to topic is an
ornamental means of presenting some intended literal meaning,
so that when one states George is a wolf, it is merely an aesthetic
way of saying that George is ferce. Te comparison approach is
less ornamental and closer to the second function described by
Aristotle in that, here, the listener must construct a way in which
properties of the vehicle are applicable to the topic. Tere have
been several variants of comparison theories proposed in the
literature over the past 50 years, but all include the notion that
the comprehension process involves the identifcation of a rele-
vant set of preexisting features shared by topic and vehicle. Tese
theories all have shortcomings, including the failure to encom-
pass the creation (and not mere identifcation) of similarity, the
problem in identifying the mechanisms that would select the
features assumed to be important for interpretation, and the fail-
ure in such theories to explain the asymmetry in meaning that
occurs when the topic and vehicle are reversed, as occurs when
one contrasts my lawyer is a shark with my shark is a lawyer.
Max Black attempted to address at least some of these short-
comings by postulating an interactive theoretical perspective in
Metaphor
485
CONCEPTUAL METAPHORS AND THE CONTEMPORARY THEORY OF
METAPHOR. Te research described in the previous sections has
undercut the diference between literal and nonliteral language
(see also Gibbs 1994), a challenge extended most notably by
cognitive linguists, especially by George Lakof , starting with the
publication in 1980 of Metaphors We Live By , co-authored with
Mark Johnson . Te main thrust of this theory is that metaphors
are matters of thought and not merely of language, thus distin-
guishing the basic mapping of a source conceptual domain to a
target conceptual domain ( conceptual metaphor ) from the
linguistic expression of this mapping. Te true source of meta-
phor is at the conceptual level. According to this theory, con-
ceptual metaphors motivate and underlie understanding of the
world, such that most of what we call literal is, by this theory,
based on underlying metaphorical mappings. Tus, conceptual
metaphors are the basis for understanding literal and nonlit-
eral, novel and conventional, poetic and mundane language
alike. Evidence for a conceptual metaphor, such as the mapping
between the conceptual domains of life and journeys (LIFE IS A
JOURNEY), is refected in a set of seemingly unrelated linguistic
expressions, such as His life is at an important crossroad and
She knows where she is going. Mappings elucidate the system-
atic set of correspondences that exist between constituent ele-
ments of the source and the target domain. For example, with
the LIFE IS A JOURNEY mapping, the person is analogous to a
traveler, purposes are destinations, means are routes, dif culties
are obstacles, achievements are landmarks, choices are cross-
roads, and so on, allowing for novel extensions of elements from
the source domain to elements of target concepts.
Te theory has had widespread acceptance, and the task of
identifying the presence and force of underlying (and hence
unconscious) cognitive mappings has entered the debates of
linguistics, cognitive science, philosophy, literary theory, and
criticism, among other disciplines . Nonetheless, the claims in
the literature for an ever-increasing number of conceptual meta-
phors indicate looseness in the theory that may make it incapable
of being disprovable and, thus, an inadequate scientifc expla-
nation. Moreover, one testable prediction made by the theory,
namely, that conceptual metaphors are activated on line during
comprehension, has not been supported consistently, with the
strongest support coming from the examination of orientational
and temporal metaphors (e.g., Boroditsky 2000) .
CONCEPTUAL BLENDING. A more recent framework, proposed by
Gilles Fauconnier and by Mark Turner , seeks to explain much
of the same linguistic data discussed in the conceptual meta-
phor literature and shares with that approach the assumption
that metaphor is a conceptual, not a linguistic, phenomenon. In
contrast with conceptual metaphor theory, however, concep-
tual blending theory is not limited to entrenched conceptual
relations or to the unidirectional mapping from source to target
or the mapping between only two mental domains. Rather, the
basic units are mental spaces representing particular scenarios
recruited from the knowledge of specifc domains constructed
while thinking or talking about situations. As such, the theory
emphasizes blending as an on-line process, which both instanti-
ates entrenched metaphors and can yield short-lived and novel
conceptualizations. Tis theory, too, has entered the literatures
a shark, the vehicle shark stands for or exemplifes a category
to which lawyers could be assigned (such as aggressive, preda-
tory, tenacious entities). In more recent expansion of the theory,
he and his colleagues have indicated how the topic plays a role
in identifying the appropriate category for which categorization
is appropriate .
THE PROCESS OF METAPHOR COMPREHENSION. Much of the the-
ory and research described here is based on of ine methodology.
Beginning in the late 1970s, researchers started to examine the
processing of metaphor online, measuring processes that were
happening during the act of comprehension. Most of the early
studies were based on the indexing of reading time or some other
measure of response latency; lately, studies have also employed
neurocognitive imaging techniques such as EEG and fMRI (see
neuroimaging ). Much of the initial theorizing has been based
on speech-act theory, especially as espoused in the work of
John Searle . Following from the distinction between literal and
nonliteral language, the assumption has concerned the process-
ing priority of literal meaning. According to what is now called the
standard pragmatic model , the model would be that the default
processing of language would be to its literal meaning, and that
those inferential processes that seek an alternative, nonliteral
interpretations are only triggered if one fails to fnd a context-
appropriate literal interpretation. Tese assumptions have been
tested in psycholinguistic research that has concretized the stan-
dard model: It assumes that priority to a default literal meaning
would be demonstrated by more rapid reading (or other indices
of processing) of a metaphor in a discourse context that is consis-
tent with its literal sense than in a context that is consistent with
the nonliteral sense; it also assumes that one should not fnd the
processing of metaphoric meaning in conditions in which the lit-
eral sense of an utterance is context appropriate.
More than 20 years of research have failed, in the main, to sup-
port the predictions arising from the standard pragmatic model,
instead showing that in appropriately elaborated contexts, one
can process the metaphoric sense as rapidly as the literal sense of
an utterance and that, using Stroop-like procedures and speed-
accuracy analyses, the initiation of metaphoric interpretation
does not depend on a failure to fnd an appropriate literal inter-
pretation. Tese fndings, though sometimes complicated by the
level of conventionality of the metaphoric expression, have led
to a set of competing theories, all of which have some support,
including models based on the notion of resolving constraint
satisfaction and those that assume that the initial processing of
a word is at an underspecifed schematic level. An increasingly
popular processing model by Rachel Giora attempts to main-
tain processing priority but places the emphasis not on literal
meaning (as Searle had it) but on the saliency of a word (as con-
cretized by familiarity, conventionality, and frequency of use).
According to this theory, one is obligated to process the salient
sense of a word (or expression), regardless of context; contextual
constraints can boost the activation and meaning access of less
salient meanings but will not do so at a cost to the activation and
access of the more salient sense. Te ultimate success and test of
these various theories are being contested, more often these days
with neuroimaging techniques that give a more fne-grained
analysis of online processing than available in the past .
Metaphor
486
Richards , I. A . 1936 . Te Philosophy of Rhetoric . New York : Oxford
University Press .
METAPHOR, ACQUISITION OF
Metaphor is a pervasive aspect of human language (Lakof and
Johnson 1997). Metaphor also plays a central role in abstract
thought by structuring concepts (Gibbs 1994) and leading to
conceptual change (Gentner and Wolf 2000). As such, its rudi-
mentary manifestation at the early ages and its continued growth
over developmental time has been the focus of scientifc inquiry
for several decades. Research on the acquisition of metaphor
followed two main lines of inquiry, each bearing on a diferent
defnition of the term. One approach defned metaphor as a simi-
larity comparison between the perceptual features of objects or
actions, and explored how early children would understand and
produce these so-called perceptual metaphors (e.g., butterfy is
(like) rainbow). Another approach defned metaphor as a con-
ceptual-linguistic mapping between the structural features of two
disparate knowledge domains a source domain, which serves
as the source of vocabulary and conceptual inferences, and a
target domain, to which vocabulary and inferences are extended
metaphorically, and it examined the age at which children begin
to develop an integrated understanding of such structural meta-
phors (e.g., time is motion along a path) as an amalgam of both
source and target domain meanings.
Following is a brief summary of the developmental changes
in childrens metaphorical ability, from the early onset of simple
perceptual metaphors to the later emergence of more complex
structural metaphors.
Metaphor as Similarity: Childrens Early Comprehension
and Production of Simple Perceptual Metaphors
Children can spontaneously produce a variety of perceptual
metaphors that highlight similarities between objects and events
during preschool years (~ages 2.05.0; e.g., Billow 1981; Gardner
et al. 1978; Winner, McCarthy, and Gardner 1980; Winner 1979).
For example, they hold up a half-peeled banana and call it a
fower (Elbers 1988), place a foot in the wastebasket and call
it a boot (Winner 1979), point to a mushroom and say like ice
cream cone (zalkan and Goldin-Meadow 2006), or describe
a ship sailing in the far distance as taking a bath (Chukovsky
1968). Tese early perceptual metaphors typically arise in emerg-
ing symbolic play contexts, in which children frst engage in
imaginative object substitutions (e.g., using a banana as if it were
a phone), and later on, they express similarities between such
objects explicitly in speech (banana is like a phone; Gardner
et al. 1978; see also Sinclair and Stambak 1993 for more informa-
tion on early symbolic play).
Children can use perceptual similarity to sort objects into
categories as early as 18 months (e.g., boxes vs . balls; see Oakes
and Madole 2000 for a review). By preschool age, they can under-
stand and make comparisons between two categorically diferent
objects based on feature-based similarities (Billow 1975; Epstein
and Gamlin 1994; Gardner et al. 1975; Mendelsohn et al. 1984;
Vosniadou and Ortony 1983; Winner, McCarthy, and Gardner
1980) and between two events based on action-based similarities
(Dent 1984). For example, when asked to pick two objects that
of a number of diverse disciplines, and although the on-line pro-
cessing implications of the theory are still ongoing, the tests to
date have been encouraging, often employing brain neuroimag-
ing techniques such as event-related potential (ERP) measure-
ment (see Coulson 2001). Nonetheless, this theory has also been
subjected to criticisms that it, too, is incapable of being disproved
and that it is too indiscriminate, inasmuch as almost anything
that enters working memory can be considered a blend.
Evaluation
Treatments of metaphor as a linguistic and as a cognitive phe-
nomenon coexist today, in much the same way that two species
of hominid have coexisted in our evolutionary history. It remains
to be seen whether the ofspring of one approach will disap-
pear. Despite the diferences, there is a convergence between
approaches that should not be undervalued: Tis convergence
includes the undercutting of the distinction between literal and
nonliteral language; a recognition by both approaches of the
need to consider the richness of examples coming from liter-
ary or philosophical analysis, as well as the controlled rigor that
comes from experimental studies; an emphasis on the role of
cognition and pragmatics (see Carston 2002, for instance, for
an exposition from a relevance theory perspective); and
the growing sentiment (however conceptualized) that the con-
struction of metaphoric meaning is fexible and involves more
of an active on-line interpretive process and less of a mere
arousal of entrenched meaning and that, ultimately, the battle-
ground for theoretical supremacy (or for synthesis of the two
approaches) will depend on data generated and based in the
neurosciences .
Albert N. Katz
WORKS CITED AND SUGGESTED FURTHER READINGS
Black , M . 1962 . Models and Metaphor . Ithaca, NY : Cornell University
Press .
Boroditsky , L . 2000 . Metaphoric structuring: Understanding time
through spatial metaphors . Cognition 75 : 1 28.
Carston , R . 2002 . Toughts and Utterances: Te Pragmatics of Explicit
Communication . Oxford : Blackwell .
Coulson , S . 2001 . Semantic Leaps . New York : Cambridge University
Press .
Eco , U . 1984 . Semiotics and the Philosophy of Language .
Bloomington : Indiana University Press .
Fauconnier G . 1997 . Mappings in Tought and Language .
Cambridge : Cambridge University Press .
Gibbs , R . 1994 . Te Poetics of Mind . New York : Cambridge University
Press .
Giora , R . 2003 . On Our Mind: Salience, Context and Figurative Language .
New York : Oxford University Press .
Giora, R., ed. 2001. Models of fgurative language. Metaphor and
Symbol 16 .3/4 (Special Issue): 145333.
Glucksberg , S . 2001 . Understanding Figurative Language . New
York : Oxford University Press .
Katz , A., C. Cacciari, R. Gibbs, and M. Turner . 1998 . Figurative Language
and Tought . New York : Oxford University Press .
Lakof , G., and M. Johnson . 1980 . Metaphors We Live By . Chicago :
University of Chicago Press .
Ortony , A ., ed. 1993. Metaphor and Tought . 2d ed. New York : Cambridge
University Press .
Metaphor Metaphor, Acquisition of
487
commonalities: Children, at all ages, have no dif culty under-
standing feature-based similarities between objects, but it is with
increasing age that they begin to understand cross-domain map-
pings based on relational structure and, accordingly, produce
explanations that refect this understanding.
Others propose a developmental progression from an under-
standing of metaphor as involving only one domain to a con-
ceptualization of metaphor as involving two domains (Asch
and Nerlove 1960; Cicone, Gardner, and Winner 1981; Schecter
and Broughton 1991; Winner, Rosenstiel, and Gardner 1976).
Tus, children initially focus only on the source domain of the
metaphorical mapping and gradually develop a more inte-
grated understanding of metaphor as involving both a source
and a target domain. For example, in explaining the metaphor-
ical statement the prison guard is a hard rock, children six
to eight years of age focused exclusively on the source domain
meaning of the mapping and provided literal interpretations for
metaphorical statements (e.g., Te guard has hard muscles),
whereas older children and adults were able to consider both the
source and target domain meanings of the mapping, thus pro-
viding explanations that captured the metaphorical meaning
(e.g., Te guard was mean and did not care about the feelings
of prisoners; Winner, Rosenstiel, and Gardner 1976). Similarly,
when asked to extend physical sensation terms onto psycholog-
ical traits (e.g., Can a person be warm/ sweet/ soft?), children
three to seven years of age focused only on the source domain
of the metaphorical mapping and provided literal explanations
for metaphorical statements (e.g., Mommy is sweet because she
cooks sweet things), whereas older children focused on both
domains simultaneously and provided explanations that treated
metaphorical meaning as a diferent but related extension of
the literal meaning (e.g., Hard things and hard people are both
unmanageable; Asch and Nerlove 1960).
Yet another group of researchers argue that the ability to
understand structural metaphors is not determined solely by a
childs age but by a host of other factors, such as the nature of
the source or the target domain (Keil 1986), and the familiarity
of the metaphorical mapping or the source domain (zalkan
2007). For example, fve-year-old children can correctly map
animate terms onto cars (e.g., the car is thirsty), but have dif -
culty understanding metaphors that involve mappings between
taste terms and people (e.g., she is a bitter person; Keil 1986).
Similarly, preschool children can both understand and explain
metaphors that are structured by motion (e.g., Time fies by,
Ideas cross my mind; zalkan 2005, 2007) a domain that
structures a wide range of abstract concepts across diferent lan-
guages of the world but have dif culty deciphering the mean-
ing of metaphors that involve extensions of object properties
(e.g., Te prison guard is a hard rock; Winner, Rosenstiel, and
Gardner 1976). From this perspective, the development of meta-
phorical ability shows diferent trajectories for diferent con-
ceptual domains and metaphorical mappings, based on ones
knowledge of the source and/or the target domain and the famil-
iarity of the metaphorical mapping .
In summary, research on childrens metaphor comprehen-
sion and production shows that children can both understand
and spontaneously produce perceptual metaphors that involve
similarity comparisons by preschool age. However, the ability
go together, children were more likely to group a cherry lollipop
with a toy stop sign, which was similar in shape and color, rather
than matching it with a dissimilar object in the same category
(Mendelsohn et al. 1984). Similarly, when presented with event
triads, children were more likely to pair two events that were
alike (ballerina spinningtop spinning) than to match two events
that were of diferent types (ballerina leapingtop spinning; Dent
1984).
Moreover, fve-year-old children could provide similarity-
based explanations when asked about metaphorical expressions
that involve comparisons between objects such as a butterfy is a
fying rainbow, or a cloud is like a sponge (Gardner et al. 1975;
Gentner 1988; Billow 1975; Malgady 1977). For example, they
would explain the statement a cloud is like a sponge by say-
ing that both clouds and sponges are round and fufy (Gentner
1988), or they would complete the statement he looks as gigan-
tic as by selecting from among multiple choice alternatives
an ending that draws on a feature-based comparison: he looks
as gigantic as a double-decker cone in a babys hand (Gardner
et al. 1975).
Tus, preschool children can both understand similarity
comparisons between two objects or events that are perceptu-
ally alike and spontaneously produce perceptual metaphors and
explanations based on such comparisons in their early com-
munications. Tis ability constitutes an important milestone in
childrens language development. Te ability to express simi-
larities between objects and events based on shared perceptual
features is considered the earliest sign of metaphorical ability in
young children, and accordingly, children are believed to have a
rudimentary level of metaphorical ability as early as preschool
age (Billow 1981; Gardner et al. 1978; Vosniadou 1987; Winner
1979) .
Metaphor as Conceptual-Linguistic Mapping: Childrens
Comprehension and Production of Complex Structural
Metaphors
Childrens early ability to produce feature-based similarity com-
parisons is considered the frst step in the development of more
complex metaphorical abilities, particularly those that involve
structural comparisons between disparate domains (Gardner
et al. 1978; Gentner 1988; Winner 1979). Not surprisingly, chil-
drens mastery of such structural metaphors takes several more
years, extending well into early adolescent years (Asch and
Nerlove 1960; Vosniadou 1987; Winner, Rosenstiel, and Gardner
1976), and diferent researchers propose diferent views con-
cerning how children make this transition.
Some researchers propose a developmental progression
from mappings based on feature-based similarities to map-
pings based on relational structure in childrens metaphorical
abilities (Billow 1975; Gentner 1988, Gentner and Rattermann
1991, Vosniadou and Ortony 1983). For example, in explaining
the metaphorical statement a cloud is like a sponge, fve-year-
old children typically rely on feature-based similarities between
the two objects (e.g., Both clouds and sponges are round and
fufy ) , while older children and adults opt for more relational
explanations (e.g., Both clouds and sponges contain water ;
Gentner 1988). In this view, what drives development is the
shift in focus from feature-based commonalities to relational
Metaphor, Acquisition of
488
zalkan , S. 2005 . On learning to draw the distinction between physi-
cal and metaphorical motion: Is metaphor an early emerging cognitive
and linguistic capacity? Journal of Child Language 32.2 : 1 28.
. 2007 . Metaphors we move by: Childrens developing under-
standing of metaphorical motion in typologically distinct languages .
Metaphor and Symbol 22.2 : 147 68.
zalkan , S. , and S. Goldin-Meadow . 2006 . X is like Y: Te emer-
gence of similarity mappings in childrens early speech and gesture.
In Cognitive Linguistics: Foundations and Fields of Application , ed. G.
Kristianssen , M. Achard, R. Dirven, and F. Ruiz de Mendoza , 229 62.
Berlin: Mouton de Gruyter .
Schecter , B., and J. Broughton. 1991 . Developmental relationships
between psychological metaphors and concepts of life and conscious-
ness . Metaphor and Symbolic Activity 6.2 : 119 43.
Sinclair , M., and M. Stambak . 1993 . Pretend play among three-year-olds .
Mahwah, NJ : Lawrence Erlbaum .
Vosniadou , S. 1987 . Children and metaphors . Child Development
58 : 870 85.
Vosniadou , S., and A. Ortony . 1983 . Te emergence of the literal-meta-
phorical anomolous distinction in young children. Child Development
54 : 154 61.
Winner , E. 1979 . New names for old things: Te emergence of meta-
phoric language . Journal of Child Language 6 : 469 91.
. 1997 . Te Point of Words: Childrens Understanding of Metaphor
and Irony . Cambridge : Harvard University Press .
Winner , E., M. McCarthy, and H. Gardner . 1980 . Te ontogenesis of met-
aphor. In Cognition and Figurative Language , ed. R. P. Honeck and
R. Hofman , 34161. Hillsdale, NJ : Lawrence Erlbaum .
Winner , E., A. K. Rosenstiel, and H. Gardner . 1976 . Te development of
metaphoric understanding . Developmental Psychology 12 : 289 97.
METAPHOR, INFORMATION TRANSFER IN
Te study of metaphor is currently dominated by concep-
tual metaphor theory. One alternative was put forth by Amos
Tversky, then further developed by Andrew Ortony and others.
Tis account begins with the idea that we understand metaphors
by scanning entries in our mental lexicon, transferring relevant
features from a source to a target (see source and target ). In
some versions, the process is viewed as involving a wider range
of information and components of cognitive architecture
beyond semantic features.
Consider the following situation. Smith monopolizes dis-
cussion in a department meeting. Afterward, Doe asks Jones
what she thought of the debate. She replies, Smith is a braying
donkey. Using standard cognitive architecture, we might ana-
lyze this as follows: Jones and Doe both have lexical entries for
donkey , bray , and Smith . Tey also have episodic memories of
the recent department meeting. Te recent events are primed
or partially activated (see priming, semantic; spreading
activation ). Te mention of Smith serves to further activate
the episodic memories of Smith in the meeting. Te lexical entry
for bray involves such elements as produce a sound using vocal
chords. Tis serves to further activate the episodic memories
of vocal chord sounds in the meeting. Specifcally, in conjunc-
tion with Smith, it serves to strongly activate episodic memories
of Smith speaking. Following principles of conversational
implicature , Doe assumes that Jones is making some positive
contribution to the conversation. Tus, Doe looks for new infor-
mation in Joness statement. Tere is no new information in what
to understand and explain more complex metaphors, namely,
those that involve structural mappings between diferent knowl-
edge domains, emerges in late childhood, somewhere between
ages 11.0 to 14.0. Nevertheless, at the same time, childrens early
metaphorical ability is strongly infuenced by the familiarity of
the source and target domains of the metaphor, with more famil-
iar domains and metaphorical relations leading to earlier onset
of metaphor comprehension and production .
eyda zalkan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Asch , S., and H. Nerlove. 1960 . Te development of double function
terms in children. In Perspectives in Psychological Teory , ed. B. Kaplan
and S. Wapner, 4760). New York : International Universities Press .
Billow , R. M. 1975 . A cognitive developmental study of metaphor com-
prehension . Developmental Psychology 11.4 : 415 23.
. 1981 . Observing spontaneous metaphor in children. Journal of
Experimental Child Psychology 31 : 430 45.
Chukovsky , K. 1968 . From Two to Five . Berkeley : University of California
Press .
Cicone , M., H. Gardner, and E. Winner. 1981 . Understanding the psy-
chology in psychological metaphors . Journal of Child Language
8 : 213 16.
Dent , C. H. 1984 . Te developmental importance of motion infor-
mation in perceiving and describing metaphoric similarity . Child
Development 55 : 1607 13.
Elbers , L. 1988 . New names from old words: Related aspects of chil-
drens metaphors and word compounds . Journal of Child Language
15 : 591 617.
Epstein , R. L., and P. J. Gamlin . 1994 . Young childrens comprehension
of simple and complex metaphors presented in pictures and words .
Metaphor and Symbolic Activity 9.3 : 179 91.
Gardner , H., M. Kircher, E. Winner, and D. Perkins . 1975 . Childrens
metaphoric productions and preferences . Journal of Child Language
2 : 125 41.
Gardner , H., E. Winner, R. Bechhofer, and D. Wolf. 1978 . Te develop-
ment of fgurative language . In Childrens Language. Vol. 1. Ed. K.
Nelson, 138. New York : Gardner Press .
Gentner , D . 1988 . Metaphor as structure mapping: Te relational shift .
Child Development 59 : 47 59.
Gentner , D., and M. J. Rattermann . 1991 . Language and the career of
similarity. In Perspectives on Language and Tought: Interrelations
in Development , ed. S. A. Gelman and J. P. Byrnes , 22577. New
Yor k: Cambridge University Press .
Gentner , D., and P. Wolf . 2000 . Metaphor and knowledge change. In
Cognitive Dynamics: Conceptual Change in Humans and Machines , ed.
E. Dietrick and A. Markman , 295342. Mahwah, NJ : Lawrence Erlbaum .
Gibbs , R. 1994 . Te Poetics of Mind . Cambridge : Cambridge University
Press .
Keil , F. C. 1986 . Conceptual domains and the acquisition of metaphor .
Cognitive Development 1 : 73 96.
Lakof , G., and M. Johnson . 1997 . Philosophy in the Flesh . New York : Basic
Books .
Malgady , R. G. 1977 . Childrens interpretation and appreciation of simi-
les . Child Development 48 : 1734 8.
Mendelsohn , E., S. Robinson, H. Gardner, and E. Winner. 1984 . Are pre-
schoolers renamings intentional category violations? Developmental
Psychology 20.2 : 187 92.
Oakes , L. M., and K. L. Madole. 2000 . Te future of infant categoriza-
tion research: A process-oriented approach . Child Developm ent
71.1 : 119 26.
Metaphor, Acquisition of Metaphor, Information Transfer in
489
Indeed, they see many conceptual metaphors as lexicalized. For
example, pass away just has die as one of its literal (lexical-
ized) meanings. It does not operate metaphorically.
One obvious advantage of this account is that it explains the
prominence of mixed metaphors . Consider a sentence such as I
tapped into the good life on the road to acing my degree. Some
elements are lexicalized here. Others are interpreted metaphori-
cally, but only to the extent required by context. In contrast,
conceptual metaphor theory might lead us to expect greater
consistency in the use of standard metaphorical mappings . Te
present account does have more dif culty explaining consis-
tency when it does occur, as when someone says I followed the
straight and narrow path to the reach my destination a degree.
However, it may be possible to account for such consistency by
ordinary processes of priming, both current and historical (see
Hogan 2002) .
One future task is to develop this account in terms of neural
substrates. Consistent with the preceding analysis, neurosci-
entifc research indicates that there is no sharp metaphorical/
literal division. Certain interpretive tasks demand greater activa-
tion of a broad range of meanings before selection. Tese tasks
often involve metaphorical interpretation, but not invariably
(see metaphor, neural substrates of ). Currently, we are
not in a position to examine semantic processing in a suf ciently
fne-grained way to consider the processes posited here. We may
distinguish diferent categories of information (e.g., perception-
related versus motor-related; see semantics, neurobiology
of ), but not precise features, defaults, and so on. Possibilities for
future research may be suggested by modeling these processes
in connectionist networks particularly the key diference
between assuming that defaults apply and assuming that they
do not .
Patrick Colm Hogan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Hogan , Patrick Colm. 2002. A minimal, lexicalist/constituent transfer
account of metaphor . Style 36 .3: 484502.
Ortony , Andrew . 1988. Are emotion metaphors conceptual or lexical?
Cognition and Emotion 2 : 95104.
. 1993 . Te role of similarity in similes and metaphors. In
Metaphor and Tought (2d ed.), ed. Andrew Ortony , 34256. New York :
Cambridge University Press .
Tversky , Amos . 1977 . Features of similarity . Psychological Review
84 : 32752.
METAPHOR, NEURAL SUBSTRATES OF
Te interest in how the brain processes metaphors traces its
origins back to a tradition that regarded fgurative language as
poetic and, hence, the opposite of literal language. Despite its
ubiquity (Lakof and Johnson 1980), the underlying assumption
has been that this diference should be refected both in behav-
ioral (Grice 1975; Searle 1979) and brain mechanisms. In this
entry, we examine this and other long-standing assumptions,
suggesting that the interactions of linguistics with empirical,
neuropsychological, and neuroscientifc research have drawn
a far more complex and, arguably, fascinating picture, not only
about metaphor but also about the brain.
we have isolated thus far that Smith used his vocal chords to
make a sound. Te new information comes with distinctive fea-
tures of the metaphorical source. Specifcally, braying does not
apply to every use of vocal chords. It applies only to a particu-
lar sort of nonlinguistic thus, meaningless and inarticulate
sound . Doe synthesizes this information in working memory .
He understands, roughly, that (in Joness view) Smiths speech
was meaningless and inarticulate.
Tus far, however, the analysis does not distinguish the
understanding of metaphor from that of literal statements. In
both cases, there is a complex synthesis of lexical and episodic
information in working memory; this leads to contextually rel-
evant inference. What, then, is the diference between a meta-
phorical statement and a literal one?
One account begins by making metaphor a matter of inter-
pretation, rather than a matter of some intrinsic linguistic prop-
erty. Specifcally, a speaker intends an utterance metaphorically
when he or she intends the addressee to interpret the utter-
ance metaphorically. What, then, constitutes metaphorical
interpretation?
Our mental lexicons are organized into clusters of infor-
mation bearing on particular objects and types of objects (see
schema, prototype ). Tis information is arranged hierar-
chically. Tere are certain things that we take to be more cru-
cial or defnitive features of a given type of object. For example,
being made from milk is a more important property of cheese
than being white or yellow. Moreover, a range of high-level
properties are default properties. If a default does not apply,
then we commonly have specifable alternatives. Tus, we
assume (as a default) that an unknown person say, Jones
has two arms. But if we learn that she does not, we assume that
she sufered some birth defect or is an amputee, these being the
standard alternatives.
When interpreting a statement literally, we assume that all
default information applies unless it is specifcally contradicted.
Moreover, if a default is contradicted, we assume that one of the
standard alternatives applies. In contrast, when interpreting a
statement metaphorically, we do not assume that default infor-
mation applies. Tat is the defnitive diference. In interpreting a
statement either metaphorically or literally, we scan lexical infor-
mation to glean what is most relevant to the topic at hand. But
when interpreting literally, we assume that unselected, default
information applies as well. We do not assume this when inter-
preting metaphorically.
Te basic diference has several consequences. One is worth
mentioning. All interpretation involves drawing on a range of
associated information, not only that included in the lexical
entries for the source and target items. In metaphorical interpre-
tation, the loosening of hierarchical structures (e.g., through the
nonassumption of defaults) may encourage the incorporation
of more distant associations, including primed emotional asso-
ciations. For example, when Jones refers to Smith as a braying
donkey, she not only characterizes Smith but also expresses and
tries to communicate a certain feeling .
Tis account is similar to conceptual metaphor theory in
stressing cognition. However, it suggests that the cognitive
efects of metaphors need not be profound. Writers adopting this
account commonly view metaphor as operating more locally.
Metaphor, Information Transfer in Metaphor, Neural Substrates of
490
metaphors distinguishes patients with schizophrenia from
healthy controls (Kircher et al. 2007).
Taken together, these fndings suggest that lateralization
in the brains hemispheres is contingent upon such factors as
novelty, semantic and conceptual mapping complexity, and
evoked range of associations, all of which seem to act indepen-
dently of fgurativeness, thus challenging as too simplistic the
notion of a preferential RH processing of stimuli solely by vir-
tue of their metaphoricity. Tese factors, however, are in accor-
dance with an alternative account the fnecoarse semantic
coding hypothesis (Beeman 1998; Jung-Beeman 2005) which
views the LH as adept at processing fnely tuned semantic rela-
tions and the RH as specialized in processing distant semantic
relationships .
NOVELTY. Recent studies indicate that the degree of novelty of
an expression is an important determinant of neural process-
ing. For instance, lesion studies (Giora et al. 2000; Kaplan et
al. 1990) and studies of individuals with Alzheimers disease
(Amanzio et al. 2008), as well as functional magnetic resonance
imaging (fMRI) studies involving healthy participants (Eviatar
and Just 2006), demonstrated that processing non-salient
( ironic, metaphoric ) interpretations relied more heavily
on the RH; processing conventional (metaphoric) meanings
involved the LH. Similarly, a series of fMRI, divided visual feld
(DVF), and event-related potential (ERP) studies demonstrated
increased activation of RH areas during processing of nonsa-
lient interpretations of novel metaphors (Arzouan, Goldstein,
and Faust 2007; Faust and Mashal 2007; Mashal and Faust 2008;
Mashal, Faust, and Hendler 2005; Mashal et al. 2007) and lit-
eral/compositional interpretations of idioms (Mashal et al.
2008). And while RH advantage was demonstrated in process-
ing nonsalient interpretations of novel metaphors during frst
exposure, repeated exposure benefted the LH (Mashal and
Faust 2009) .
COMPLEXITY. Tat RH recruitment increases with complex sen-
tences has been demonstrated by a number of studies (Jung-
Beeman 2005). Tis has also been seen as typifying conceptual
mapping complexity (Coulson and Van Petten 2002), thus intro-
ducing another parameter that may determine processing and
operate regardless of metaphoricity. Further work is awaited to
establish this view .
RANGE OF SEMANTIC ASSOCIATIONS. Range of semantic asso-
ciations, also termed degree of open-endedness, can be seen
as determined by the extent to which a stimulus evokes a wide
network of semantic associations (Black 1993). In a fMRI study,
Stringaris et al. (2006) showed that deciding that a given probe
was unrelated to a previous neutral context triggered activation
of frontal RH areas following open-ended (metaphoric) contexts
( Some answers are straight ) but not following more restricted
(literal) contexts ( Some answers are emotional ). In the case of
the open-ended primes (see priming , semantic ), both nega-
tive and positive decisions elicited the same neural responses.
Indeed, higher degree of open-endedness may lead to increased
RH activation, probably because of the evocation of remotely
related associations (Jung-Beeman 2005). As shown by Mashal
Is Metaphor Really So Different?
Since the 1970s, the assumption that metaphors are processed
diferently from literals has come under close scrutiny. For
example, on the basis of psycholinguistic experiments, it has
been argued that in the presence of rich and supportive con-
text, metaphors and literals are processed along the same routes
(Gibbs 1994; Ortony et al. 1978).
Although some metaphoric and literal expressions require
similar processes (Glucksberg 2001), it has also become increas-
ingly evident that the categories used are in themselves heteroge-
neous. For instance, some literals ( the ring was made of tin, with
a pebble instead of a gem ) require more complex (metaphor-like)
conceptual mapping processes than others ( Tat stone we saw
in the natural history museum is a gem ; Coulson and Van Petten
2002). Others ( curl up and dye ) are more appealing though harder
to process than metaphoric equivalents ( curl up and die ; Giora
2003). Metaphors are not all alike either; some are novel, hav-
ing nonsalient metaphoric interpretations that are usually more
appealing yet harder to process than those that are conventional
and salient (Giora et al. 2004). Furthermore, some metaphoric
stimuli, though relatively conventional, may still be more open-
ended than others and, when functioning as a context, give rise
to a wider range of associations (Stringaris et al. 2006).
In fact, recent fndings indicate that notions such as degree
of salience, complexity, or open-endedness may be more suit-
able for describing the complexity of some of the phenomena
in question and span the metaphor-literal divide. Furthermore,
while these notions may, to an extent, overlap, none of them is
specifc to metaphor.
Is Metaphor Processed Differently in the Brain?
Consistent with the prevailing view of the right hemisphere
(RH) as being more adept at creativity than the left hemi-
sphere (LH), early lesion studies have been interpreted as
evidence that metaphors rely more heavily than their literal
counterparts on regions in the RH (Winner and Gardner 1977).
However, Ellen Winner and Howard Gardners study actually
reveals that patients with RH lesions were not insensitive to met-
aphor (1977, 725) when ofering verbal explications to fgurative
stimuli, although they tended to erroneously select literal over
metaphoric interpretations in a picture-matching task. Similarly,
the results of the earliest imaging study in the feld (Bottini et al.
1994) were also seen as supporting a RH predominance for meta-
phor comprehension. However, alternative explanations may be
more appropriate, given that the linguistic items used also dif-
fered on categories other than sensu strictu metaphoricity .
Indeed, subsequent studies have challenged the purported
predominance of the RH by demonstrating that when conven-
tional metaphors compared to literals are processed, the LH is
more active (Ahrens et al. 2007; Lee and Dapretto 2006; Oliveri,
Romero, and Papageno 2004), perhaps refecting retrieval from
semantic stores. In fact, most recent research suggests that in
the absence of a rich biasing context, the hemispheres are insen-
sitive to fgurativeness. Rather, the RH is more sensitive than the
LH to novel, nonsalient interpretations and poetic associations,
to complexity, and to open-endedness (Blasko and Kazmerski
2006; Giora 2007). Tis is corroborated by a recent fmri study
showing that failure to recruit RH areas when processing novel
Metaphor, Neural Substrates of
491
Coulson , Seana, and Cyma Van Petten . 2002 . Conceptual integration
and metaphor comprehension: An ERP study . Memory & Cognition
30 : 958 68.
. 2007 . A special role for the right hemisphere in metaphor com-
prehension? ERP evidence from hemifeld presentation . Brain
Research 1146 : 128 45.
Eviatar , Zohar, and Marcel Just . 2006 . Brain correlates of discourse
processing: An fMRI investigation of irony and metaphor comprehen-
sion . Neuropsychologia 44 : 2348 59.
Faust , Miriam, and Nira Mashal . 2007 . Te role of the right cerebral
hemisphere in processing novel metaphoric expressions taken from
poetry: A divided visual feld study . Neuropsychologia 45 : 860 70.
Gibbs , W. Raymond, Jr . 1994 . Te Poetics of Mind . Cambridge : Cambridge
University Press .
Giora , Rachel . 2003 . On Our Mind: Salience, Context and Figurative
Language . New York : Oxford University Press .
Giora , Rachel , ed. 2007. Is Metaphor Unique? Neural Correlates of
Nonliteral Language. Brain and Language 100 : 2.
Giora , Rachel, Ofer Fein, Ann Kronrod, Idit Elnatan, Noa Shuval, and
Adi Zur . 2004 . Weapons of mass distraction: Optimal innovation and
pleasure ratings . Metaphor and Symbol 19 : 11541 .
Giora , Rachel, Eran Zaidel, Nachum Soroker, Gila Batori, and Asa
Kasher . 2000 . Diferential efects of right- and left-hemisphere dam-
age on understanding sarcasm and metaphor . Metaphor and Symbol
15 : 63 83.
Glucksberg , Sam . 2001 . Understanding Figurative Language: From
Metaphors to Idioms . New York : Oxford University Press .
Grice , H. Paul . 1975 . Logic and conversation. In Speech Acts: Syntax
and Semantics. Vol. 3. Ed. Peter Cole and Jerry Morgan , 41 58. New
York : Academic Press .
Jung-Beeman , Mark . 2005 . Bilateral brain processes for comprehending
natural language . Trends in Cognitive Sciences 9 : 512 18.
Kacinik , A. Natalie, and Christine Chiarello . 2007 . Understanding meta-
phors: Is the right hemisphere uniquely involved? Brain and Language
100 : 188 207.
Kaplan , Joan A., Hiram H. Brownell, Janet R. Jacobs, and Howard
Gardner . 1990 . Te efects of right hemisphere damage on the prag-
matic interpretation of conversational remarks . Brain and Language
38 : 315 33.
Kircher , T. J. Tilo, Dirk T. Leube, Michael Erb, Wolfgang Grodd, and
Alexander M. Rapp . 2007 . Neural correlates of metaphor processing
in schizophrenia . NeuroImage 34 : 281 9.
Kutas , Marta . 2006 . One lesson learned: Frame language processing
literal and fgurative as a human brain function . Metaphor and
Symbol 21: 285 325.
Lakof , George, and Mark Johnson . 1980 . Metaphors We Live By .
Chicago : University of Chicago Press .
Lee , S. Susan, and Mirella Dapretto . 2006 . Metaphorical vs. literal word
meanings: fMRI evidence against a selective role of the right hemi-
sphere . NeuroImage 29 : 536 44.
Mashal , Nira, and Miriam Faust . 2008 . Right hemisphere sensitivity to
novel metaphoric relations: Application of the signal detection the-
ory . Brain and Language 104 .2: 10312.
. 2009. Conventionalization of novel metaphors: A shift in hemi-
spheric asymmetry. In Laterality 14.6:57389.
Mashal , Nira, Miriam Faust, and Talma Hendler . 2005 . Te role of
the right hemisphere in processing nonsalient metaphorical mean-
ings: Application of principal components analysis to fMRI data .
Neuropsychologia 43. 14: 2084 100.
Mashal , Nira , Miriam Faust , Talma Hendler , and Mark Jung-Beeman .
2007 . An fMRI investigation of the neural correlates underlying the
processing of novel metaphoric expressions . Brain and Language
100 : 115 26.
et al. (in press), RH areas were uniquely involved when novel lit-
eral interpretations of familiar idioms (involving their familiar
idiomatic meanings as well) were deliberated on .
CONTEXTUAL INFORMATION. Contextual factors involved in
processing (such as biasing information, task, mood, or expe-
rience) further argue against a specifc and invariant brain
locus for metaphor (Kutas 2006). Tey show that recruitment
of neural networks depends upon factors other than metapho-
ricity per se. For instance, in Coulson and Van Petten (2007),
RH advantage in processing novel metaphors disappears in
the presence of biasing information. In Kacinik and Chiarello
(2007), both hemispheres were activated by metaphors, but
only the LH-response was context sensitive, thereby restricting
the range of possible alternatives. Conversely, the response in
the RH indicated retention of alternatives available for process-
ing. Findings in Rapp et al. (2007) indicate that the type of task
is an additional determinant of processing. When participants
had to judge the emotional valence of connotations, meta-
phors elicited LH regions, despite their novelty. In Stringaris
et al. (2006), familiar metaphors activated RH areas when a
coherence judgment was required; however, when a meaning-
fulness judgment was required, same stimuli evoked LH areas
(Stringaris et al. 2007). In Blasko and Kazmerski (2006), it was
individual diferences in experience that mattered: Poets and
nonpoets difered as to which brain areas were recruited when
reading poetry .
In sum, recent research, involving a wide range of method-
ologies, does not provide support for the long-assumed special
status of metaphor in language. Instead, it shows that the pro-
cessing of metaphors in the brain depends on a great number of
factors beyond fgurativeness .
Rachel Giora and Argyris K. Stringaris
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ahrens , Kathleen, Ho-Ling Liu, Chia-Ying Lee, Shu-Ping Gong, Shin-Yi
Fang, and Yuan-Yu Hsu. 2007 . Functional MRI of conventional and
anomalous metaphors in Mandarin Chinese . Brain and Language
100 : 163 71.
Amanzio , Martina, Giuliano Geminiani, Daniela Leotta, and Stefano
Cappa . 2008 . Metaphor comprehension in Alzheimers dis-
ease: Novelty matters . Brain and Language 107 .1: 110.
Arzouan , Yossi, Abraham Goldstein, and Miriam Faust . 2007 . Brain
waves are stethoscopes: ERP correlates of novel metaphor compre-
hension . Brain Research 1160 : 69 81.
Beeman , Ma rk. 1998. Coarse semantic coding and discourse compre-
hension. In Right Hemisphere Language Comprehension: Perspectives
from Cognitive Neuroscience, ed. Mark Beeman and Christine Chiarello ,
25584. Mahwah, NJ : Erlbaum .
Black , Max . 1993 . More about metaphor. In Metaphor and Tought (2d
ed.), ed. Andrew Ortony , 1941. Cambridge : Cambridge University
Press .
Blasko , G. Dawn, and Victoria A. Kazmerski . 2006 . ERP correlates of
individual diferences in the comprehension of nonliteral language .
Metaphor and Symbol 21.4 : 267 84.
Bottini , Gabriella., Corcoran Rhiannon , Roberto Sterzi , Eraldo Paulesu,
P. Schenone, P. Scarpa , et al. 1994. Te role of the right hemisphere
in the interpretation of fgurative aspects of language: A positron emis-
sion tomography activation study . Brain 117 : 1241 53.
Metaphor, Neural Substrates of
492
of evidence from a number of linguists who are native speak-
ers of the respective languages, Zoltn Kvecses (2000) points
out that English, Japanese, Chinese, Hungarian, Wolof, Zulu,
Polish, and others possess the metaphor AN ANGRY PERSON
IS A PRESSURIZED CONTAINER, to various degrees. Ning Yus
(1995, 1998) work indicates that that the metaphor HAPPINESS
IS UP is also present not only in English but also in Chinese. Te
system of metaphors called the event structure metaphor (Lakof
1993) includes such submetaphors as CAUSES ARE FORCES,
STATES ARE CONTAINERS, PURPOSES ARE DESTINATIONS,
ACTION IS MOTION, DIFFICULTIES ARE IMPEDIMENTS
(TO MOTION), and so forth. Remarkably, this set of submeta-
phors occurs in such widely diferent languages and cultures as
Chinese (Yu 1998) and Hungarian (Kvecses 2005), in addition
to English. Eve Sweetser (1990) noticed that the KNOWING IS
SEEING and the more general the MIND IS THE BODY meta-
phors can be found in many European languages and are prob-
ably good candidates for (near-)universal metaphors. As a fnal
example, George Lakof and Mark Johnson (1999) describe the
metaphors used for ones inner life in English. It turns out that
metaphors such as SELF-CONTROL IS OBJECT POSSESSION,
SUBJECT AND SELF ARE ADVERSARIES, and THE SELF IS A
CHILD are shared by English, Japanese, and Hungarian. Given
that ones inner life is a highly elusive phenomenon and, hence,
would seem to be heavily culture and language dependent, one
would expect a great deal of signifcant cultural variation in such
a metaphor. All in all, then, we have a number of cases that con-
stitute near-universal or potentially universal conceptual meta-
phors, though not universal metaphors in the strong sense .
How Can We Have (Near-)Universal Metaphors?
How is it possible that such conceptual metaphors exist in diverse
languages and cultures? After all, the languages belong to very
diferent language families and represent very diferent cultures
of the world. Several answers to this question lend themselves
for consideration. First, we can suggest that by coincidence, all
these languages developed the same conceptual metaphors for
happiness, time, purpose, and so on. Second, we can consider
the possibility that languages borrowed the metaphors from
one another. Tird, we can argue that there may be some uni-
versal basis for the same metaphors to develop in the diverse
languages.
Let us take as an example the HAPPINESS IS UP concep-
tual metaphor, frst discussed by Lakof and Johnson (1980) in
English. Tis conceptual metaphor can be seen in such linguis-
tic expressions as feeling up , being on cloud nine , being high , and
others. Yu (1995, 1998) noticed that the conceptual metaphor
can also be found in Chinese. And evidence shows that it also
exists in Hungarian. Following are some linguistic examples (Yu
used the grammatical abbreviations PRT = particle and ASP =
aspect marker):
C hinese:
happy is up
Ta hen gao-xing.
he very high-spirit
He is very high-spirited/happy.
. 2008 . Hemispheric diferences in processing the literal interpre-
tation of idioms: Converging evidence from behavioral and fMRI stud-
ies. Cortex 44 .7: 84860.
Oliveri , Massimiliano., Leonor Romero, and Costanza Papagno . 2004 .
Left but not right temporal involvement in opaque idiom comprehen-
sion: A repetitive transcranial magnetic stimulation study. Journal of
Cognitive Neuroscience 16 : 848 55.
Ortony , Andrew, Diane L. Schallert, Ralph E. Reynolds, and Stephen J.
Antos . 1978 . Interpreting metaphors and idioms: Some efects of
context on comprehension . Journal of Verbal Learning and Verbal
Behavior 17 : 465 77.
Rapp , M. Alexander , Dirk T. Leube , Michael Erb , Wolfgang Grodd , and
Tilo T. J. Kircher . 2007 . Laterality in metaphor processing: Lack of evi-
dence from functional magnetic resonance imaging for the right hemi-
sphere theory . Brain and Language 100 : 142 9.
Searle , John . 1979 . Expression and Meaning . Cambridge : Cambridge
University Press .
Schmidt , L. Gwen , Casey J. DeBuse , and Carol A. Seger . 2007 . Right
hemisphere metaphor processing? Characterizing the lateralization
semantic processes . Brain and Language 100 : 127 41.
Stringaris , K. Argyris, Nicholas C. Medford, Vincent C. Giampietro,
Michael J. Brammer, and Anthony S. David . 2007 . Deriving mean-
ing: Distinct neural mechanisms for metaphoric, literal, and non-
meaningful sentences . Brain and Language 100 : 150 62.
Stringaris , K. Argyris, Nicholas C. Medford, Rachel Giora, Vincent C.
Giampietro, Michael J . Brammer , and Anthony S. David . 2006 . How
metaphors infuence semantic relatedness judgments: Te role of the
right frontal cortex . NeuroImage 33 : 784 93.
Winner , Ellen, and Howard Gardner . 1977 . Te comprehension of meta-
phor in brain-damaged patients . Brain 100 : 717 29.
METAPHOR, UNIVERSALS OF
Universal Metaphors?
Native speakers of all languages use a large number of meta-
phors when they communicate about the world (Lakof and
Johnson 1980). Such metaphorically used words and expres-
sions may vary considerably across diferent languages. For
example, the idea that is expressed in English with the words
spending your time is expressed in Hungarian as flling your time .
Te images that diferent languages and cultures employ to
code meanings can be extremely diverse. Given this diversity, it
is natural to ask: Are there any universal metaphors at all, if by
universal we mean those linguistic metaphors that occur in each
and every language? Tis question is dif cult because it goes
against our everyday experiences and intuitions as regards meta-
phorical language in diverse cultures; it would also be extremely
dif cult to study, given that there are 4,0006,000 languages spo-
ken around the world today.
If we go beyond looking at metaphorically used linguistic
expressions in diferent languages, however, and look at con-
ceptual metaphors instead of linguistic metaphors, we begin
to notice that many conceptual metaphors appear in a wide
range of languages. For example, Hoyt Alverson (1994) found
that the TIME IS SPACE conceptual metaphor can be found
in such diverse languages and cultures as English, Mandarin
Chinese, Hindi, and Sesotho. Many other researchers suggested
that the same conceptual metaphor is present in a large num-
ber of additional languages. Several other conceptual metaphors
appear in a large number of diferent languages. On the basis
Metaphor, Universals of
493
metaphors from primary metaphors. His idea was that com-
plex metaphors (e.g., THEORIES ARE BUILDINGS) are com-
posed of primary metaphors (e.g., LOGICAL ORGANIZATION IS
PHYSICAL STRUCTURE). Te primary metaphors consist of cor-
relations of a subjective experience with a physical experience.
As a matter of fact, it turned out that many of the conceptual
metaphors discussed in the cognitive linguistic literature
are primary metaphors in this sense. For instance, HAPPY IS UP
is best viewed as a primary metaphor, wherein being happy is a
subjective experience and being physically up is a physical one
that is repeatedly associated with it. Other primary metaphors
include MORE IS UP, PURPOSES ARE DESTINATIONS, and
INTIMACY IS CLOSENESS. On this view, it is the primary meta-
phors that are potentially universal.
Primary metaphors function at a fairly local and specifc level
of conceptualization, and, hence, in the brain. At the same time,
we can also assume the existence of much more global meta-
phors (see also generic- and specific-level metaphors ).
For example, animals are commonly viewed as humans and
humans as animals; humans are commonly conceptualized as
objects and objects as humans, and so on. A famous example of
the objects as humans metaphor was described by Keith Basso
(1967), who showed that in the language of the Western Apache,
cars are metaphorically viewed in terms of the human body. In
addition, the work of Bernd Heine and his colleagues (Heine,
Claudi, and Hnnemeyer 1991; Heine 1995; Heine and Kuteva
2002) reveals other large-scale metaphorical processes that peo-
ple seem to employ nearly universally; for example, spatial rela-
tions are commonly understood as parts of the human body (e.g.,
the head means up and the feet mean down). Tese conceptual
metaphors seem to be global design features of the brain/mind
of human beings .
It seems to be clear at this point that commonality in human
experience is a major force shaping the metaphors we have. It is
this force that gives us many of the metaphors that we can take
to be near-universal or potentially universal. But commonality
in human experience is not the only force that plays a role in
the process of establishing and using metaphors. Tere are also
countervailing forces that work against universality in metaphor
production .
Causes of Metaphor Variation
Heines work also shows that not even such global metaphors as
SPATIAL RELATIONS ARE PARTS OF THE BODY are universal in
an absolute sense. Tere are languages in which spatial relations
are conceptualized not as the human but as the animal body.
Heine points out that such languages function in societies where
animal husbandry is a main form of subsistence. Tis leads us
to the question: What causes our metaphors to vary as they do?
It is convenient to set up two large groups of causes: diferential
experience and diferential cognitive preferences. Diferential
experience involves diferences in the social-cultural context, in
social and personal history, and in what we can term social and
personal concern or interest (see Kvecses 2005).
One example of how the social-cultural context can shape
conceptual metaphors is provided by Dirk Geeraerts and Stephan
Grondelaers (1995). Tey note that in the Euro-American tradi-
tion, it is the classical-medieval notion of the four humors from
Ta xing congcong de.
he spirit rise-rise PRT
His spirits are rising and rising./Hes pleased and excited.
Zhe-xia tiqi le wo-de xingzhi.
this-moment raise ASP my mood
Tis time it lifted my mood/interest.
H ungarian:
happiness is up
Ez a flm feldobott.
this the flm up-threw-me
Tis flm gave me a high.-Tis flm made me happy.
Majd elszll a boldogsgtl.
almost away-fies-he/she the happiness-from
He/she is on cloud nine.
English, Mandarin Chinese, and Hungarian (a Finno-Ugric
language) belong to diferent language families , which
developed independently for much of their history. It is also
unlikely that the three languages had any signifcant impact on
one another in their recent history. Tis is not to say that such an
impact never shapes particular languages as regards their meta-
phors (e.g., the processes of globalization and the widespread
use of the Internet may popularize certain conceptual meta-
phors, such as TIME IS A COMMODITY), but only to suggest
that the particular HAPPINESS IS UP metaphor does not exist in
the three languages because, say, Hungarian borrowed it from
Chinese and English from Hungarian.
So how did the same conceptual metaphor emerge, then, in
these diverse languages? Te best answer seems to be that there is
some universal bodily experience that led to its emergence. Lakof
and Johnson argued early that English has the metaphor because
when we are happy, we tend to be physically up, move around, be
active, jump up and down, smile (i.e., turn up the corners of the
mouth), rather than down, inactive, and static, and so forth. Tese
are undoubtedly universal experiences associated with happiness
(or, more precisely, joy), and they are likely to produce potentially
universal (or near-universal) conceptual metaphors. Te emer-
gence of a potentially universal conceptual metaphor does not,
of course, mean that the linguistic expressions themselves will be
the same in diferent languages that possess a particular concep-
tual metaphor (Barcelona 2000; Maalej 2004).
Kvecses (1990, 2000) proposed, furthermore, that the uni-
versal bodily experiences can be captured in the conceptual
metonymies associated with particular concepts. Specifcally,
in the case of emotion concepts such as happiness, anger, love,
pride, and so forth, the metonymies correspond to various kinds
of physiological, behavioral, and expressive reactions. Tese
reactions provide us with a profle of the bodily basis of emotion
concepts. Tus, the metonymies give us a sense of the embodied
nature of concepts, and the embodiment of concepts may be
overlapping, that is, (near-)universal, across diferent languages
and language families. Such universal embodiment may lead to
the emergence of shared conceptual metaphors.
Joseph Grady (1997a, 1997b) developed the Lakof-Johnson
view further by proposing that we need to distinguish complex
Metaphor, Universals of
494
metaphorical conceptualization of a target domain. A case in
point is the conceptualization of anger in English and Chinese.
As studies of the physiology of anger across several unrelated
cultures show, increase in skin temperature and blood pres-
sure are universal physiological correlates of anger (Levenson
et al. 1992). Tis accounts for the ANGER IS HEAT metaphor in
English and in many other languages. However, Kings and Yus
work mentioned earlier suggest that the conceptualization of
anger in terms of heat is much less prevalent in Chinese than it
is in English. In Chinese, the major metaphors of anger seem to
be based on pressure not heat. Tis indicates that speakers of
Chinese have relied on a diferent aspect of their physiology in
the metaphorical conceptualization of anger than speakers of
English. Te major point is that in many cases, the universality
of the experiential basis does not necessarily lead to universally
equivalent conceptualization at least not at the specifc level of
hot fuids .
Are there any diferences in the way the cognitive processes
of metaphor versus metonymy are used in diferent languages
and cultures? Jonathan Charteris-Black (2003) examined in great
detail how and for what purpose three concepts mouth, tongue,
and lip are fguratively utilized in English and Malay. He found
similarities in metaphorical conceptualization. For example,
in both languages, the same underlying conceptual metaphor
(e.g., MANNER IS TASTE) accounts for expressions like honey-
tongued and lidah manis (tongue sweet), and in both languages
such expressions are used for the discourse function of evaluat-
ing (especially negatively) what a person says. However, he also
found that the fgurative expressions involving the three concepts
tended to be metonymic in English and metaphoric in Malay. In
English, more than half of the expressions were metonyms, while
in Malay the vast majority of them showed evidence of metaphor
(often in combination with metonymy). For example, while met-
onymic expressions like tight-lipped abound in English, such
expressions are much less frequent in Malay. It seems that, at
least in the domain of speech organs, the employment of these
concepts by means of fgurative processes is partially culture
specifc.
In sum, metaphorical linguistic expressions may vary widely
cross-culturally, but many conceptual metaphors appear to be
potentially universal or near-universal. Tis happens because
people across the world share certain bodily experiences.
However, even such potentially universal metaphors may dis-
play variation in their specifc details because people do not
use their cognitive capacities in the same way from culture to
culture. Moreover, shared conceptual metaphors may vary
cross-culturally in the frequency of their use. Finally, many con-
ceptual metaphors are unique to particular (sub)cultures or sets
of cultures because of diferences in such factors as the social-
cultural context, history, or human concern that characterize
these cultures .
Zoltn Kvecses
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Alverson , Hoyt . 1994 . Semantics and Experience: Universal Metaphors
of Time in English, Mandarin, Hindi, and Sesotho . Baltimore : Johns
Hopkins University Press .
which the Euro-American conceptualization of anger (as well as
that of emotion in general) derived. Te humoral view maintains
that the four fuids (phlegm, black bile, yellow bile, and blood)
and the temperatures associated with them regulate the vital
processes of the human body. Te humors were also believed to
determine personality types (such as sanguine, melancholy, etc.)
and account for a number of medical problems. Te humoral
view exerted a major impact on the emergence of the European
conception of anger as a hot fuid in a pressurized container. By
contrast, Brian King (1989) and Yu (1995 and 1998) suggest that
the Chinese concept of nu (corresponding to anger) is bound up
with the notion of qi , that is, the energy that fows through the
body. Qi in turn is embedded in not only the psychological (i.e.,
emotional) but also the philosophical and medical discourse of
Chinese culture and civilization. When qi rises in the body, there
is anger (nu). Without the concept of qi, it would be dif cult to
imagine the view of anger in Chinese culture. Tus, emotion
concepts, such as anger in English, dh in Hungarian (the two
representing European culture), and nu in Chinese, are in part
explained in the respective cultures by the culture-specifc con-
cepts of the four humors and qi, respectively. It appears that the
culture-specifc key concepts that operate in particular cultures
account for many of the specifc-level diferences among the vari-
ous anger-related concepts and the PRESSURIZED CONTAINER
metaphor .
As an example of how diferences in human concern can
create new metaphors, consider some well-known concep-
tual metaphors for sadness: SADNESS IS DOWN, SADNESS IS
A BURDEN, and SADNESS IS DARK. Te counterpart of sad-
ness is depression in a clinical context. Linda McMullen and
John Conway (2002) studied the metaphors that people with
episodes of depression use and, with one exception, found the
same conceptual metaphors for depression that nondepressed
people use for sadness. Tey identifed the unique metaphor as
DEPRESSION IS A CAPTOR. Why dont merely sad people talk
about sadness as being a captor? Most people do not normally
talk about being trapped by, wanting to be free of, or wanting
to break out of sadness, although these are ways of talking and
thinking about depression in a clinical context. It makes sense to
suggest that people with depression use this language and way of
thinking about their situation because it faithfully captures what
they experience and feel. Teir deep concern is with their unique
experiences and feelings that set them apart from people who
do not have them. It is this concern that gives them the CAPTOR
metaphor for depression (see also emotion and language ) .
People can employ a variety of diferent cognitive operations
in their efort to make sense of experience. For example, what I
call experiential focus can have an impact on the specifc details
of the conceptual metaphors used, and what is conceptualized
metaphorically in one culture can predominantly be conceptu-
alized by means of metonymy in another (Kvecses 2005). Te
universal bodily basis on which universal metaphors could be
built may not be utilized in the same way or to the same extent
in diferent languages. What experiential focus means is that
diferent peoples may be attuned to diferent aspects of their
bodily functioning in relation to a metaphorical target domain
(see source and target ) or that they can ignore or downplay
certain aspects of their bodily functioning with respect to the
Metaphor, Universals of
495
METER
Verse is text that is divided into lines ( verse lines ). One of the
subtypes of verse is metrical verse . In metrical verse, the length
of the lines is controlled by a set of rules (indirectly, all metrical
rules count syllables ). Te lines in metrical verse are usually
subject to other restrictions as well, most commonly restrictions
on the rhythm of the line (based on stress, syllable weight,
or lexical tone), and/or on a requirement that a syllable in a
specifc line-internal location be word-initial or word-fnal (a
caesura rule). Some meters also include rules about rhyme or
alliteration.
Although verse is probably a universal (see poetic form,
universals of ), found in all oral or literary traditions, there are
poetic traditions without metrical verse, of which perhaps the
best known is the Hebrew poetry of the Old Testament, which
is based on syntactic parallelism rather than on counted syl-
lables. Metrical verse is found in European literatures (Greek,
English, the various Celtic, Germanic, Romance, and Slavic lit-
eratures, also Finnish), in Arabic and Islamic literatures (e.g.,
Persian, Urdu, Turkish, Hausa), and in literatures less clearly
infuenced by Arabic (such as Berber and Somali), in the litera-
tures of South Asia (e.g., Sanskrit, Pali, Hindi, Malayalam, Tamil),
of Southeast Asia (e.g., Tai, Burmese, Vietnamese), and of East
Asia (e.g., Chinese, Korean, Japanese). Metrical verse is reported
to be largely or completely absent in the poetry of ancient Semitic
literatures and of Australia, non-Islamic Africa, the Americas,
and New Guinea, but this may just be because researchers have
not been looking for it (feldworkers far too rarely ask questions
about the poetics or poetic practice of a culture).
Te variety of meters can be illustrated by some examples.
English iambic pentameter requires a 10- or 11-syllable line,
with even-numbered syllables tending to have stress . Te French
alexandrin requires a line of 12 or 13 syllables, with the sixth syl-
lable both stressed and word-fnal . Swahili shairi requires a line
of 16 syllables, with the eighth syllable word-fnal (and no con-
trol over rhythm) . Greek iambic trimeter requires a line of 12 syl-
lables, with even-numbered syllables heavy (containing a long
vowel or ending in two consonants), and the third, seventh and
eleventh syllables light (containing a short vowel ending in at
most one consonant) . Arabic kamil requires a line of between 12
and 15 syllables with a complex rhythmic control (in the short-
est line the third, seventh and eleventh syllables are light and the
others heavy) . Sanskrit sardulavikridita requires a line of 19 syl-
lables with the aperiodic rhythm heavy heavy heavy light light
heavy light heavy light light light heavy (word boundary) heavy
heavy light heavy heavy light heavy. Japanese meters require
lines of fve or seven light syllables (but permit a heavy syllable
to substitute for two light syllables). A genre of Vietnamese pairs
a six-syllable line with an eight-syllable line, in which the sec-
ond, and sixth (and eighth) syllables belong to one tonal class
and the fourth to another . Germanic alliterative meter requires
between two and four stressed syllables, at least two of which
must alliterate .
In literary studies, meter is usually discussed as an aid to
interpretation, and less attention has been paid to the the-
ory that underlies the meter than is desirable. Te approach to
meter most common in such studies is the foot combination and
Barcelona , Antonio . 2000 . On the plausibility of claiming a metonymic
motivation for conceptual metaphor. In Metaphor and Metonymy at
the Crossroads , ed. A. Barcelona , 3158. Berlin : Mouton de Gruyter .
Basso , Keith H . 1967 . Semantic aspects of linguistic acculturation .
American Anthropologist , n.s., 69.5 : 471 7.
Charteris-Black , Jonathan . 2003 . Speaking with forked tongue: A com-
parative study of metaphor and metonymy in English and Malay
phraseology . Metaphor and Symbol 18.4 : 289 310.
Geeraerts , Dirk, and Stephan Grondelaers . 1995 . Looking back at
anger: Cultural traditions and metaphorical patterns . In Language
and the Cognitive Construal of the World , ed. J . Taylor and R. MacLaury ,
15379. Berlin : G ruyter .
Grady , Joseph . 1997 a. Foundations of meaning: Primary metaphors and
primary scenes . Ph.D. d i ss., University of California at Berkeley .
. 1997 b. Teories are buildings revisited . Cognitive Linguistics
8 : 267 90.
Haspelmath , Martin . 1997 . From Space to Time: Temporal Adverbials in
the Worlds Languages . Munich and Newcastle : Lincom Europa .
Heine , Bernd . 1995 . Conceptual grammaticalization and prediction. In
Language and the Cognitive Construal of the World , ed. J. Taylor and R.
MacLaury , 11935 . Berlin : Mouton de Gruyter .
Heine , Bernd, Ulrike Claudi, and Friederike Hnnemeyer . 1991 .
Grammaticalization: A Conceptual Framework . Chicago : University of
Chicago Press .
Heine , Bernd, and Tania Kuteva . 2002 . World Lexicon of
Grammaticalization . Cambridge : Cambridge University Press .
King , Brian . 1989 . Te conceptual structure of emotional experience in
Chinese . Ph.D. diss., Ohio State University .
Kvecses , Zoltn . 1990 . Emotion Concepts . Berlin and New York : Springer-
Verlag .
. 2000 . Metaphor and Emotion . New York and Cambridge :
Cambridge University Press .
. 2005 . Metaphor in Culture: Universality and Variation . Cambridge
and New York : Cambridge University Press .
Lakof , George . 1993 . Te contemporary theory of metaphor. In
Metaphor and Tought , ed. A. Ortony, 20251. Cambridge : Cambridge
University Press .
Lakof , George, and Mark Johnson . 1980 . Metaphors We Live By .
Chicago : University of Chicago Press .
. 1999 . Philosophy in the Flesh . New York : Basic Books .
Levenson , R. W., P. Ekman, K. Heider, and W. V. Friesen . 1992 . Emotion
and autonomic nervous system activity in the Minangkabau of West
Sumatra . Journal of Personality and Social Psychology 62 : 972 88.
Maalej , Zouhair . 2004 . Figurative language in anger expressions in
Tunisian Arabic: An extended view of embodiment . Metaphor and
Symbol 19.1 : 51 75.
McMullen , Linda, and John Conway . 2002 . Conventional metaphors for
depression. In Verbal Communication of Emotion: Interdisciplinary
Perspectives , ed. S. Fussell , 16781. Mahwah, NJ : Lawrence
Erlbaum .
Mithen , Steven . 1996 . Te Prehistory of the Mind: A Search for the Origin
of Art, Science and Religion . London and New York : Tames and
Hudson .
. 1998 . A creative explosion? Teory of mind, language, and the
disembodied mind of the Upper Paleolithic. In Creativity in Human
Evolution and Prehistory , ed. S. Mithen, 16591. London and New
York : Routledge .
Sweetser , Eve . 1990 . From Etymology to Pragmatics . Cambridge and New
York : Cambridge University Press .
Yu , Ning . 1995 . Metaphorical expressions of anger and happiness in
English and Chinese . Metaphor and Symbolic Activity 10 : 59 92.
. 1998 . Te Contemporary Teory of Metaphor in Chinese: A
Perspective from Chinese . Amsterdam : John Benjamins .
Metaphor, Universals of Meter
496
elements. In Sanskrit and later Indian meters, morae are thus
counted, but some meters count morae while also controlling
syllables: In the gana -counting meters, syllables form (typically)
four-mora groups that are respected in composing the line.
Some song meters (e.g., Tongan, Ugandan) also use mora count-
ing as an organizing principle, but this may be a secondary efect
in song traditions where heavy syllables match two beats, light
syllables match one, and the number of beats is musically con-
trolled. Here, a better understanding of the independent metri-
cal status of text and tune is required.
In addition to controlling the length of the line, metrical rules
also often control a pattern based on putting the syllables into
two classes, one marked and the other unmarked. It is of par-
ticular interest that metrical rules diferentiate two kinds of syl-
lable but apparently never three or more kinds, even though this
greater diferentiation is phonetically possible in many lan-
guages. For example, in English metrical verse, the only strictly
controlled syllables are those that carry main stress in a poly-
syllable; other syllables, whether stressed or not, are not strictly
controlled, and this is why English metrical verse is rhythmically
fairly variable (see the discussion in music, language and ).
Tis means that as regards the strict regulation of syllable types
in English meters, the syllable carrying the main stress in a poly-
syllabic word is in one class, and all other syllables, whether
stressed or unstressed, are in the other class. Yet English distin-
guishes several degrees of stress in longer words, such as autobi-
ographical or onomatopoeic, and there is no question that in the
perception of the rhythm of the line, we perceive more than two
degrees of stress . In the quantitative meters of Greek, Sanskrit, or
Arabic, syllable placement depends on whether a syllable is light
or is heavy. In Vietnamese, there are phonologically six distinct
kinds of tone, but the six types of syllable are grouped into just
two tonal classes for metrical purposes. It is interesting in this
connection to consider alliterative meters, such as the meter of
Beowulf; in the normative line with four stressed syllables, the
third must alliterate with the frst and/or second but not with the
fourth. Here, stressed syllables are partitioned into two types
alliterating and not alliterating.
A patterned distribution based on two metrical classes of syl-
lable, such as the heavy and light syllables in Greek verse, is often
thought of as the basis of the rhythm of the line. A major way in
which theories of meter diverge is in their account of the relation
between rhythm and meter. For example, in English iambic pen-
tameter, there is a general tendency for odd-numbered syllables
to be unstressed and even-numbered syllables to be stressed,
but the actual pattern of stressed and unstressed syllables var-
ies constantly from line to line, thus, lines in the same meter can
vary in their rhythm. Some accounts of English meters attempt to
explain the full range of rhythmic variation by building statistical
tendencies into the metrical rules . In a diferent approach, Derek
Attridge (1982) incorporates rhythm fully into his account of
metrical verse, so that meter and rhythm are accounted for by a
single theory. In his account, the metrical template also includes
elements that match silences in the text (ofbeats), thus build-
ing temporal notions into the metrical theory. Tis and similar
accounts must cope with the fact that lines with the same met-
rical pattern can be realized with diferent rhythmic patterns,
and vice versa. If rhythm is not explained by the metrical rules,
substitution approach. In this approach, a meter such as iambic
pentameter is a template made by combining fve iambic feet
each of which is composed of a sequence of an unmarked sylla-
ble followed by a syllable that is marked. Te resulting template
is matched to a line whose syllables are unstressed or stressed,
so that stressed syllables occupy marked positions in the tem-
plate and unstressed syllables the unmarked positions. For
lines that are not fully periodic (e.g., in an iambic pentameter
line where the rhythm does not involve a uniform repetition of
unstressed-stressed throughout), the template itself is changed
by substituting a foot of a diferent kind (e.g., a spondee for an
iamb) to match the stress pattern of the variant part of the line.
Tis approach only describes the actual rhythm of the line, and
though it ofers a convenient vocabulary for the literary critic, it
tells us nothing about the organization of the meter of the line
or why some variations are possible in this meter and some not.
Most recent theoretical accounts express strong reservations or
total rejection of this approach.
Recent theoretical approaches to meter are based primarily
on linguistic theory, particularly on the theory of phonology ,
following the foundational work of Morris Halle and Samuel Jay
Keyser (1971). For metrical purposes, most such theories adopt
mechanisms that are used in the theory of phonology, particu-
larly the theory of word stress. Following Mark Libermans (1975)
insight that stress is a matter of the relation between syllables, not
a feature of syllables, diferent approaches explored the use of
trees and grids as representations in accounts both of word stress
and of metrical poetry (Kiparsky 1977; Hayes 1983). Te phono-
logical theory of optimality theory has also been adapted for
use in metrical verse (e.g., Golston and Riad 2005). Nigel Fabb
and Halle (2008) develop their account of poetic meter from a
formalism proposed for word stress by William Idsardi (1992); it
groups the syllables with the help of unpaired parentheses both
in phonology (word stress) and in lines (metrical verse). While in
most approaches the metrical representation is a template built
by special rules and then matched to the line, for Fabb and Halle
the metrical representation is generated from the line (much as
in generative syntax the syntactic representation is generated
from the terminal elements, such as words or morphemes ) .
In metrical verse, as noted, the length of the line is controlled.
In most cases, the basic unit of measurement is the syllable.
However, in many metrical traditions, some syllables are part of
the line but uncounted. In a common convention, a syllable end-
ing in a vowel precedes a syllable beginning in a vowel, but only
one of the two syllables is counted for metrical purposes (though
both are usually pronounced). Te latter fact shows something
important; it shows that the grouping and counting of syllables
for metrical purposes is not directly dependent on the phonology
of the lines. It also poses evident problems for other approaches,
such as that of Kristin Hanson and Paul Kiparsky (1996), which
attempt to account for variation in number of syllables by refer-
ring to the specifc phonology of the language.
In Japanese, some Indian meters, and some other metrical
traditions, morae are counted, a heavy syllable counting as two
morae and a light syllable as one mora. It is often argued that
the heavy syllable actually consists phonologically of two morae,
but this is not necessary for an explanation of the meter, which
can refer just to the syllable as projecting one or two metrical
Meter
497
mental states, this is impossible because Putnams thought refers
to H
2
O whereas his Twins thought refers to XYZ. Fodors conclu-
sion from this argument is that mental states are to be construed
narrowly, without reference to the external state of the world.
Only the so-called narrow content and structure of a belief deter-
mines behavior, not whether the belief is about H
2
O or XYZ or
even whether its true or not .
In addition to mental states being internal, Fodor argues,
mental processes have to be computational, that is, work on the
formal, syntactical properties of mental states, rather than on
their semantical properties, which are forbidden in an inter-
nalist account. Tis is called the formality condition .
Fodors main argument for MS is a negative one: Adhering
to its counterpart renders psychology practically impossible
because it assumes the availability of a full description of the
relevant aspects of the world in physical terms, such being
necessary to individuate mental states. Tat is, we would need
to have the physical description of water available to tell us
what a mental state containing water is about. However, such
physical descriptions are often unavailable. MS is diferent
from functionalism (as proposed by Putnam) in that it adds the
requirement that mental states are formal, symbolic entities on
which computational processes can work. Functionalism defnes
mental states to be determined by their functional role, that is,
their place in a causal network of other mental states, sensory
inputs, and behavior resulting from them. Functionalism sets
apart mental states from their physical substratum, whereas MS
divorces mental states from their causal antecedents in the world
and proposes that mental states are to be treated as syntactical
rather than semantical entities .
Ingmar Visser
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Fodor , Jerry A. 1980 . Methodological solipsism considered as a research
strategy in cognitive psychology . Behavioral and Brain Sciences
3 : 63 109.
Putnam , Hilary. 1975 a. Te meaning of meaning. In Mind, Language
and Reality: Philosophical Papers , II: 21571. Cambridge : Cambridge
University Press .
. 1975 b. Te nature of mental states. In Mind, Language and
Reality: Philosophical Papers , II: 42940. Cambridge : Cambridge
University Press .
Tuomela , Raima . 1989 . Methodological solipsism and explanation in
psychology . Philosophy of Science 56.1 : 23 47.
METHODOLOGY
Te topic of methodology most generally involves exploring the
range of responses to the following questions that any researcher
in the language sciences must answer: What sort of empirical
data are you collecting, how are you collecting it, and how do
you hope it will bear on the research question(s) you are trying
to answer? Addressing these questions for a particular special-
ization within the language sciences falls under the purview
of its experts. My goal here is a more general examination of
issues that arise when language scientists assess alternative data
sources and means of data collection and seek to interpret their
data. Tese high-level choices require more fnesse in language
several types of explanation are possible (and can be combined).
For example, Fabb (2002) argues that the perception of rhythmic
regularity involves pragmatic processes of pattern matching that
are distinct from metrical rules (which govern those aspects of
the line that are strictly controlled and, like other kinds of implicit
linguistic rules, are not directly perceived). It is also possible that
rhythmic patterns might be independently represented, per-
haps by grids similar to those found in metrical verse. Te link
between the metrical form and the rhythmic form of the verse
then may fall under a theory of text-to-tune matching .
Nigel Fabb and Morris Halle
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Attridge , Derek . 1982 . Te Rhythms of English Poetry . Harlow,
UK : Longman .
Fabb , Nigel . 2002 . Language and Literary Structure: Te Linguistic
Analysis of Form in Verse and Narrative . Cambridge : Cambridge
University Press .
Fabb , Nigel, and Morris Halle . 2008 . Meter in Poetry: A New Teory .
Cambridge : Cambridge University Press .
Golston , Chris, and Tomas Riad . 2005 . Te phonology of Greek lyric
meter . Journal of Linguistics 41 : 77 115.
Halle , Morris, and Samuel Jay Keyser . 1971 . English Stress: Its Form, Its
Growth and Its Role in Verse . New York : Harper and Row .
Hanson , Kristin, and Paul Kiparsky . 1996 . A parametric theory of poetic
meter . Language 72 : 287 335.
Hayes , Bruce . 1983 . A grid-based theory of English meter . Linguistic
Inquiry 14 : 357 94.
Idsardi , William . 1992 . Te computation of stress. Ph.D. diss.,
Massachusetts Institute of Technology.
Kiparsky , Paul . 1977 . Te rhythmic structure of English verse . Linguistic
Inquiry 8 : 189 247.
Liberman , Mark . 1975 . Te intonational system of English. Ph.D. diss.,
Massachusetts Institute of Technology.
METHODOLOGICAL SOLIPSISM
Methodological solipsism (MS) is the thesis that mental (or
psychological) states are to be individuated solely by referring
to their relationships with other mental states and the physical
state of someones brain, but not by referring to the physical
world outside the individual to whom those states are ascribed.
Tis phrase was coined by Hilary Putnam (1975a) in an essay
about meaning externalism and internalism , but its
main advocate (Putnam opposes the thesis) is Jerry Fodor , who
defends it as part of the representational and computational the-
ory of mind (Fodor 1980).
Te main issue that MS is concerned with is the relation-
ship between mental states and the outside world or, rather, the
absence of such relationships so far as explanation in scientifc
psychology is concerned. In Putnams (1975a) Twin Earth
thought experiment, the question is whether Putnam and his
Twin are in the same mental state when thinking about water,
given that on Earth water is H
2
O, whereas on Twin Earth water
has the chemical formula XYZ (though it behaves otherwise iden-
tically to water on Earth). If mental states are to explain behavior,
Fodor argues, it needs to be the case that Putnam and his Twin
are in the same mental state when they are thinking I would like
to take a dive into the deep waters. In an externalist account of
Methodological Solipsism Methodology
498
an otherwise comparable population is one of the major chal-
lenges of research. For example, if you want to study specific
language impairment (SLI), you presumably want to com-
pare children with SLI to unafected children, but which ones?
If you use children of the same chronological age, you will surely
fnd many diferences in their speech, but this will mainly con-
frm that a speech pathologist was correct in diagnosing the frst
group with SLI. More interesting would be to fnd that younger
unafected children whose language is similar to that of the
children with SLI in some respects (e.g., mean length of utter-
ance) nonetheless is more advanced in others (e.g. correct use of
infectional morphology ).
Tere are numerous ways of classifying speakers into groups
that have seemed fruitful: age (see aging and language ),
gender (see gender and language ), handedness, education,
native versus non-native speaker, mono- versus bi-/multilingual
(see bilingualism and multilingualism ), socioeconomic
status (see sociolinguistics ), and many more. Interpreting
any correlations one fnds between one of these variables and
some language phenomenon is rarely straightforward, how-
ever. For example, if we fnd that increasing age correlates with
increasing frequency of tip-of-the-tongue states, does that impli-
cate a general decline in memory retrieval with age, or rather an
ability to partially retrieve words that younger people could not
retrieve at all, thanks to greater exposure to these words over the
course of a lifetime (Gollan and Brown 2006)?
RELEVANCE OF ATYPICAL SPEAKERS TO THE STUDY OF TYPICAL
SPEAKERS. A second reason for studying a particular population
is to allow us to learn things about typical language that typical
speakers do not. For example, it has been suggested that certain
language disorders represent an otherwise intact language sys-
tem from which one circumscribed grammatical mechanism has
been removed or rendered inoperative, as in Yosef Grodzinskys
1986 account of agrammatic aphasia, according to which
traces of movement are missing from otherwise normal syn-
tactic representations. No unafected speakers would provide us
with the opportunity to pose the question What does syntax
look like if you take out just the traces? Similarly, SLI in later
life could, on certain views, allow us to ask how an incompletely
developed morphosyntax behaves when coupled with adult-
sized open-class vocabulary and general cognitive capacities,
such as working memory (see working memory and lan-
guage processing ) permitting us to test, for example, the
behavior of complex sentences in such circumstances. Te
congnitive immaturity of (typically developing) children pre-
cludes this kind of test.
Tere is a major caveat when dealing with atypical popula-
tions, however particularly when analyzing them using theo-
ries based on typical populations or drawing conclusions about
such populations. We do not know how circumscribed their
deviation from the norm really is. For example, in the case of
focal brain damage, it is not an innocent assumption to posit that
the speakers subsequent use of language is simply the output of
an otherwise normal brain (as it was before the lesion occurred)
minus whatever function(s) used to be performed by the lost
neural structures. Rather, the speakers language use is the prod-
uct of a damaged brain that has recovered from injury in ways
sciences than in others, perhaps, because (for the purposes of
most researchers) humans are the only creatures that display the
target phenomenon, namely, language. Tis creates ubiquitous
challenges owing to the great complexity of the human organ-
ism (as compared, say, to a fruit fy) and to ethical considerations
that prevent us from carrying out potentially informative proce-
dures that are used with other species.
I structure the discussion around a taxonomy of the ways
in which data can be collected by language scientists. Practical
exigencies limit the discussion to data about (human) language,
though many language scientists need to gather other sorts of
data as well (e.g., computational linguists collect simulation
data, anthropological linguists collect cultural data, dialectolo-
gists collect geographical data, etc.). I further narrow the focus
to data collected for research purposes, leaving aside issues par-
ticular to clinical (see speech-language pathology ), foren-
sic (see forensic linguistics ), and other applications. I also
omit discussion of instrumental, statistical, or other formal treat-
ment of data: Tis is important, but ultimately futile if ones data
do not properly address the questions to be answered.
My taxonomic framework divides empirical methods along
two dimensions. One dimension along which data collection
can be characterized is by the population of speakers/ hearers
(hereafter simply speakers , by which I mean also users of sign
languages ) from whom one is collecting data: adult native
speakers of a particular language, children growing up in bilin-
gual households, a creole community, the last surviving
speaker of a dying language (see extinction of languages ),
the unknown author(s) of an ancient text, the editorial board of a
dictionary, and so on. A second dimension, more or less orthog-
onal to the frst in principle, is how the language data get from
the speaker(s) to the researcher(s). Researchers may observe the
speaker while he or she is doing something involving language,
or they may gather data produced as the result of a prior event
involving language, via an artifact or another person.
As illustrations, I use hypothetical fndings that should not
be taken as statements of fact. Because their importance here
lies in clarifying conceptual points, I have not restricted myself
to attested uncontroversial results, though I believe these hypo-
thetical results not to be wildly implausible.
Different Populations
INTRINSIC INTEREST IN SUBGROUPS. Obviously, if ones research
questions are about a particular population (e.g., the language
of autistic children [see autism and language ], speakers of
tone languages), then this is an excellent reason for collecting
data from that population, but it is not the only possible reason,
as will be discussed presently. Research on any group other than
the default (healthy adult native speakers) is virtually always
comparative, if only implicitly: In order to know whether one is
really discovering properties of population X, rather than simply
heretofore unknown properties of human language in general,
X must be compared with population Y with regard to the same
properties. For example, fnding that some group of bilinguals
has an average vocabulary size of n in their dominant language
would be most interesting in the context of knowing that an
otherwise-comparable group of monolinguals has an average
vocabulary size of, say, 1.75 n . Determining what constitutes
Methodology
499
(ERP), magnetoencephalography (MEG). Useful results have
been obtained on some measures without giving subjects any
task at all, simply by exposing them to language auditorily, but
depending on the technique, mental tasks or even ones involving
responses such as button pressing are possible. A serious meth-
odological issue arises when we want to interpret the resulting
data, however. For example, consider eye-tracking data from a
sentence reading task. Te assumption has usually been that
the longer a reader spends looking at a particular word or group
of words, the harder they found those words to process or under-
stand. While that is true in many cases, one situation in which
people may spend a very short time looking at some word is
when it signals the need to reanalyze an earlier part of the sen-
tence and triggers an immediate regressive eye movement. Tis
clearly should not be taken to indicate ease of processing. Due
to challenges of this sort, there are now a half dozen or more
measures of fxation times commonly reported in eye-tracking
studies, but their interpretation is not agreed upon and may
well depend on the particulars of what is being read. Here, and
especially for brain measures, as data become richer they do not
necessarily become more informative until foundational results
establish how the basic response features are to be interpreted.
It is sometimes thought that we do not actually need to
understand these detailed properties of brain activity in order to
make productive use of these measures: So long as we can show
that stimulus Y patterns like stimulus X while stimulus Z pat-
terns diferently, then we have evidence that whatever manipu-
lation we used in creating the stimuli classifes X with Y to the
exclusion of Z. For example, someone might try to ask whether
binding theory (see also anaphora ) is really part of syntax or
part of semantics by creating a sentence that violates a clearly
syntactic principle (X), one that violates a clearly semantic prin-
ciple (Z), and a binding violation (Y), and then seeing whether
Y patterns like X or like Z in ERPs (or neither, in which case
no conclusion can be drawn). But it is impossible to construct
sentences that are identical in all respects ( phonology, mor-
phology, sequence of word classes , etc.) except for these vio-
lations, and so if we know nothing about what the observed brain
patterns actually mean, all that this kind of experiment can tell
us is that some property shared by X and Y is lighting up, and Z
does not share that property. (Although ERP researchers speak
of components sensitive to syntactic violations versus semantic
anomaly, the basis for this is a very small range of sentence types,
and semantic really refers to real-world implausibility, not vio-
lations of principles of formal semantics .)
Turning now to conscious reaction tasks, the most common
of course involve psychologys favorite technique, measuring
reaction time to press a button, say a word, and so on. Within
certain schools of linguistics, grammaticality judgments
are the most favored (increasingly encompassed by the broader
term well-formedness ratings , as they are also applied to individ-
ual words and are elicited on multipoint or open-ended scales).
In mentioning these two types of data collection side by side, my
intent is to emphasize their similarities. Tey both involve col-
lecting behavioral measures in immediate response to some lin-
guistic stimulus. Grammaticality judgments can be recorded and
timed by computer. Contrariwise, experiments normally carried
out by computer can be done interview style, for example, with
that we do not yet know how to ascertain. Some functions previ-
ously performed in the damaged area may have been taken over
by intact areas, which may in turn have lost some of their original
functionality; areas that were inhibited by the damaged area may
now be free to come into play; and so forth.
Another kind of atypical population includes the expert users
of language: authors, comedians, songwriters, journalists, poets,
playwrights, preachers, politicians, and so on. Tey can be taken
as proof by example of what it is possible for humans to do with
language, but beyond that we know very little about how they
come by their expertise, and so it is hard to say how they might
inform the study of language in nonexpert speakers.
A special reason for choosing particular speakers to study is
because of their genetic relationship to other speakers. Tis is
most obvious for language disorders suspected to have a herita-
ble component, such as SLI. But genetic relationships, in partic-
ular between twins, can be used in language sciences (as in many
sciences) to approach issues concerning the possible contribu-
tions of the genotype to aspects of language in the phenotype.
Te standard methodology is to compare monozygotic to dizy-
gotic twin pairs, whereby the former share (on average) twice as
much genetic material. Any phenomenon of language where the
monozygotic pairs are more similar is taken to be shaped more
heavily by prewired brain structures (cf. Ganger 1998) .
Types of Data Collection: Immediate Versus Delayed
By immediate collection of language-relevant data I mean that
researchers obtain data from a speaker while he or she is engaged
in some language-related behavior (though this may not involve
any action or even any awareness of language on the speakers
part). By delayed data collection I mean that researchers
obtain language-related data after the fact, including by studying
artifactual records of previous language-related behaviors (e.g.
transcriptions, recordings, corpora [see corpus linguistics ],
grammars, etc.). Tis distinction is crucial because it bears on
how much researchers can know about the original event. In
what follows I exemplify numerous approaches in each category
and suggest advantages and disadvantages.
KINDS OF IMMEDIATE DATA COLLECTION. Tere are really only two
sorts of data one can collect from speakers who are doing a lin-
guistic task (meant broadly, as shorthand for doing something
that involves language). One is to collect data on what they are
(deliberately) doing: for example, if they are talking, what they are
saying; if they are listening, when they are nodding, when they are
smiling, and so on. Te other is to collect some other physical mea-
sure that will (one hopes) provide evidence about language inside
them. Tis can take the form of voluntary behavioral measures or
involuntary physiological or brain measures. (Te data may be pre-
served for later analysis, e.g., on videotape. What is crucial in count-
ing it as immediate data is that it captures the speakers immediate
response. Even a questionnaire can fall into this category if speakers
report their immediate reactions, e.g., Yes/No or numeric ratings.)
In the category of involuntary responses, we fnd such tech-
niques as measuring galvanic skin response, pupil diameter,
and eye movements, plus indicators of brain activity from neu-
roimaging positron emission tomography (PET), functional
magnetic resonance imagining (fMRI), event-related potential
Methodology
500
member of an isolated society that does not welcome outsiders
may report that its leader uses special vocabulary; if research-
ers are not members of this society, they must take the word of
someone else .
Evidently there are some situations in which the use of delayed
language data is unavoidable, for example, when studying dead
languages or when speakers are not accessible. Also, quantita-
tive measures such as word frequency could not practically be
calculated entirely from immediate interactions with individual
speakers. More generally, the use of delayed data afords us a
much larger sample of language material, hence, potential expo-
sure to rare phenomena that we might otherwise never become
aware of. However, it is misguided to think that the availability
of billions of words of computer-searchable text has eliminated
the need for explicit data-gathering tasks: For the vast majority of
the worlds languages, the quantity of existing written materials
(if there is a writing system at all) is many orders of magnitude
smaller than for the languages that dominate the information
age, and much of it is not on computer .
Conclusion
Part of what makes the study of language both fascinating and
frustrating is that language can never truly be studied in isola-
tion: It inexorably traces back to the bodies and brains of human
beings, which both are always doing myriad things. Rather than
trying to ignore this as an inconvenience, researchers would do
well to keep it in mind whenever they have methodological deci-
sions to make. Sometimes, as with twin studies and certain lan-
guage disorders, it can even be turned into an advantage .
Carson T. Schtze
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Botha , Rudolph P . 1981 . Te Conduct of Linguistic Inquiry: A Systematic
Introduction to the Methodology of Generative Grammar . Te
Hague : Mouton .
Cowart , Wayne . 1997 . Experimental Syntax: Applying Objective Methods
to Sentence Judgments . Tousand Oaks, CA : Sage .
Ganger , Jennifer B . 1998 . Genes and environment in language acquisi-
tion: A study of early vocabulary and syntactic development in twins.
Ph.D. diss., Massachusetts Institute of Technology.
Gollan , Tamar H., and Alan S. Brown . 2006 . From tip-of-the-tongue
(TOT) data to theoretical implications in two steps: When more TOTs
means better retrieval . Journal of Experimental Psychology: General
135 : 462 83.
Grodzinsky , Yosef . 1986 . Language defcits and the theory of syntax .
Brain and Language 27 : 135 59.
Labov , William . 1972 . Some principles of linguistic methodology .
Language in Society 1 : 97 120.
Matthewson , Lisa . 2004 . On the methodology of semantic feldwork .
International Journal of American Linguistics 70 : 369 415.
Newmeyer , Frederick J . 1983 . Grammatical Teory, Its Limits and Its
Possibilities . Chicago : University of Chicago Press .
Resnik , Philip, Aaron Elkiss, Ellen Lau, and Heather Taylor. 2006. Te
Web in Teoretical Linguistics Research: Two Case Studies Using the
Linguists Search Engine. In Proceedings of the 31st Annual Meeting of
the Berkeley Linguistics Society , 26576.
Schtze , Carson T . 1996 . Te Empirical Base of Linguistics: Grammaticality
Judgments and Linguistic Methodology . Chicago : University of Chicago
Press .
people who cannot read. Interviews lose fne-grained timing
information (experimenters should still note gross diferences
in response times) but gain elsewhere, including the allowance
for open-ended narrative responses and the possibility of asking
follow-up questions contingent thereon. It is important to note
that the presence/absence of a laboratory setting, electronic
equipment, statistical analysis, and so on has no bearing on the
conceptual/epistemological nature of the data collected .
Finally in this category are data from speakers who are actu-
ally using language with no extra task imposed on them. It is
surprising prima facie how little time most language scientists
spend actually observing just these events. Te reason is largely
practical. Most research necessarily concentrates on a quite spe-
cifc aspect of some linguistic phenomenon; waiting for it to arise
by chance is too resource intensive. Nonetheless, it is important
to keep in mind that every step away from the real situations we
are interested in will introduce both random errors and system-
atic distortions for example, in the case of transcripts, due to
imperfect recording quality and the transcribers subconscious
assumptions about what is being said, respectively.
KINDS OF DELAYED DATA COLLECTION. I classify as a delayed data
situation one in which the object of the researchers measure-
ments, observations, and so on is not a person at a time when he
or she is engaging with language, but rather some indication of
what may have happened at such a time: an artifact produced by
that person, or a behavior observed by someone other than the
researcher. Tere are two major subclasses of such data.
One subtype comprises any instances of written language,
whether created by an original act of writing or representing an
attempt to transcribe or otherwise keep record of language that
was originally spoken. (Although this distinction is important,
all written material, including phonetic transcription, loses
much information found in spoken language.) Tis includes
documents from now-dead languages, dictionaries and gram-
mars, dialect atlases, poetry, song lyrics, scripts, and so on
(in some of which the writers intent may be to sound unlike
his own or anyone elses natural speech or prose writing), as
well as corpora amassed specifcally for academic purposes.
Any text found on the World Wide Web falls into this category
as well. One can, of course, treat textual material as an object of
study unto itself, ignoring how it was created, but if one wants to
use it as evidence bearing on human language in general, then
considering the many diferences between writing and talking
becomes paramount. Most signifcant is the ability to edit writ-
ten material after initially producing it (in most situations). Te
Web, increasingly used as a corpus because of its size, comes
with many special problems: It can be hard to ascertain who
actually wrote any given passage; it is usually impossible to
establish the native language(s), gender, age, and so on of the
author; the intended meaning and discourse function is often
unclear; and so forth.
Te second subtype of delayed data is hearsay, that is, reports
given by someone about language phenomena witnessed, told
about, or engaged in personally. For example, elderly speak-
ers might report that their parents used to use some expression
but that they themselves never used it. Tis is information that
researchers have no way of independently verifying. Likewise, a
Methodology
501
Like lexical metonymies, grammatical metonymies operate
both on the synchronic and diachronic levels. Te coercive pro-
cess in metonymy is particularly striking in cases where gram-
matical meaning conficts with lexical meaning. For example,
stative predicates, such as the verb be , may be used in construc-
tions that normally require action predicates, such as impera-
tives. Tus, the slogan of the American news network CNN Be the
frst to know is interpreted as the efect of an intentional act to
be carried out by the hearer: Do something [viz. watch CNN] so
that, as a result, you are the frst to know. Te conceptual shift
at work here is based on the result for action metonymy. On
the diachronic level, metonymy plays a crucial role in gram-
maticalization processes. For example, the lexical item go (in
conjunction with the present progressive) in the phrase be going
to has grammaticalized into a future marker. Human motion is
typically directed toward a goal and, hence, is strongly associ-
ated with the intention of reaching the goal. Since the goal can
only be reached in the future, the intention to reach the goal may
stand for the future itself .
Looked at from a pragmatic point of view, metonymy can
be regarded as a matter of inferencing. We can distinguish the
following three types of metonymic inference: inferences about
a referential item ( referential metonym), inferences about a
predicate ( predicational metonymy), and inferences about the
speech-act meaning ( illocutionary metonymy) (Panther
and Tornburg 1998). Referential metonymy is a means of indi-
rect reference. For example, the use of subway in Te subway is on
strike invites the inference that the subway personnel is meant .
Predicational metonymy is exemplifed by utterances such as
Te saxophone player had to leave early , which in many contexts
induces the metonymic inference that the saxophone player left
early. In this case, an obligation to leave is interpreted as an actu-
ally occurring action . Illocutionary metonymy is illustrated by
utterances such as Can you lend me ten dollars? Te speaker liter-
ally poses a question about the hearers ability to lend the speaker
$10, but this question gives rise to the metonymic inference that
the hearer is being asked to lend $10 to the speaker; it is under-
stood as a request. Conventional indirect requests like these are
not just random substitute forms for the direct request Lend me
ten dollars . Te literal meaning of the metonymic expression has
an important communicative function in this indirect request .
It addresses a potential obstacle: Te hearer might be unable
to carry out the requested action because he or she needs the
money, too (Gibbs 1994). In fact, the example illustrates an
important general point: Te literal meaning of a metonymy is
always relevant to the interpretation of metonymic expressions.
It thus provides strong evidence against the view that metonymy
is merely the substitution of one word for another .
Gnter Radden and Klaus-Uwe Panther
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Gibbs , Raymond W., Jr . 1994 . Te Poetics of Mind: Figurative Tought,
Language, and Understanding . Cambridge : Cambridge University
Press .
Koch , Peter . 1999 . Frame and contiguity: On the cognitive bases of
metonymy and certain types of word formation . In Panther and
Radden 1999, 139 67.
. 2005 . Tinking about what we are asking speakers to do. In
Linguistic Evidence: Empirical, Teoretical, and Computational
Perspectives, ed. Stephan Kepser and Marga Reis , 45784.
Berlin : Mouton de Gruyter .
METONYMY
Metonymy (Greek change of name) is one of the
major fgures of speech recognized in classical rhetoric . Te
Roman treatise Rhetorica ad Herennium defnes metonymy as
a trope that takes its expression from near and close things by
which we can comprehend a word that is not denominated by
its proper word . Tis ancient characterization already points to
two criterial notions of metonymy, contiguity and substitution ,
which still occur in most present-day defnitions of metonymy
as the substitution of one word for another with which it is
associated.
Recent studies in cognitive linguistics have shown that
metonymy is not just a matter of words and their substitution
but is part of human thinking and reasoning. Te conceptual
nature of metonymy has been demonstrated by George Lakof
(1987). For example, the term mother makes many people think
of a housewife mother. Te relationship between mothers and
housewives is metonymic and operates only on the conceptual
level: Te category mother is metonymically associated with the
subcategory housewife mother as one of its members.
Various cognitive linguists have described the conceptual
basis of metonymy using the notion conceptual frame . Frames
are packages of knowledge about coherent segments of experi-
ence. Te elements of a frame are conceptually contiguous: Any
element evokes the frame as a whole and, concomitantly, other
elements within the frame network. For example, the concept
author establishes a frame that includes literary works, a pub-
lisher, biographical information, etc. Since these elements are
conceptually contiguous, they may be exploited by metonymy.
Tus, we may metonymically refer to a book by naming its author,
as in We are reading Shakespeare . Typically, a metonymic inter-
pretation is coerced when there is a conceptual confict between
expressions belonging to the same frame. In the previous exam-
ple, the verb read requires an object that denotes a linguistically
coded content, such as a book or a letter. Te conceptual confict
is resolved by understanding Shakespeare as a reference point
that provides mental access to Shakespeares literary work
(Langacker 1993; Radden and Kvecses 1999) .
Studies in metonymy have traditionally focused on words.
Standard examples on the synchronic level include Te ket-
tle is boiling ( container for content) and Jonathan is in the
phone book ( person for name) . Metonymic processes on the
diachronic level have been long noted by historical lin-
guists and amply demonstrated since the nineteenth century.
Metonymic shifts have been observed cross-linguistically in a
number of conceptual frames (Koch 1999). For example, in the
marriage frame, a preparatory status of being engaged may
stand for the state of being married. Tus, the Latin word sponsus/
sponsa with the meaning fanc/fance shifted its meaning to
bride/bridegroom and ended up with the meaning husband/
wife, as in Spanish esposo/esposa , French poux/ pouse , and
English spouse .
Metonymy
502
inserted throughout the course of the syntactic derivation, via
generalized transformations, rather than all in one initial block.
Te derivation proceeds bottom up with the most deeply
embedded structural unit created, then combined, via merge ,
with the head of which it is the complement to create a larger
unit, and so on. Consider the derivation of Te woman will see
the man : Te noun (N) man is combined with the determiner (D)
the to form the determiner phrase (DP) the man . Tis DP then
combines with the verb see to produce an intermediate projec-
tion (in the sense of x-bar theory ), V-bar. Te DP the woman
is created in the same fashion as the man , and is combined with
the V-bar to produce the VP. Next, this VP merges with the Inf
will producing I-bar. Te DP the woman fnally moves (leaving
a TRACE t ) to the specifer position of I, yielding the full clausal
projection IP, schematically illustrated in (2) (by labeled bracket-
ing, a notational variant of tree representation):
(2) [
IP
Te woman [
I
will [
VP
t [
V
see [
DP
the man]]]]]
In this model, there is no one representation following all lexi-
cal insertion and preceding all singulary transformations. Tat
is, there is no D-structure .
Some Minimalist Goals
So far, S-structure persists: If there is a point where the deriva-
tion divides, branching toward LF on one path and toward PF on
the other, that point is S-structure. Te more signifcant question
is whether there are any crucial conditions defned on it as in
the GB framework, for example, with respect to binding theory
(Chomsky 1981 ). One goal of the minimalist research program is
to establish that these further properties are actually properties
of LF, as suggested in the mid-1980s (Chomsky 1986 ), contrary to
previous arguments (Chomsky 1981 ).
Another goal is to reduce all constraints on representation to
bare output conditions , determined by the properties of the men-
tal systems that LF and PF must interface with. For instance, the
motor system determines that a phonetic representation must
be linearly ordered.
Internal to the computational system, the desideratum is
that constraints on transformational derivations be reduced to
general principles of economy. Derivations beginning from the
same choice of lexical items are compared in terms of number
of steps, length of movements, and so on, with the less economi-
cal ones being rejected. An example is the minimalist deduction
of the superiority condition , which demands that when multiple
items are available for wh-movement in a language, like English,
allowing only one to move, it is the highest one (one closest to
the root of the phrase structure tree) that will move:
(3) Who t will read what
(4) *What will who read t [* indicates ungrammaticality]
Economy, in the form of shortest move , selects (3) over (4) since
the subject is higher than the object, hence, closer to the sen-
tence initial target of wh-movement than the object is .
Te simplifying developments in the theory leading toward
the minimalist approach generally led to greater breadth and
depth of understanding of both how human languages are orga-
nized (descriptive adequacy) and how they develop in childrens
Lakof , George . 1987 . Women, Fire, and Dangerous Tings: What Categories
Reveal about the Mind . Chicago : University of Chicago Press .
Lakof , George, and Mark Johnson . 1980 . Metaphors We Live By. Chicago
and London : University of Chicago Press .
Langacker , Ronald . 1993 . Reference-point constructions . Cognitive
Linguistics 4 : 1 38.
Panther , Klaus-Uwe, and Gnter Radden , eds. 1999. Metonymy in
Language and Tought . Amsterdam and Philadelphia : Benjamins.
Panther , Klaus-Uwe, and Linda L. Tornburg . 1998 . A cognitive approach
to inferencing in conversation . Journal of Pragmatics 30 : 755 69.
Radden , Gnter, and Zoltn Kvecses . 1999 . Towards a theory of meton-
ymy . In Panther and Radden 1999, 17 59.
MINIMALISM
Minimalism, extending earlier work in transformational
grammar and generative grammar , conjectures that the
computational system central to human language is a per-
fect solution to the task of relating sound and meaning. Recent
research has investigated the complexities evident in earlier
models and attempted to eliminate them, or to show how they
are only apparent, following from deeper and simpler proper-
ties. Major examples of this work include the reduction of the
number of linguistic levels of representation in the model and
the deduction of certain constraints on syntactic derivations
from general considerations of economy and computational
simplicity .
Like earlier versions of generative grammar, the minimalist
program (MP) (Chomsky 1995b , 2000 , 2004 , 2005 ) maintains that
linguistic competence is a computational system creating and
manipulating structural representations. MP further proposes
that the derivations and representations conform to economy
criteria, demanding that they be minimal in a sense determined
by the language faculty (perhaps ultimately by general proper-
ties of organic systems): no extra steps in derivations, no extra
symbols in representations, and no representations beyond
those that are conceptually necessary.
Reduction of Levels
Minimalism developed out of the government and binding
(GB) or principles and parameters model (Chomsky 1981 ,
1982 ; Chomsky and Lasnik 1993 ). In that model, there are four
signifcant levels of representation, related by derivation:
(1)
Given that a human language is a way of relating sound (or,
more generally, gesture, as in sign languages) and meaning,
the interface levels PF and LF were assumed to be ineliminable.
Minimalism begins with the hypothesis that there are no other
levels .
Structure Building
Minimalism, in a partial return to the apparatus of pre-1965
transformational theory (Chomsky 1955 ), has lexical items
D(eep)-Structure
S(urface)-Structure
PF
(Phonetic Form)
LF
(Logical Form)
Minimalism
503
as a result of the movement, but that is simply a benefcial side
efect of the satisfaction of the requirement of the attractor. Te
earlier minimalist approach to the driving force of movement
was called Greed by Chomsky. Tis later one developed out of
what Howard Lasnik ( 1995 ) called Enlightened Self Interest.
The Syntactic Similarity of Languages
One recurrent theme in GB and minimalist theorizing, motivated
by the quest for explanatory adequacy, is that human languages
are syntactically very similar. Te standard GB and early mini-
malist instantiation of this claim was the proposal that superfcial
diferences result from potential derivational timing diferences
among languages, with the same transformation applying in
overt or covert syntax. Under both circumstances, LF refects the
results of the transformation. For example, the wh-movement
operative in English interrogative sentences is overt movement
to specifer of C(omplementizer). In many other languages,
including Chinese and Japanese, interrogative expressions seem
to remain in situ, unmoved, as seen in the contrast between (9)
and its English translation in (10):
(9) ni xihuan shei [Chinese]
you like who
(10) Who do you like
C.-T. Huang ( 1981 /1982) argued that even in such languages
there is movement, by showing that well-established locality
constraints on wh-movement, such as those of John Robert Ross
( 1967 ), also constrain the distribution and interpretation of cer-
tain seemingly unmoved wh-expressions in Chinese. Tis argu-
ment was widely infuential and laid the groundwork for much
GB and minimalist research.
Along related lines, Chomsky argued that V-raising, overt in
virtually all of the Romance languages, among others, operates
covertly in English, as in the following examples from English
and their translations into French:
(11) a. John often kisses Mary
b. *John kisses often Mary
(12) a. *Jean souvent embrasse Marie
b. Jean embrasse souvent Marie
Te assumption is that the position of the verb vis--vis the adverb
indicates whether the verb has raised overtly. For V-raising, the
feature driving the movement is claimed to be one that resides
in Inf. Te feature might be strong , forcing overt movement (as
in French), or weak. Similarly, the feature demanding overt wh-
movement in English is a strong feature of C. Te principle pro-
crastinate disallows overt movement except when it is necessary
(i.e., for the satisfaction of a strong feature as in Chomsky 1993 ;
Lasnik 1999 a).
Procrastinate invited a question. Why is delaying an operation
until LF more economical than performing it earlier? Further,
many of the hypothesized instances of covert movement do
not have the semantic efects (with respect to quantifer scope,
anaphora, etc.) that corresponding overt movements have, as
discussed by Lasnik ( 1999 b, Chapters 6 and 8). To address these
questions, Chomsky ( 2000 ; 2001 ) argues for a process of agree-
ment (potentially at a substantial distance) that relates the two
minds (explanatory adequacy) (see descriptive, obser-
vational, and explanatory adequacy ). Tis success
led Noam Chomsky to put forward the audaciously minimal-
ist conjecture that we are now in a position to go even beyond
explanatory adequacy: Te human language faculty might be
a computationally perfect solution to the problem of relating
sound and meaning, the minimal computational system given
the boundary conditions provided by other modules of the mind.
Tis conjecture leads to a general minimalist critique of syntactic
theorizing, including Chomskys own earlier minimalist theo-
rizing. Consider frst the leading idea that multiple derivations
from the same initial set of lexical choices are compared. Tis
introduces considerable complexity into the computation, espe-
cially as the number of alternative derivations multiplies. It thus
becomes desirable to develop a model whereby all relevant deri-
vational decisions can be made in strictly Markovian fashion: At
each step, the very next successful step can be determined, and
determined easily. Tis arguably more tractable local economy
model was suggested by Chomsky ( 1995a ) and developed in
detail by Chris Collins ( 1997 ) .
The Last Resort Nature of Syntactic Movement
From its inception in the early 1990s, minimalism has insisted
on the last resort nature of movement: Movement must hap-
pen for a formal reason. Te case flter (see filters ), which
was a central component of the GB system, was thought to pro-
vide one such driving force. A standard example involves subject
raising :
(5) John is certain [ t to fail the exam]
(6) It is certain [that John will fail the exam]
In (5), as in (6), John is the understood subject of fail the exam .
Tis fact is captured by deriving (5) from an underlying structure
much like that of (6), except with an infnitival embedded sen-
tence instead of a fnite one:
(7) __ is certain [John to fail the exam]
John in (7) is not in a position appropriate to any case. By raising
to the higher subject position (specifer of the higher Inf), it can
avoid a violation of the case flter, since the raised position is one
where nominative case is licensed. But if the case requirement of
John provides the driving force for movement, the requirement
will not be satisfed immediately upon the introduction of that
nominal expression into the structure, under the assumed bot-
tom-up derivation. Rather, satisfaction must wait until the next
cycle, when a higher layer of structure is built, or, in fact, until an
unlimited number of cycles later, as raising confgurations can
iterate:
(8) John seems [ to be certain [ to fail the exam]]
A minimalist perspective favors an alternative whereby
the driving force for movement can be satisfed immediately.
Suppose that Inf has a feature that must be checked against
the NP. Ten as soon as that head has been introduced into the
structure, it attracts the NP or DP that will check its feature.
Movement is then seen from the point of view of the target, rather
than the moving item itself. Te case of the NP does get checked
Minimalism
504
representational theory. As the derivation proceeds, always
merging together pairs of items, sisterhood and domination are
the only immediately available primitives. And X (asymetrically)
c-commands Y if and only if Y is dominated by the sister of X.
Tese notions are illustrated in (13), where B and C are sisters as
are D and E, A dominates B, C, D, and E, and C dominates D and
E. B asymmetrically c-commands D and E.
(13)
Multiple Spell-Out efectively deals with a range of reconstruc-
tion phenomena. For example, an anaphor normally requires an
antecedent that c-commands it:
(14) John criticized himself
(15) *Himself criticized John
But when the anaphor is fronted from a position c-commanded
by an antecedent to a position not in that structural relation, the
anaphoric connection is nonetheless possible:
(16) Himself, John criticized
Tis follows straightforwardly if anaphora can be interpreted
prior to movement .
Chomsky has also explored another kind of approach to
reconstruction, based on a condition that he calls Inclusiveness
Chomsky ( 1995a ). Tis condition demands that a syntactic deri-
vation merely combine elements of the numeration. No new
entities can be created. Traces, as traditionally conceived, vio-
late this condition. Chomsky therefore concludes that a trace of
movement is actually a copy of the item that moved, rather than
a new sort of entity. Tis is yet another return to earlier genera-
tive approaches (wherein movement was seen as a compound of
copying and deletion). Te copy left behind is normally deleted
in the phonological component (though Boskovic 2001 pres-
ents arguments that under certain circumstances, lower copies
are pronounced in order to rescue what would otherwise be
PF violations) but could persist for semantic purposes, such as
the licensing of anaphoric connection. Danny Fox ( 2000 ) pres-
ents an analysis of scope and anaphora reconstruction efects in
terms of the copy theory .
An infuential research line, initiated by Richard S. Kayne
( 1994 ), extends the impact of c-command to PF as well. Kayne
hypothesizes that the linear order that is manifest in PF (as it must
be, given properties of the phonetic system) comes about via his
linear correspondence axiom (LCA), which states that asym-
metric c-command is mapped onto PF linear order. Tis has the
far-reaching consequence that structures must always be right-
branching. Subject-verb-object (SVO) languages like English are
broadly consistent with this requirement, but subject-object-verb
(SOV) languages like Japanese are not. Kaynes antisymmetry
approach reanalyzes SOV languages as underlyingly SVO (as all
languages must be by this hypothesis), with the SOV order derived
by (leftward) movement. One crucial unanswered question is the
source of the driving force for the required movements .
items that need to be checked against each other. Many of the
phenomena that had been analyzed as involving covert move-
ment are reanalyzed as involving no movement at all, just the
operation Agree (though Huangs argument indicates that there
are at least some instances of covert movement). Overt phrasal
movement (such as subject raising) is then seen in a diferent
light: It is not driven by the need for case or agreement features
to be checked (since that could take place via Agree). Instead, it
takes place to satisfy the requirement of certain heads (includ-
ing Inf) that they have a specifer (in the X-bar theoretic sense).
Such a requirement was already formulated by Chomsky ( 1981 ),
and dubbed the extended projection principle (EPP) in Chomsky
( 1982 ). To the extent that long distance A-movement (basically,
movement to a higher subject position) as in (8) proceeds suc-
cessive-cyclically through each intermediate subject position,
the EPP is motivated, since, as observed earlier, these intermedi-
ate positions are not case-checking positions .
An important question at this point is why language has the
seeming imperfection of movement processes at all. We can
distinguish two major types of movement, phrasal movement
and head movement. Chomsky conjectures that phrasal move-
ment is largely to convey topic-comment information (and pos-
sibly to make scope relations more transparent), and that the
EPP is the way the computational system formally implements
this. V-movement, on the other hand, is conjectured to have
PF motivation (guaranteeing that the Inf af x will ultimately
be attached to a proper host, V), and may even be a PF process.
Another possibility is that movement is simply a generalization
of the merge operation combining smaller structures into larger
ones. Given that merge is ineliminable, perhaps move is not an
imperfection after all .
Syntactic Interfaces
Te connection between syntactic derivation and semantic and
phonological interfaces has long been a central research area. In
minimalism, interpretation could be distributed over many struc-
tures in the course of transformational cycles . Already decades
ago, Joan W. Bresnan ( 1971 ) argued that the rule responsible for
assigning English sentences their intonation contour applies
following each cycle of transformations, rather than at the end
of the syntactic derivation. Ray Jackendof ( 1972 ) put forward
similar proposals for semantic phenomena involving scope and
anaphora. Chomsky ( 2000 , 2001 ) argues for a general instanti-
ation of this distributed approach, sometimes called Multiple
Spell-Out, based on Epstein ( 1999 ) and Uriagereka ( 1999 ).
At the end of each cycle (or phase in Chomskys more recent
work), the syntactic structure thus far created can be encapsu-
lated and sent of to the interface components for phonological
and semantic interpretation. Tus, even the levels of PF and LF
fade away. Samuel D. Epstein argues that such a move represents
a conceptual simplifcation (in the same way that elimination of
D-structure and S-structure did), and both Juan Uriagereka and
Chomsky provide empirical justifcation. Te role of syntactic
derivation, always important in Chomskian theorizing, becomes
even more central on this view. Epstein reasons that the central-
ity of (asymmetric) c-command (as opposed to one of a whole
range of other conceivable geometric relations) in syntax is
predicted on this strongly derivational view, but not in a more
A
B C
D E
Minimalism
505
Kayne , Richard S. 1994 . Te Antisymmetry of Syntax . Cambridge, M A: MIT
Press .
Lasnik , Howard. 1995 . Case and expletives revisited: On Greed and
other human failings. Linguistic Inquiry 26 : 615 33. Repr. in Lasnik
1999b, 7496.
. 1999 a. On feature strength: Tree minimalist approaches to overt
movement. Linguistic Inquiry 30 : 197 217. Repr. Howard Lasnik,
Minimalist Investigations in Linguistic Teory ( London ; Routledge,
2003), 83102.
. 1999 b. Minimalist Analysis . Oxford : Blackwell .
Lasnik , Howard , and Juan Uriagereka , with Cedric Boeckx . 2005 . A Course
in Minimalist Syntax: Foundations and Prospects . Oxford : Blackwell .
Ross , John Robert . 1967 . Constraints on Variables in Syntax . Ph.D. diss.,
Massachusetts Institute of Technology. Published as Infnite Syntax!
( Norwood, N.J . : Ablex , 1986).
Uriagereka , Juan . 1998 . Rhyme and Reason: An Introduction to Minimalist
Syntax . Cambridge, M A: MIT Press .
. 1999 . Multiple spell-out. In Working Minimalism , ed. Samuel
David Epstein and Norbert Hornstein , 251 82. Cambridge, M A: MIT
Press .
MIRROR SYSTEMS, IMITATION, AND LANGUAGE
Any normal child reared in human society will acquire lan-
guage. Some argue that basic structures of grammar are innate ,
so that the child need simply hear a few sentences to set the
parameter for each key principle of the grammar of his or her
frst language (Baker 2001 ; Chomsky and Lasnik 1993 ). Others
have argued that the modern child receives rich language stimuli
within social interactions in learning these key principles. In any
case, the child must acquire the particular sounds ( phonol-
ogy ) of the language, an ever-increasing stock of words, and
constructions for arranging words to compound their meanings.
Te infant acquiring maternal phonology does not imitate the
caregiver (Y. Yoshikawa and colleagues [ 2003 ] model how the
process may use associative learning), but learning how to put
sounds together to form a word that achieves the childs commu-
nicative goal seems to involve imitation. Imitation also lies at the
heart of the acquisition of syntax and semantics (see syntax,
acquisition of; semantics, acquisition of ). Even within
the principles and parameters approach, the child must imitate
words and combinations, as well as set parameters, to come to
speak the language (see principles and parameters the-
ory and language acquisition ) .
Monkeys have little or no capacity for imitation, and apes
(chimpanzees, gorillas, bonobos, orangutans) have a capac-
ity for simple imitation, whereas humans are the only primates
capable of complex imitation. We describe these forms of imita-
tion, then argue that increasing imitative skills, and the relation
of mirror neurons to these imitative skills, were at the heart of the
evolution of the language-ready brain.
Simple and Complex Imitation
M. Myowa-Yamakoshi and T. Matsuzawa ( 1999 ) observed that
chimpanzees took 12 or so trials to learn to imitate a behavior in
a laboratory setting, focusing on bringing an object into relation-
ship with another object or the body, rather than the actual move-
ments involved. R. W. Byrne and J. M. E. Byrne ( 1993 ) found that
gorillas learn complex feeding strategies but may take months to
do so. Consider eating nettle leaves. Skilled gorillas grasp the stem
Conclusion
Chomsky constantly emphasizes that minimalism is as yet still
just an approach, a set of questions and a conjecture about
how human language works (perfectly), and a general program
for exploring the questions and developing the conjecture. Te
descriptive and explanatory success attained thus far gives some
reason for optimism that the approach can be developed into an
articulated theory of human linguistic ability and of why it has
the exact properties it does .
Howard Lasnik
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Boeckx , Cedric . 2006 . Linguistic Minimalism: Origins, Methods, Concepts,
and Aims . Oxford : Oxford University Press .
Boskovic , Zeljko . 2001 . On the Nature of the Syntax-Phonology
Interface: Cliticization and Related Phenomena . Amsterdam : Elsevier
Science .
Bresnan , Joan W. 1971 . Sentence stress and syntactic transformations .
Language 47 : 257 81.
Chomsky , Noam . 1955 . Te logical structure of linguistic theory.
Manuscript, Harvard University and Massachusetts Institute of
Technology. Revised 1956 version published in part by Plenum, New
York, 1975 and by University of Chicago Press, Chicago, 1985.
. 1981 . Lectures on Government and Binding . Dordrecht, the
Netherlands : Foris .
. 1982 . Some Concepts and Consequences of the Teory of Government
and Binding . Cambridge, M A: MIT Press .
. 1986 . Knowledge of Language . New York : Praeger .
. 1993 . A minimalist program for linguistic theory. In T he View
from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger ,
ed. Kenneth Hale and Samuel J. Keyser, 152. Cambridge, M A: MIT
Press. Repr. in Chomsky 1995b, 167217.
. 1995 a. Categories and transformations. In Te Minimalist
Program , 219394. Cambridge, M A: MIT Press.
. 1995 b. Te Minimalist Program . Cambridge, M A: MIT Press .
. 2000 . Minimalist Inquiries: the framework. In Step by Step: Essays
on Minimalist Syntax in Honor of Howard Lasnik , ed. Roger Martin ,
David Michaels , and Juan Uriagereka , 89155. Cambridge, M A: MIT
Press .
. 2001 . Derivation by phase. In Ken Hale: A Life in Language , ed.
Michael Kenstowicz , 152. Cambridge, M A: MIT Press .
. 2004 . Beyond explanatory adequacy. In Structures and Beyond
the Cartography of Syntactic Structure . Vol. 3. Ed. Adriana Belletti ,
10431. Oxford : Oxford University Press .
. 2005 . Tree factors in language design. Linguistic Inquiry
36 : 1 22.
Chomsky , Noam , and Howard Lasnik . 1993 . Te theory of principles and
parameters. In Syntax: An International Handbook of Contemporary
Research . Vol. 1. Ed. Joachim Jacobs , Arnim von Stechow , Wolfgang
Sternefeld , and Teo Vennemann , 50669. Berlin : Walter de Gruyter .
Reprinted in Chomsky 1995b, 13127.
Collins , Chris . 1997 . Local Economy . Cambridge, M A: MIT Press.
Epstein , Samuel D. 1999 . Un-principled syntax: Te derivation of syn-
tactic relations. In Working Minimalism , ed. Samuel D. Epstein and
Norbert Hornstein , 31745. Cambridge, M A: MIT Press .
Fox , Danny . 2000 . Economy and Semantic Interpretation . Cambridge,
M A: MIT Press .
Huang , C.-T. James . 1981 /1982. Move wh in a language without wh -
movement. Linguistic Review 1 : 369 416.
Jackendof , Ray . 1972 . Semantic Interpretation in Generative Grammar .
Cambridge, M A: MIT Press .
Minimalism Mirror Systems, Imitation, and Language
506
area, traditionally thought of as a speech area but which has been
shown by brain imaging studies (see neuroimaging ) to be active
when humans both execute and observe grasps. It is posited that
the mirror system for grasping was also present in the common
ancestor of humans and monkeys (perhaps 20 million years ago)
and that of humans and chimpanzees (perhaps 5 million years
ago). Moreover, the mirror neuron property accords well with the
parity requirement for language that what counts for the speaker
must count approximately the same for the hearer. In addition,
normal face-to-face speech involves manual and facial as well as
vocal gestures and, moreover, signed languages are fully devel-
oped human languages (see sign language ). Tese fndings
ground the Mirror System Hypothesis (Arbib and Rizzolatti 1997 ;
Rizzolatti and Arbib 1998 ):
Te parity requirement for language in humans is met because
Brocas area evolved atop the mirror system for grasping, which
provides the capacity to generate and recognize a set of actions.
Recent work (see Arbib 2005 for a review and commentaries
on current controversies) has elaborated the hypothesis, defn-
ing an evolutionary progression of seven stages, S1 through S7:
(S1) Cortical control of hand movements.
(S2) A mirror system for grasping, shared with the common
ancestor of human and monkey.
A mirror system does not provide imitation in itself. A monkey
with an action in its repertoire may have mirror neurons active
both when executing and observing that action. Te monkey
does not repeat the observed action nor, crucially, does it use
observation of a novel action to add that action to its repertoire.
Nonetheless, the mirror system may serve the monkey well both
in providing feedback during close observation of handobject
relations during dextrous actions and in allowing its recognition
of others actions to infuence social behavior. In any case, the
data on primate imitation support the hypothesis that a monkey-
like mirror system becomes embedded in more powerful sys-
tems in the next two stages of evolution.
(S3) A simple imitation system for grasping, shared with
common ancestor of human and apes.
(S4) A complex imitation system for grasping.
frmly, strip of leaves, remove petioles bimanually, fold leaves
over the thumb, pop the bundle into the mouth, and eat. Teaching
is virtually never observed in apes (Caro and Hauser 1992 ), and
the young seem to look at the food, not at the methods of acquisi-
tion (Corp and Byrne 2002 ). Moreover, chimpanzee mothers sel-
dom, if ever, correct and instruct their young (Tomasello 1999 ).
Te challenge for acquiring such skills is compounded because
the sequence of atomic actions varies greatly from trial to trial.
Byrne ( 2003 ) implicates imitation by behavior parsing , a pro-
tracted form of statistical learning whereby certain subgoals (e.g.,
nettles folded over the thumb) become evident from repeated
observation as being common to most performances. Apparently,
the young ape, over many months, may acquire the skill by com-
ing to recognize the relevant subgoals and derive action strategies
for achieving subgoals by trial and error.
Te ability to learn the overall structure of a specifc feeding
behavior over many, many observations, however, is very diferent
from the human ability to understand any sentence of an open-
ended set as it is heard and to generate another novel sentence
as an appropriate reply. In many cases, humans need just a few
trials to make sense of a relatively complex behavior and can then
repeat it under changing circumstances, if the constituent actions
are familiar and the subgoals these actions must achieve are read-
ily discernible. (Te next section places this facility for complex
imitation in an evolutionary and neurological perspective.) It is
interesting to note that even newborn infants can perform certain
acts of imitation, but this capacity for neonatal imitation such
as poking out the tongue on seeing an adult poke out a tongue
(Meltzof and Moore 1977 ) is quantitatively diferent from that
for complex imitation (see communication, prelinguistic ) .
The Mirror System Hypothesis
Te system of the macaque brain for visuomotor control of grasp-
ing has its premotor outpost in an area called F5 ( Figure 1 left),
which contains a set of neurons, mirror neurons , such that each
one is active not only when the monkey executes a specifc grasp
but also when the monkey observes a human or other monkey
execute a more-or-less similar grasp (Rizzolatti et al. 1996 ). Tus,
macaque F5 contains a mirror system for grasping that employs
a similar neural code for executed and observed manual actions.
Te homologous region of the human brain is in or near Brocas
Monkey [Not to scale] Human
Homology
Parietal Lobe
Temporal Lobe
Occipital
Lobe
Frontal
Lobe
Brocas
Area
Wernickes
Area
Figure 1. A comparative side view of the monkey brain (left) and human brain (right), not to scale. The left view
emphasizes premotor area F5; the right view emphasizes Brocas area and Wernickes area, considered crucial for
language processing. F5 and Brocas area are considered homologous.
Mirror Systems, Imitation, and Language
507
categories. Tis is supported by the observation that nouns are often
marked for case, number, gender (see gender marking ), size,
shape, defniteness, and possession, while verbs are often marked
for tense, aspect , mood , modality, transitivity, and agree-
ment . On the latter view, once protolanguage was established, dif-
ferent peoples developed (and later shared) diferent strategies for
talking about things and actions and then developed these strat-
egies in diverse ways to talk about more and more of their world.
Tis view is based on the fact that there are further aspects of lan-
guage diversity hard to reconcile with natural selection of brain
mechanisms. Some languages, like Vietnamese, lack all infection,
precluding the use of infectional criteria for identifying grammati-
cal categories; other languages employ infection in unusual ways.
For example, the language of the Makah of the northwestern coast
of the United States applies aspect and mood markers not only to
words for actions that are translated into English as verbs but also
to words for things and properties .
Complex imitation has two parts: i) the ability to perceive that
a novel action may be approximated by a composite of known
actions associated with appropriate subgoals, and ii) the abil-
ity to employ this perception to perform an approximation of
the observed action, which may then be refned through prac-
tice. Both parts come into play when the child is learning a lan-
guage; the former predominates in adult use of language as the
emphasis shifts from mastering novel words and constructions
to fnding the appropriate way to continue a dialogue .
Michael A. Arbib
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Arbib , M. A. 2005 . From monkey-like action recognition to human lan-
guage: An evolutionary framework for neurolinguistics (with com-
mentaries and authors response). Behavioral and Brain Sciences
28 : 105 67.
Arbib , M. A., K. Liebal , and S. Pika. 2008. Primate vocalization, ges-
ture, and the evolution of human language. Current Anthropology
59 .6: 105376.
Arbib , M. A. , and G. Rizzolatti . 1997 . Neural expectations: A possible
evolutionary path from manual skills to language . Communication
and Cognition 29 : 393 424.
Baker , M. 2001 . Te Atoms of Language: Te Minds Hidden Rules of
Grammar . New York : Basic Books .
Byrne , R. W. 2003 . Imitation as behavior parsing . Philosophical
Transactions of the Royal Society of London (B) 358 : 529 36.
Byrne , R. W. , and J . M. E. Byrne. 1993 . Complex leaf-gathering skills of
mountain gorillas ( Gorilla g. beringei ): Variability and standardiza-
tion . American Journal of Primatology 31 : 241 61.
Caro , T. M. , and M. D. Hauser . 1992 . Is there teaching in nonhuman ani-
mals? Quarterly Review of Biology 67 : 151 74.
Chomsky , N. , and H. Lasnik. 1993 . Te theory of principles and parame-
ters. In Syntax: An International Handbook of Contemporary Research ,
I: 50669. Berlin : de Gruyter .
Corp , N. , and R. W. Byrne . 2002 . Ontogeny of manual skill in wild chim-
panzees: Evidence from feeding on the fruit of saba forida. Behavior
139 : 137 68.
Kemmerer , D. 2005 . Against innate grammatical categories. Behavioral
and Brain Sciences 28 . Available online at: http://www.bbsonline.org/
Preprints/Arbib-0501 2002/Supplemental/ .
Meltzof , A. N. , and M. K. Moore . 1977 . Imitation of facial and manual
gestures by human neonates . Science 198 : 75 8.
Each of these changes can be of evolutionary advantage in
supporting the transfer of novel skills among the members
of a community, though involving praxis rather than explicit
communication.
M. A. Arbib, K. Liebal, and S. Pika (2008) summarize data
suggesting that manual gestures have greater openness than
vocalizations in nonhuman primates. Monkey vocalizations are
innately specifed (though occasions for using a call may change
with experience), whereas a group of apes may communicate
with novel gestures. M. Tomasello and colleagues ( 1997 ) argue
that novel gestures may develop through ontogenetic ritualiza-
tion , wherein repeated interaction between two individuals
establishes a conventionalized form of an action as a signal for
the action for example, a beckoning movement may become
recognized as short for the physical action of pulling the other
toward oneself . Tese gestures may then be propagated by social
learning. Tis supports the hypothesis that it was gesture rather
than primate vocalizations that created the opening for
greatly expanded gestural communication once complex imi-
tation had evolved for practical manual skills. R. M. Seyfarth,
D. L. Cheney, and T. J. Bergman ( 2005 ) advance the opposing
view, but the Mirror System Hypothesis postulates that evolution
proceeded via the next two stages:
(S5) Protosign , a manual-based communication system
breaking through the fxed repertoire of primate vocalizations
to yield an open repertoire .
(S6) Protolanguage as Protosign and Protospeech : an expand-
ing spiral of conventionalized manual, facial, and vocal com-
municative gestures .
Te transition from complex imitation and the small reper-
toires of ape gestures (perhaps 10 or so novel gestures shared
by a group) to protosign involves pantomime, frst of grasp-
ing and manual praxic actions, then of nonmanual actions
(e.g., fapping the arms to mime the wings of a fying bird).
Pantomime transcends the slow accretion of manual gestures
by ontogenetic ritualization, providing an open semantics
for a large set of novel meanings (Stokoe 2001 ). However, such
pantomime is inef cient both in the time taken to produce it
and in the likelihood of misunderstanding. Conventionalized
signs extend and exploit more ef ciently the semantic rich-
ness opened up by pantomime. Processes like ontogenetic
ritualization can convert elaborate pantomimes into a con-
ventionalized shorthand, just as they do for praxic actions.
Tis capability for protosign rather than elaborations intrin-
sic to the core vocalization systems may then have provided
the essential scafolding for protospeech and evolution of the
human language-ready brain .
(S7) Language : the development of syntax and composi-
tional semantics.
Te fnal stage the transition from protolanguage to lan-
guage may have rested primarily on biological evolution (Pinker
and Bloom 1990 ), but may instead result from cultural evolution
(historical change) alone (Arbib 2005 ; Kemmerer 2005 ). On the
former view, the brain might have innate biological mechanisms
for processing nouns and verbs, as well as principles and param-
eters for combining them with words and morphemes of other
Mirror Systems, Imitation, and Language
508
Te modern approach to the semantics of modal logic, devel-
oped by Saul Kripke and others, is a form of truth condi-
tional semantics based on possible worlds . For example,
a sentence of the form S is true, at a given world w , if S is
true at every accessible world v. Diferent meanings of and
are represented by establishing diferent sets of worlds as acces-
sible. For example, suppose we use to represent it is morally
required that; then, at any world w , the accessible worlds are
those that are ideal, from the point of view of morality in w :
(1) ( the rich give money to the poor ) is true at w if the rich give
money to the poor in every accessible world v .
A world v is accessible if and only if it is a perfect world, from
the point of view of the moral principles holding in w .
Te semantics of is given by replacing every with some in (1) .
LINGUISTIC THEORIES BASED ON POSSIBLE WORLDS. Most lin-
guistic theories of modal semantics are based on possible worlds.
For example, Angelika Kratzer ( 1981 ) refnes the approach by
defning the set of accessible worlds in terms of two conversa-
tional backgrounds. According to Kratzer, the conversational
backgrounds for (2) are i) relevant facts and ii) moral principles.
Simplifying somewhat:
(2) Te rich must give money to the poor is true at w if the rich
give money to the poor in every world v which is i) consis-
tent with the relevant facts in w and ii) as good as possible
from the point of view of relevant moral principles in w .
NON-TRUTH CONDITIONAL THEORIES. Many philosophers and
linguists have argued that epistemic modals (see the follow-
ing) lack truth conditions. Instead, they are said to indicate the
speakers level of commitment to what he or she is saying (e.g.,
Palmer 2001 ). Dynamic modal logic (Groenendijk, Stockhof, and
Veltman 1996 ) combines ideas from possible worlds semantics
with a non-truth conditional analysis of epistemic modality. Te
fundamental semantic concept of dynamic logic is update poten-
tial , the capacity of a sentence to afect an information state (for
example, someones knowledge state or the information shared
in a conversation). Although the update potential of some sen-
tences can be defned in terms of truth conditions, that of an epi-
stemic sentence cannot be .
FUNCTIONAL THEORIES. functional linguistics has
made important contributions to our understanding of the
history (e.g., Traugott and Dasher 2002 ) and typology (e.g.,
Bybee, Perkins, and Pagliuca 1994 ) of modality. cognitive
linguistics ofers a theory of modality based on metaphor
(e.g., Talmy 1988 ) .
Varieties of Modality
SENTENTIAL MODALITY. Most linguists take as the central cases of
modality examples in which some expression combines with a
nonmodal sentence, making it modal. For example, English must
can be analyzed as: must + ( the rich give money to the poor ). Tere
are many distinct subtypes of sentential modality, including:
(1) Deontic modality: having to do with rules, including
morality and law (example: Criminals must be punished ).
Myowa-Yamakoshi , M. , and T. Matsuzawa . 1999 . Factors infuencing
imitation of manipulatory actions in chimpanzees ( Pan troglodytes ).
Journal of Comparative Psychology 113 : 128 36.
Pinker , S. , and P. Bloom . 1990 . Natural language and natural selection .
Behavioral and Brain Sciences 13 : 707 84.
Rizzolatti , G. , and M. A. Arbib. 1998 . Language within our grasp . Trends
in Neuroscience 21.5 : 188 94.
Rizzolatti , G., L . Fadiga , V. Gallese , and L. Fogassi . 1996 . Premotor cor-
tex and the recognition of motor actions . Cognitive Brain Research
3 : 131 41.
Seyfarth , R. M. , D. L. Cheney , and T. J. Bergman . 2005 . Primate social
cognition and the origins of language . Trends in Cognitive Sciences
9.6 : 264 6.
Stokoe , W. C. 2001 . Language in Hand: Why Sign Came Before Speech .
Washington, DC : Gallaudet University Press .
Tomasello , M. 1999 . Te human adaptation for culture . Annual Review
of Anthropology 28 : 509 29.
Tomasello , M., J. Call, J. Warren, T. Frost, M. Carpenter , and K. Nagell .
1997 . Te ontogeny of chimpanzee gestural signals. In Evolution
of Communication , ed. S. Wilcox, B. King, and L. Steels, 22459.
Amsterdam and Philadelphia: John Benjamins.
Yoshikawa , Y., M. Asada, K. Hosoda , and J. Koga . 2003 . A constructivist
approach to infants vowel acquisition through mother-infant interac-
tion . Connection Science 15 : 245 58.
MODALITY
Defnition
In its broadest sense, this term encompasses all means by which
we can talk about hypothetical situations. Te conception of
modality includes the following, plus more:
(1) Expressions of necessity and possibility, in any sense
of these terms (English examples: necessary, possible, must,
may ).
(2) Expressions of knowledge, belief, desire, and so on ( know,
believe, want, must ).
(3) Expressions used to indicate how strongly the speaker is
committed to what he or she is saying ( perhaps, might ).
(4) Expressions used to say that some action is obligatory or
permissible ( have [ to ] , must, may, allowed, permit ).
(5) Conditional sentences ( If then ) .
Tere is a range of narrower senses of the term, each used to
describe a grammatical category or set of related categories. For
example, English can express necessity, possibility, obligation,
permissibility, and ability (and other concepts) by means of a
grammatically special set of auxiliary verbs ( must, may, should,
can, etc.) . So, when studying English, it is reasonable to defne
modality as the range of meanings expressed by these verbs. But
other languages do not have this grammatical category, and so
its also reasonable to defne modality diferently when studying
these languages .
Semantic Theories
MODAL LOGIC. Much research on modality in natural language
has been inspired by modal logic (see Blackburn, de Rejke, and
Venema 2001 for a brief history). Modal logic typically has two
modal operators, (necessarily, must, or obligatory) and
(possibly, may, permissible), which attach to sentences.
Modality
509
MODERN WORLD-SYSTEM, LANGUAGE AND THE
Ever since the modern world-system came into existence in the
long sixteenth century, language has been a primary political
concern and a locus of major political struggle. In particular,
the issue of the language or languages that one will require to be
learned and used has been a subject of decisions by states in
their constitutions, their legislation, and/or their executive poli-
cies (see language policy ).
In the modern world-system, all included territory falls within
the jurisdiction of individual states. All states are linguistically
heterogeneous in the languages used within households, though
some much more so than others. Seeking to be a strong state,
most states have proclaimed an of cial language, meaning that
laws are written, governmental processes conducted, and edu-
cation ofered in the of cial language. Sometimes, but rarely,
it means that no other language may be used in public locales,
including in signage.
Some states have had more than one of cial language, and
some distinguish between the of cial language and one or more
national languages (which have some more restricted legal
rights). As an overall rule, almost every state has been pressed to
adopt a single of cial language. Te usual argument, aside from
the convenience, is that a single language favors national inte-
gration, part of a process of turning a state into a nation-state.
Integration is particularly an issue when there are large immi-
grant groups who speak a diferent language.
In many states, speakers of so-called minority languages, in
the name of cultural rights, resist eforts to impose a single of -
cial language. In particular, they demand the right to use other
languages in governmental business and schools. Whether
states yield to such demands is largely a question of the inter-
nal balance of power and demographic strength of the dominant
linguistic group, as well as the degree of support a minority lan-
guage may have from powerful neighboring states in which this
states minority language is the neighbors majority language. In
multilingual states, there is often social resistance by the users of
the language second in strength to learning well and using the
primary language .
Te problem is compounded beyond the boundaries of a sin-
gle state. Strong regional powers favor the learning of their lan-
guage by states that they consider to fall within their orbit. Tey
use direct political pressure, the benefts of economic ties, or
cultural liaison. Adoption of particular alphabetic or ideographic
systems also favors the dominant regional power. If a small state
breaks politically with a regional power and allies itself with
another world power, it often seeks to demonstrate and cement
the new ties by adopting a new secondary language or (if rele-
vant) changing the alphabetic system.
It is at the world or continental level that the issue becomes
most contentious. Tere are practical benefts in using as few lan-
guages as possible fnancial costs, ease of communication, and
savings in the time and efort required for either translation or
interpretation. However, the political implications of eliminating
a particular language as a legitimate option in interstate commu-
nication are very large. Te United Nations now has six of cial
languages. Two are working languages English and French.
Te inclusion of French has been the result of continuing and
(2) Epistemic modality: having to do with knowledge ( It
must be raining ).
(3) Subjective modality: having to do with the speakers
point of view (overlapping with 1 and 2).
(4) Dynamic modality: having to do with ability or the laws of
the natural world ( Ducks can swim ).
Much work in syntax has studied the representation of sen-
tential modality. Two important issues are the extent to which
modality is represented in the same ways across languages and
whether diferent subtypes are realized in diferent grammatical
positions (e.g., Cinque 1999 ) .
SUBSENTENTIAL MODALITY. Broad defnitions of modality will
include verbs and adjectives such as know and likely . Tey will
also include mood , the category of expressions that refect the
presence of modal meaning in the sentence but that do not intro-
duce modal meaning themselves (for example, indicative and
subjunctive verb forms; e.g., Farkas 1985 ) .
DISCOURSE MODALITY. Some varieties of modality operate at the
discourse level. evidentials are forms indicating the speakers
source or quality of information (e.g., Willett 1988 ). Te concept
of illocutionary force is connected to modality as well
(e.g., imperative sentences direct the addressee to perform a
hypothetical action) . Discourse modality overlaps with senten-
tial and subsentential modality. For example, in some languages
the subjunctive mood can operate as an imperative .
Paul Portner
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Blackburn , Patrick , Maarten de Rejke , and Yde Venema . 2001 . Modal
Logic . Cambridge : Cambridge University Press .
Bybee , Joan , Revere Perkins , and William Pagliuca . 1994 . Te Evolution of
Grammar: Tense, Aspect, and Modality in the Languages of the World .
Chicago : University of Chicago Press .
Cinque , Guglielmo . 1999 . Adverbs and Functional Heads: A Cross-
Linguistic Perspective . Oxford : Oxford University Press .
Farkas , Donka . 1985 . Intensional Descriptions and the Romance
Subjunctive Mood . New York : Garland .
Garson , James . 2007 . Modal logic. In The Stanford Encyclopedia
of Philosophy (summer ed.), ed. Edward N. Zalta . Available
online at: http://plato.stanford.edu/archives/sum2007/entries/
logic-modal/ .
Groenendijk , Jeroen, Martin Stockhof , and Frank Veltman . 1996 .
Coreference and modality. In Te Handbook of Contemporary
Semantic Teory , ed. S. Lappin , 179213. Oxford : Blackwell.
Kratzer , Angelika . 1981 . Te notional category of modality. In Words,
Worlds, and Contexts , ed. H.-J. Eikmeyer and H. Rieser , 3874.
Berlin : de Gruyter .
Palmer , F. 2001 . Mood and Modality . Cambridge : Cambridge University
Press .
Portner, P. 2009. Modality. Oxford: Oxford University Press.
Talmy , Leonard . 1988 . Force dynamics in language and cognition .
Cognitive Science 12 : 49 100.
Traugott , Elizabeth , and Richard Dasher . 2002 . Regularity in Semantic
Change . New York : Cambridge University Press .
Willett , Tomas . 1988 . A cross-linguistic survey of the grammaticaliza-
tion of evidentialit y. Studies in Language 12.1 : 51 97.
Modality Modern World-System, Language and the
510
the central system responsible for problem solving and abstract
thought. Modularity is in most striking contrast with theories
such as connectionism that treat cognition as the emergent
product of an unstructured neural network.
Teories of modularity come in a variety of favors with some-
times incompatible properties, making evaluation of the general
hypothesis dif cult. I begin with the best-known example Jerry
Fodors ( 1983 ) Modularity of Mind and then contrast it with
alternative views.
For Fodor, modules, or input systems, convert sensory
inputs into representations on which the central system of
the mind can operate . Incoming stimuli of a visual, auditory, tac-
tual, or other kind are converted into a form that, in conjunction
with knowledge drawn from memory, is adequate for problem
solving or the fxation of belief. Such beliefs are typically nei-
ther complex nor profound: Hearing a whining noise and seeing
a wagging tail may activate enough encyclopedic knowledge to
make you fx the belief that the dog wants to go out.
Fodor argued that input systems (corresponding to the
senses) all share a number of properties. Each has a specifc
domain of operation (vision, audition, and so on); they act fast
and mandatorily (you have no choice but to see a dog as a dog);
they are subserved by dedicated neural architecture and, hence,
are subject to idiosyncratic pathological breakdown (you can be
blind without being deaf, and vice versa); they are innately deter-
mined (hence, universal and uniform across the species); and,
most importantly, they are informationally encapsulated (that
is, the operation of the input systems proceeds independently
of information stored in memory). You may know that railway
lines dont really converge in the distance but your visual system
still makes them look as if they do. Fodor then suggested that any
system that shared the properties of the sensory input systems
was by defnition a module, with the result that language was
included as a module just like vision .
Tis claim highlights a radical distinction between Fodors
version of modularity and Noam Chomskys earlier one ( 1975 )
that treats modules as knowledge structures, rather than as pro-
cessing systems. Te language faculty is a system of knowledge
that can be accessed by both input and output systems: We pro-
duce as well as understand language. A further diference is that
Fodor is pessimistic about the possibility of saying anything inter-
esting about the structure of the inscrutable central system,
whereas Chomsky is more optimistic, suggesting that the central
system too is modular, with moral judgment, music, and other
faculties all having specifc (if not localized) areas of the mind
dedicated to them. On a point of terminology, it is also impor-
tant to note that Chomsky (and linguists more generally) use the
term module for the various subparts of the language faculty (the
lexicon and the computational system with components such as
binding, control, movement, etc.).
It is clear that even if they share some of the Fodorian prop-
erties such as innate specifcation and domain specifcity, moral
judgment and the sense of smell are radically diferent. Tis
has led to the suggestion that we need a distinction between
(Fodorian) modules and quasi - modules , or modules of the cen-
tral system (Smith and Tsimpli 1995 ), where these are defned
in terms of the properties (such as informational encapsula-
tion) that they possess and the kind of vocabulary, perceptual or
very strong political pressure from France. Te European Union
has decided that any of cial language of a member state may be
used. Since this number is very large, and the costs of translating,
for example, Maltese into Finnish are enormous, the result has
been a creeping usage of English as the de facto but not de jure
of cial language.
In the history of the modern world-system, as Latin fell out
of diplomatic usage, French took its place. Since 1945, given the
hegemony of the United States in the world-system, English has
displaced French. Te story in international scientifc discourse
is diferent. In the nineteenth century, German was the favored
lingua franca. After 1918 and especially after 1945, because of
defeats on the battlefeld, it lost this status. Before 1939, at an
international scholarly congress, participants felt free to deliver
their papers in English, French, German, and usually Italian as
well. Tere was normally no translation, and it was assumed that
scholars could understand at least three of the four languages.
After 1945, international scholarly organizations dropped
German and Italian entirely. In the 50 years since then, the use of
French has declined but is still permitted, and Spanish has joined
French as a permitted but seldom-used language. Te inclusion
of Spanish is the direct result of the fact that there are 19 states in
which it is an of cial language .
In commerce, there have always been lingua francas. Anyone
going to a local market in a major center of a country in the global
South will see merchants capable of conducting their business in
multiple relevant languages. If one looks at discussions among
personnel of large corporations, there has been an increasing
tendency to use English. Nonetheless, it is probably still the
case that the ability to use a widely spoken (of cial) language
other than English is an advantage to persons doing business in
countries outside the English linguistic zone. In commerce, the
decision on linguistic use is less a matter of coercion than of opti-
mizing the ability to engage in proftable business .
Finally, we should notice the consequences for the geocul-
ture of the world-system of the existence of dominant languages.
Te widespread use of English in the twenty-frst century is very
advantageous for native English speakers. It is not merely conve-
nient but tends to turn English linguistic eyes into world linguis-
tic eyes. It also, however, has its negative side for native English
speakers. Tey are often the only ones cut of from the internal
communications of other linguistic zones, as well as from the
possibilities of seeing the world through other linguistic eyes .
It is quite possible that the increasing role of the Internet
in communications of all kinds, along with the declining
power of the United States in the world-system, will lead to the
reemergence of a multipolar linguistic situation, with fve to
seven world languages that diplomats, scholars, and business
executives will feel the need to master and use .
Immanuel Wallerstein
MODULARITY
Modularity is the claim that human cognition is compartmental-
ized into a number of discrete components or modules poten-
tially including vision, audition, moral judgement, theory of
mind, and language. Tese specialized modules contrast with
Modularity
511
. 2000 . Te Mind Doesnt Work Tat Way: Te Scope and Limits of
Computational Psy chology . Cambridge MA : MIT Press .
Karmilof-Smith , A. 1992 . Beyond Modularity . Cambridge, MA : MIT
Press .
Plomin , R. , and P. Dale 2000 . Genetics and early language develop-
ment: A UK study of twins. In Speech and Language Impairments
in Children: Causes, Characteristics, Intervention and Outcome , ed.
D. Bishop and L. Leonard , 3551. Philadelphia : Psychology Press .
Smith , N. 2003 . Dissociation and modularity: Refections on language
and mind. In Mind, Brain and Language , ed. M. Banich and M. Mack ,
87111. Mahwah, NJ : Lawrence Erlbaum . Tis article treats in greater
depth many of the issues discussed here.
Smith , N. , and I.-M. Tsimpli . 1995 . Te Mind of a Savant: Language-
Learning and Modularity . Oxford , Blackwell .
Sperber , D. 2002 . In defense of massive modularity. In Language, Brain
and Cognitive Development: Essays in Honor of Jacques Mehler , ed.
E. Dupoux , 4757. Cambridge, MA : MIT Press .
MONTAGUE GRAMMAR
Montague grammar is a theory of semantics and the syntax-
semantics interface developed by the logician Richard Montague
(193071) and subsequently modifed and extended by linguists,
philosophers, and logicians. Classical Montague grammar
had its roots in logic and the philosophy of language; it quickly
became infuential in linguistics, and linguists played a large
role in its evolution into contemporary formal semantics .
Te most constant features of the theory over time have been
the focus on truth conditional aspects of meaning (see truth
conditional semantics ), a model-theoretic conception of
semantics, and the methodological centrality of the principle of
compositionality .
History
Montague was a student of Alfred Tarski (190283), a pioneer in
the model-theoretic semantics of logic. Montague developed an
intensional logic with a rich type theory and a model-theoretic
possible worlds semantics , incorporating certain aspects
of (formal) pragmatics, including the treatment of indexical
words and morphemes like I, you and the present tense. In the
late 1960s, Montague turned to the project of universal grammar ,
which for him meant a theory of syntax and semantics encom-
passing both formal and natural languages.
Montagues idea that a natural language could be formally
described using logicians techniques was radical. Most logi-
cians considered natural languages too unruly for precise
formalization, while most linguists either had no awareness
of model-theoretic techniques in logic or doubted the appli-
cability of logicians methods to natural languages (Chomsky
1955 ).
At the time of Montagues work, generative grammar was
established, linguists were developing approaches to semantics,
and the relation of semantics to syntax had become central. Te
linguistic wars between generative semantics and inter-
pretive semantics were in full swing (Harris 1993 ). In introdu-
cing Montagues work to linguists, Barbara Partee ( 1973 , 1975 )
and Richmond Tomason ( 1974 ) argued that Montagues work
ofered some of the best aspects of both warring approaches,
with added advantages of its own .
conceptual, over which they are defned. An extreme version of
this position is the claim (cf. Sperber 2002 ) that the mind is mas-
sively modular, with everything from individual concepts like
dog , to Fodorian modules like vision, to our general pragmatic
ability to interpret utterances being modules. It is unclear what
the identity criteria for a module are in such theories, and Fodor
himself is vehemently opposed to the claim (cf. Fodor 2000 ).
A rival view (e.g., Karmilof-Smith 1992 ) accepts that the
minds structure is modular but denies that it is innately deter-
mined, suggesting instead that the (adult) modular structure
arises as a result of a process of modularization on the basis
of interaction with the environment during development.
Connectionists (e.g., Elman et al. 1996 ) are more radical and
deny the validity of modularity and its use of rules and represen-
tations entirely, relying instead on the ability of neural networks
to simulate the properties of rule-based systems.
Te major evidence for modularity in all its guises is (dou-
ble) dissociation. Although it is typically the case that abilities
and disabilities cut across domains (if youre good at one sub-
ject youre likely to be good at others, hence the possibility of
assigning people an intelligence quotient ), the existence of dis-
sociations demonstrates the intrinsic separability and autonomy
of the various components of the mind. For instance, intelligence
and language may doubly dissociate. It is possible to combine
high intelligence and good language (you), low intelligence and
good language (linguistic savants like Christopher [Smith and
Tsimpli 1995 ]), high intelligence and poor (or nonexistent) lan-
guage, as in some kinds of aphasia, and low intelligence and poor
language (as in typical Down syndrome subjects) .
Evidence for some version of innate modularity versus modu-
larization due to interaction with the environment comes from
the developmental trajectory of normal children. R. Plomin and
P. Dale ( 2000 ) demonstrate that when tested over time, children
typically start with diferent abilities in the verbal and nonver-
bal domains and then gradually converge so that their abilities
are similar across domains. Tis is exactly the opposite of what
one would expect on a modularization story. Similarly, connec-
tionist claims that modularity is unnecessary are undermined
by the implications of connectionisms uniform reliance on sta-
tistics. Te mind exploits statistical regularities in the input dif-
ferently in diferent domains, and Neil Smith and I.-M. Tsimpli
( 1995 ) demonstrate that connectionist models are undesirably
powerful in that they can infer statistical regularities that normal
humans cannot .
Modularity, of some kind, is still the most successful theory
of cognition there is. It has rivals and it has problems, but it is
indispensable .
Neil Smith
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , N. 1975 . Refections on Language . New York : Pantheon . A sem-
inal source for modularity.
Elman , J., E. Bates, M. Johnson, A. Karmilof-Smith, D. Parisi , and
K. Plunkett . 1996 . Rethinking Innateness: A Connectionist Perspective
on Development . Cambridge, MA : MIT Press . A sustained alternative
to modularity.
Fodor , J. 1983 . Te Modularity of Mind . Cambridge, MA : MIT Press . Te
classic and best-known statement of the modularity thesis.
Modularity Montague Grammar
512
a formal system; Montagues thesis was that English can be
described as an interpreted formal system .
Barbara H. Partee
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bach , Emmon . 1989 . Informal Lectures on Formal Semantics. New
York : State University of New York Press .
Chomsky , Noam . 1955 . Logical syntax and semantics: Teir linguistic
relevance . Language 31 : 36 45.
Dowty , David, Robert E. Wall , and Stanley Peters , Jr. 1981. Introduction to
Montague Semantics . Dordrecht, the Netherlands : Reidel. Te classic
textbook on Montague grammar .
Gamut , L. T. F. 1991 . Logic, Language, and Meaning. Vol. 2. Intensional
Logic and Logical Grammar . Chicago : University of Chicago Press . A
good, rigorous introduction to Montague grammar and its logic.
Harris , Randy Allen . 1993 . Te Linguistics Wars . Oxford : Oxford University
Press .
Lewis , David . 1970 . General semantics . Synthese 22 : 18 67.
Montague , Richard . 1970 . Universal Grammar. Teoria 36 : 373 98.
Repr. in Montague 1974, 22246.
. 1973 . Te proper treatment of quantifcation in ordinary English.
In Approaches to Natural Language , ed. K. J. J. Hintikka et al., 22142.
Dordrecht, the Netherlands : Reidel. Repr. in Montague 1974, 24770.
. 1974 . Formal Philosophy: Selected Papers of Richard Montague .
Ed. R. Tomason . New Haven, CT : Yale University Press .
Partee , Barbara . 1973 . Some transformational extensions of Montague
grammar . Journal of Philosophical Logic 2 : 509 34.
. 1975 . Montague grammar and transformational grammar .
Linguistic Inquiry 6 : 203 300.
Partee , Barbara H., with Herman L. W. Hendriks . 1997 . Montague gram-
mar. In Handbook of Logic and Language , ed. J. van Benthem and A.
ter Meulen , 591. Amsterdam and Cambridge, MA : Elsevier and MIT
Press . A fuller history and explication of Montague grammar and its
impact.
Tomason , Richmond . 1974 . Introduction. In Formal
Philosophy: Selected Papers of Richard Montague , ed. R. Tomason ,
169. New Haven, CT : Yale University Press .
MOOD
Mood forms part of the nonspatial setting of an event, alongside
modality , reality status, tense , aspect , and evidentiality .
Mood refers to a type of speech-act , with three basic choices.
Many languages have a special verb form marking commands,
which is known as imperative mood . In Latin, the second per-
son imperative dic means (you) say! and is diferent from the
statement dicis , you say. Declarative mood (sometimes called
indicative ) is used in statements . Many more categories tend to
be expressed in declarative clauses than in either interrogative or
imperative. Interrogative mood occurs in questions as in West
Greenlandic where every question is marked with a special suf x
on verbs (Fortescue 1984 , 49, 28798) .
In traditional uses, the notion of mood applied to sets of
infectional verb forms. Te Western classical tradition, based on
Greek and Latin, identifed three moods: indicative, subjunctive,
and imperative, which only partially correspond to the afore-
mentioned three speech-acts. Further meanings associated with
mood involve optative and dubitative (see Lyons 1977 , 725848;
Sadock and Zwicky 1985 ). Some scholars consider conditional
modality which marks a clause in a conditional sentence and
The Theory and Substance of Montague Grammar
It was the short but densely packed PTQ (Te proper treatment
of quantifcation in ordinary English, Montague 1973 ) that had
the most impact on linguists and on the development of formal
semantics. Montague grammar has often meant PTQ and its
extensions by linguists and philosophers in the 1970s and 1980s.
But it is the broader algebraic framework of UG (Universal
Grammar, Montague 1970 ) that constitutes Montagues theory
of grammar. Crucial features of that theory include the truth con-
ditional foundations of semantics, the algebraic interpretation of
the principle of compositionality, and the power of a higher-or-
der typed intensional logic.
Before Montague, semanticists focused on the explication of
ambiguity , anomaly, and semantic relatedness ; data were often
subjective and controversial. Te introduction of truth condi-
tions and entailment relations as core data profoundly afected
the adequacy criteria for semantics and led to a great expansion
of semantic research. While some cognitively oriented linguists
reject the relevance of truth conditions and entailment relations
to natural language semantics, many today seek a resolution of
meaning externalism and internalism by studying mind-
internal intuitions of mind-external relations, such as refer-
ence and truth conditions.
In UG, Montague formalized the Fregean principle of com-
positionality as the requirement of a homomorphism between a
syntactic algebra and a semantic algebra. Te nature of the ele-
ments of both the syntactic and the semantic algebras is open to
variation; what is constrained by compositionality is the relation
of the semantics to the syntax, making compositionality as rel-
evant to representational and conceptual theories of meaning as
it is to model-theoretic semantics .
Te richness of Montagues logic made possible a composi-
tional semantic interpretation of independently motivated syn-
tactic structure (see autonomy of syntax ), which was key in
overcoming the problems that underlay the linguistic wars. Tis
was well illustrated in PTQ, where a typed higher-order logic with
lambda-abstraction made it possible to interpret noun phrases
(NPs) like every man, the man, a man uniformly as semantic
constituents ( generalized quantifers ), an idea simultaneously
advocated by David Lewis ( 1970 ). PTQ also contained innovative
treatments of quantification and binding, intensional tran-
sitive verbs, phrasal conjunction, adverbial modifcation, and
more. Montagues type theory introduced to linguists Freges
strategy of taking function-argument application as the basic
semantic glue for combining meanings, giving renewed sig-
nifcance to categorial grammar .
Montagues logic was an intensional logic, developing
Gottlob Freges distinction between sense and reference
and Rudolf Carnaps distinction between intension and
extension, using possible world semantics to treat the phe-
nomenon of referential opacity, pervasive in propositional
attitude sentences and many other constructions (see
intentionality ) .
Details of Montagues analyses have been superseded,
but in overall impact, PTQ was as profound for semantics as
Noam Chomskys Syntactic Structures was for syntax. Emmon
Bach ( 1989 , 8) summed up their cumulative innovations
thus: Chomskys thesis was that English can be described as
Montague Grammar Mood
513
and in Caddo, from North America, realis marks statements and
commands, while irrealis expresses future, possibility, and con-
dition. Tis shows that the realisirrealis distinction is language
specifc and that it is distinct from mood (see Mithun 1999 ,
17880). Mood, modality, and reality status are distinct from evi-
dentiality (q.v.) whose primary meaning is information source .
Mood is often an obligatory infectional category of the verb,
marked with af xes (suf xes or prefxes, rarely infxes); it is never
expressed derivationally. In languages of an isolating profle,
mood can be expressed through particles. Modalities are not
obligatory and, thus, do not constitute part of an infectional par-
adigm. Modal verbs express modalities rather than moods (this
is the case in English and many other familiar Indo-European
languages).
Forms of mood marking can develop additional meanings
overlapping with modalities. Imperative forms can be used to
express optative and conditional, while indicative forms may
develop overtones of certainty (associated with epistemic modal-
ity). Indicative forms for instance, future can be used as com-
mand strategies, with diferences in illocutionary force .
Alexandra Y. Aikhenvald
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aikhenvald , Alexandra Y. 2004 . Evidentiality. Oxford : Oxford University
Press .
Chung , Sandra, and Alan Timberlake . 1985 . Tense, aspect and mood.
In Language Typology and Syntactic Description. Vol. 3: Grammatical
Categories and the Lexicon . Ed. Timothy Shopen , 20258.
Cambridge : Cambridge University Press .
Dixon , R. M. W. 1994 . Ergativity . Cambridge : Cambridge University
Press .
Elliott , Jennifer R. 2000 . Realis and irrealis: Forms and concepts of the
grammaticalisation of reality . Linguistic Typology 4 : 55 90.
Fortescue , Michael. 1984 . West-Greenlandic . Beckenham, UK : Croom
Helm .
Jespersen , Otto. 1924 . Te Philosophy of Grammar . London : George Allen
and Unwin .
Lyons , John. 1977 . Semantics . Vol. 2 . Cambridge : Cambridge University
Press .
Mithun , Marianne. 1999 . Te Languages of Native North America .
Cambridge : Cambridge University Press .
Palmer , F. R. 1986 . Mood and Modality . Cambridge : Cambridge University
Press .
Sadock , Jerrold M. , and Arnold M. Zwicky . 1985 . Speech act distinctions in
syntax. In Language Typology and Syntactic Description. Vol. 1: Clause
Structure . Ed. Timothy Shopen , 15596. Cambridge : Cambridge
University Press .
MORPHEME
Tis term has been used in two ways: In Leonard Bloomfelds
sense, a morpheme is a minimal meaningful form; in Zellig
Harriss and Charles F. Hocketts later usage, a morpheme is an
abstract unit of analysis realized by a morph (= minimal mean-
ingful form) or by a set of synonymous morphs in complemen-
tary distribution. In Bloomfelds sense, the plural suf xes -s
and - en are distinct morphemes; in the latter sense, they are
distinct morphs realizing the same morpheme. Te term is not
always used consistently in morpheme-based approaches to
subjunctive modality typically, a form expressing desire or
uncertainty on a par with moods. Tis is problematic since
the distinction between moods as speech-acts and clause types
(which include division between main and subordinate clauses,
where conditional forms would be used) is blurred. Te intro-
duction of interrogative mood into the system is largely due to
the existence of languages that have an overtly marked verbal
paradigm used for the interrogative speech-act, as in a number
of languages of Amazonia. Further formal distinctions between
moods as clause types involve prosody and constituent order.
Both imperative and interrogative are characterized by a typi-
cal intonatio n contour. Imperatives often have fewer catego-
ries than corresponding declaratives. Te English imperative is
perhaps the simplest form in the language: It consists of the base
form of the verb without any tense infection, whose subject
typically, the addressee can be and often is omitted. In con-
trast, many languages of North and South America and Siberia
distinguish delayed versus immediate imperatives and proximal
versus distal imperatives . Te universal property of imperatives
is having the second person as subject, of a transitive or intran-
sitive verb (Dixon 1994 , 13142). A prototypical imperative is
agentive, and this is why in numerous languages imperative can-
not be formed on passive and stative verbs. Other moods do not
have such restrictions. Imperatives directed at the frst person
(e.g., Lets go! ), also known as hortatives, are often expressed dif-
ferently from second person imperatives. Imperatives directed at
the third person (e.g., Long live the king! ), also known as jussives,
may share similarities with frst person imperatives, or have
properties diferent from all other imperative forms. Further,
minor moods include exclamative (as in Tats so tacky! ) and
expressive types, such as imprecatives (or curses, often cast as
commands but without a command meaning) .
Mood interacts with modality, understood as a means used
by the speaker to express his or her opinion or attitude towards
the proposition that the sentence expresses or the situation
that the proposition describes (Lyons 1977 , 452). Expressions
of probability, possibility, and belief are epistemic modalities,
and expressions of obligation are deontic modalities. In English,
these meanings are conveyed by modal verbs, e.g., he might
come or he must have come (epistemic), he must come (deontic)
(see Palmer 1986 , 51125; Jespersen 1924 , 3201). Further modal
distinctions include desiderative (unachievable desire), optative
(achievable desire), conditional, hypothetical, potential, pur-
posive, and apprehensive (lest). Languages with a rich verbal
morphology may have special marking for each distinction.
An alternative (rare) cover term for both mood and modality is
mode (Chung and Timberlake 1985 ) .
Some languages have an af x with a general meaning of
irrealis covering possibility, future, negative statements, and
commands. Tese languages have the category of reality status ,
the grammaticalized expression of an events location either in
the real world or in some hypothetical world (see Elliott 2000 , for
its cross-linguistic validity). In Maung, an Australian language,
statements in the present, past, and future are marked with rea-
lis suf xes. Potential meanings I can do X are expressed with
irrealis, as are commands. In Manam, an Oceanic language, irre-
alis covers future, probable, and counterfactual statements, pos-
itive commands, and habitual actions. But in Yuman languages
Mood Morpheme
514
once operated between words can give rise to alternations with
morphological value; thus, the consonant mutations of Irish
Gaelic, such as b/v/m in ba:d , va:d , ma:d her, his, their
boat, respectively, result ultimately from the diferential efects
of former word-fnal consonants of possessors *as , *a , *an on
word-initial b . Te efects of sound changes, such as the erosion
of fnal syllables whereby English singular and plural forms
oxe , oxene became ox , oxen , can lead to the reanalysis of internal
word structure so that -en is interpreted as a (new) plural suf-
fx. Loss of fnal t in the pronunciation of French argent silver
caused argent-ier silversmith to be reinterpreted as argen-tier ,
and allowed the new pattern to be extended to create derivatives
such as bijou-tier jeweler from bijou .
Syntax may supply the source of new morphology , as phrases
are reinterpreted as single words (by a process called univerbation ).
Former clitics may be reanalyzed as af xes in a process often
described as a kind of grammaticalization with accompany-
ing functional changes. Tus in Tocharian, new case suf xes (with
meanings such as toward, through, with, from) were cre-
ated by fusing former postpositions with nouns in the oblique case.
Complex new infectional markers can be created, such as French
aim-eras you will love, where the suf x includes part of the Latin
infnitive suf x, auxiliary verb have, second-singular (2sg) sub-
ject marker as the Romance future was grammaticalized from a
construction have to VERB. Te univerbation of phrases can even
lead to word-internal infection, with grammatical markers becom-
ing trapped between erstwhile lexical elements, for example, in the
slightly archaic English whomever and whose-ever , where ever was
once a separate word, or in Old Irish atotch sees you (vs. atch
sees), where ot continues an earlier pronoun that was positioned
between the two words that together meant see.
Much morphological change involves only rearrangements
within the morphology itself, within and across paradigms, and
involving either stems or af xes. In leveling , one variant of a
stem is extended to all cells in an infectional paradigm; thus, in
Ancient Greek, the prehistoric k
w
of * leik
w
- leave developed by
regular sound change into t or p before diferent vowels, but the
leip- variant was later generalized to the whole paradigm. Stem
variants may be redistributed according to a pattern prevalent
in other paradigms by a process called analogical change (see
analogy; synchronic and diachronic ). Tus, in the early
modern German verb give the variant gib- , which arose by
sound change in all the singular forms, was later confned to the
second and third persons singular because many other verbs, for
example, sleep, had a pattern where only these two forms had
a diferent stem vowel. (See Table.)
morphology . In paradigm-based approaches, no linguistic
principle is assumed to make essential reference to morphemes
as a unifed class of elements .
Gregory Stump
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bloomfeld , Leonard . 1933 . Language . New York: Henry Holt and Co.
Harris , Zellig S. 1942. Morpheme alternants in linguistic analysis .
Language 18 : 169 80.
Hockett , Charles F. 1947 . Problems of morphemic analysis . Language
23 : 321 43.
MORPHOLOGICAL CHANGE
Morphological change involves alterations made by speakers over
time to the analysis of complex words, or to the relations between
a lexical base and its compounds and derivatives, or to the set
of infected words that share a common lexical base. Te main
mechanisms of morphological change are reanalysis (the reinter-
pretation of forms) and the extension of patterns to create new
forms. Te impetus for reanalysis often comes from changes to
the semantics , phonology , or syntax of the afected forms.
Semantic shift may afect the function of a grammatical ele-
ment. Tus, in various Australian languages, a subordinating
purposive suf x (in order to VERB), typically -ku , came to be
used in independent clauses as a marker of intentional mood
(may VERB), then further shifted to express future tense (will
VERB). Functional shift in a grammatical af x also took place
in some Karnic (Australian) languages, in which a locative (at)
case suf x -nga came to mark the dative (for) function. Meaning
changes may lead to the reinterpretation of compound words as
simple lexemes (e.g. hlford loaf + ward > lord , shep-herd
sheep+ herder > shepherd ), or as a lexical stem plus a deriva-
tional af x ( king-dom ) with downgrading of a bound lexical form
-dom meaning condition to a derivational suf x .
Sound changes often create new allomorphs: For example,
the earlier English plural suf x -z split into three variants -z , - z ,
- s , and the stem long developed a variant leng - in length with the
e conditioned by the following i in the former derivational suf x
- ith . As the relationship between words becomes obscured by
the accumulation of sound changes, some phonological difer-
ences become morphologized, that is, reinterpreted as partial or
even sole signals of a morphological property. Tus, the vowel e
in slep-t (vs. ee in sleep ) helps to mark past tense, and ee in feet
(vs. oo in foot ) alone marks plural. Even sound changes that
Pre-Greek (Doric) Greek EMGerman ModGerman German
leave leave give give sleep
1Sg leip leip Gib geb-e schlafe
2Sg leiteis leipeis gib-st gib-st schlfst
3Sg leitei leipei gib-t gib-t schlft
1Pl leipomen leipomen geb-en geb-en schlafen
2Pl leitete leipete geb-t geb-t schlaft
3Pl leiponti leiponti geb-en geb-en schlafen
Morphological Change
515
ev -ler- im - iz -den
house -PL -POSS. 1-PL -ABL
Te third type of language expresses diferences in morphosyn-
tactic and lexicosemantic properties through contrasting modi-
fcations, or infections of a words stem. Tese are infectional
or fusional languages. Te classical languages, Greek, Latin, and
Sanskrit, belong to this type. In Latin you (sg[singular]) loved
is expressed by various modifcations of the root am- love to
yield amvist : stem formative -v to express perfect, and - ist to
express perfect (again) + 2d person + singular. Typically, proper-
ties are fused in one exponent: Here aspect, person and num-
ber agreement are expressed together. Equally, a property can
be expressed by more than one exponent: Here perfect is being
expressed twice .
Tere has been general unease among modern linguists with
the classical typology. One reason is that languages rarely fall
cleanly into one of these types. For example, Mandarin Chinese
productively uses what looks like a derivational suf x to build
agentive nouns, the word q mechanism: sn-r q cooler,
jin-c q monitor, yng-shng q speaker; compare the
English -er / -or agentive suf x (Hippisley, Cheng, and Ahmad
2005 ). More importantly, there is some doubt that the typology
ofers any theoretical insight, a point argued as far back as Sapir
( 1921 ). Part of the reason is that morphological type is really a
function of other grammatical structures worthy of typologi-
cal investigation, and is, therefore, epiphenomenal (Anderson
1990 ).
A more promising approach is to focus on much more nar-
rowly defned word structures and to investigate how they
cross-cut languages that may or may not be genetically or typo-
logically related. Te result is then a typology of narrowly defned
structures of words that answer the question What is a possible
word? Tis is the approach taken by Greville G. Corbett and col-
leagues, who look at unusual morphology such as suppletion,
deponency, and defectivenesss, recording such structures in a
large number of individual languages and inducing diachronic
and synchronic models of their appearance and use in syntax
(e.g. Corbett 2007 ; Baerman and Corbett 2007 ) .
Andrew Hippisley
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Steven . 1990 . Sapirs approach to typology. In Contemporary
Morphology , ed. W. Dressler et al., 27795. Berlin : Mouton de Gruyter .
Baerman , Matthew , and Greville G. Corbett . 2007 . Linguistic typol-
ogy: Morphology . Linguistic Typology 11 : 115 17.
Corbett , Greville G . 2007 . Canonical typology, suppletion and possible
words . Language 83 : 8 42.
Hippisley , Andrew , David Cheng , and Khurshid Ahmad . 2005 . Te
head modifer principle and multilingual term extraction . Natural
Language Engineering 11.2 : 129 57.
Sapir , Edward . 1921 . Language . New York : Harcourt, Brace and World .
MORPHOLOGY
Morphology and Words
While the lexicon of a language lists basic forms and their content
(meanings and grammatical properties), a languages complex
Where there are diferent infectional classes, one class is
usually dominant and its infectional pattern tends to infuence
the others. Tus, in early Italic languages, noun stems in -, i-,
u- remodeled their former ablative singular forms on the pattern
of d in the dominant o -stem class, creating new endings in -d ,
-d , -d , respectively. Words are often transferred from an irregu-
lar infectional class to the dominant one: Tus, English drag , a
former strong verb with past drug , changed to the weak class
with regular past infection dragged .
Iconicity has been emphasized by Natural Morphologists
as one of the principles motivating morphological change.
Iconically organized paradigms code more complex grammati-
cal meanings by means of phonologically larger markers and
simpler meanings with smaller markers, and the most basic
meanings (singular number, nominative case, present tense,
third person agreement , etc.) by no marker at all. In some
Slavic languages, after sound changes created zero case-number
suf xes in both the nominative singular and genitive plural of
o -stem nouns, the paradigm was repaired only in the genitive
plural (by substituting an overt suf x -ov from another infec-
tional class), whereas the iconic zero marking was retained in
the nominative singular. Iconicity in the syntagmatic dimension
is increased by changes that reorder a form like (Australian)
Arrernte me-k(e)-atye mother-to-my, where atye was origi-
nally an enclitic pronoun, to the sequence m(e)-atye-ke moth-
er-my-to, which better mirrors the semantic scopal relations
between the elements .
Harold Koch
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Stephen R. 1992 . Morphological change. In A-morphous
Morphology , 33672. Cambridge : Cambridge University Press .
Joseph , Brian D. 1998 . Diachronic morphology. In Te Handbook
of Morphology , ed. Andrew Spencer and Arnold M. Zwicky , 35173.
Oxford : Blackwell.
Koch , Harold . 1996 . Reconstruction in morphology. In Te Comparative
Method Reviewed: Regularity and Irregularity in Language Change , ed.
Mark Durie and Malcolm Ross , 21863. New York : Oxford University
Press .
MORPHOLOGICAL TYPOLOGY
typology has its origins in nineteenth-century morphologi-
cal typology, a method of grouping languages not according
to genetic relatedness but to structural similarity, where the
structure was specifically word structure (see morphology ).
Traditionally, there are three possibilities for phonologically
expressing morphosyntactc (inflectional) and lexicosemantic
(derivational) properties at the level of the word. In an iso-
lating or analytical language, complex words are built from
existing words, free forms. Mandarin Chinese could be viewed
as an isolating language. Productive coining of new terms is
through compounding. The word for Internet is h-lin
wng with h inter + lin related + wng net. In agglu-
tinating languages, the pieces of a complex word map onto
specific meaning elements biuniquely, both at the lexical and
grammatical level. Turkish evlerimizden from our houses is
glossed as
Morphological Typology Morphology
516
comprises two subsystems . Te derivational subsystem derives
one lexeme from another; for instance, the derivational subsys-
tem of English derives the verbal lexeme crystallize from the
nominal lexeme crystal . Te compounding subsystem defnes
complex lexemes through the combination of other lexemes;
thus, the compounding subsystem of English creates the com-
pound lexeme mountain lion from the lexemes mountain and
lion . A number of criteria have been adduced to distinguish
infection from derivation and to distinguish compounds from
syntactic combinations; see Booij ( 2000 ) and Matthews ( 1991 ),
respectively.
Incremental and Realizational Approaches to Morphology
Structuralist approaches to morphology gave primacy to
morphemes (minimal pairings of form with meaning) as the basic
units of morphological analysis. Tey were incremental in orienta-
tion, in that they presumed that the content of a word is the sum
of the content of its component morphemes. Tese structuralist
assumptions have been very persistent in modern linguistic theory.
Teir widespread acceptance has led many linguists to assume that
all morphological phenomena can be accounted for by indepen-
dently needed principles of syntax and phonology; thus, since the
advent of generative linguistics, morphological issues have often
been addressed as a part of syntax (Selkirk 1982 ; Lieber 1992 ) or as
a part of phonology (Kiparsky 1982 ).
Incrementalist theories of morphology are problematic, how-
ever (Stump 2001 , 3 f). First, there are words whose content can-
not be factored into that of their component morphemes; that
is, the content of a words individual morphemes may underde-
termine that of the word itself. Te aorist verb form krd-o-x I
stole in Bulgarian is unambiguous despite the fact that none of
its three component morphemes expresses frst-singular (1sg)
subject agreement; compare 2sg/3sg krde , 1pl krd-o-x-me ,
2pl krd-o-x-te , and 3pl krdo-x-a . To account for such forms,
proponents of incrementalist theories must postulate zero mor-
phemes, which lack overt phonological realization but purport-
edly supply the missing content.
Second, syncretism (the use of the same morphology to
express distinct content) is problematic for incrementalist theo-
ries. In Sanskrit, the accusative singular suf x -m is also used as
a nominative singular suf x in the paradigms of neuter a -stem
nouns: Compare the masculine noun horse (nom. sg. ava- ,
acc. sg. ava-m ) with the neuter noun gift (nom./acc. sg.
dna-m ). Incrementalist theories must attribute syncretism to
homonymy (e.g., to the existence of two distinct -m suf xes in
Sanskrit), but in doing so miss important generalizations (e.g.,
the fact that the nominative and accusative are always syncre-
tized in the paradigms of Sanskrit neuter nouns, regardless of
what the exponence of these cases might be) .
Finally, incrementalist assumptions give no explanation for
the incidence of extended exponence (the appearance, within a
single word, of more than one morpheme expressing the same
content). In Nyanja (Niger-Congo; Malawi), adjectives exhibit
noun-class agreement with the noun they modify, and members
of one group of adjectives exhibit two agreement prefxes, as in
the case of ci-pewa ca-ci-kulu large hat, where kulu large
agrees with the class 7 noun -pewa hat by means of two distinct
prefxes. On incrementalist assumptions, the prefx ci- in ca-ci-
words neednt be invariably listed, since their form and content
are often partially or wholly deducible from those of their parts
by means of regular principles. Tis system of principles is the
languages morphology .
In morphological theory, it is useful to distinguish three senses
of word . In one sense, the form come is the same word in (1) and
(2); in another sense, come in (1) is a diferent word from come
in (2). Tis apparent paradox arises because word can be used
to refer to either a phonological or a grammatical unit: In (1) and
(2), come represents the same phonological unit (phonetically
[km]) but two distinct grammatical units: the unmarked infni-
tive form of the verb come in (1) and the past participial form of
this verb in (2).
(1) Sandy should come home.
(2) Sandy has already come home.
Tus, it is useful to distinguish phonological words such as [km]
from grammatical words such as the past participle of come.
Moreover, there is a third theoretically relevant interpretation
of word according to which go and gone are diferent forms of
the same word . Here, word refers neither to a phonological word
nor to a grammatical word but to the abstract lexical element of
which go and gone are both realizations; abstract elements of this
sort are referred to as lexemes .
Tese three senses of word are related in the following
way: the pairing of a lexeme with an appropriate set of morpho-
syntactic properties defnes a grammatical word, as in (3), and
the phonological realization of a grammatical word is a phono-
logical word, as in (4).
Correspondingly, a lexemes paradigm is the full set of grammat-
ical words associated with it; a morphosyntactic property sets
exponence is its phonological realization; and where G is a gram-
matical word in the paradigm of lexeme L, Ls root is the phono-
logical form (if such can be identifed) with which the exponence
of Gs property set combines in the phonological word realizing
G. Tus, in the realization of the grammatical word (3b) in the
paradigm of the nominal lexeme dog , the exponence [z] of (3b)s
property set {plural} combines with dog s root [dg]. Te dis-
tinction between a words exponence and its root is, of course,
sometimes dif cult to draw, as in the realization of be, {1sg
[frst-singular] present indicative} as the portmanteau [m] .
Branches of Morphology
A languages morphology comprises two systems. Te infectional
system defnes the phonological realization of the grammati-
cal words in a lexemes paradigm; for instance, the infectional
system of English specifes that the third-singular (3sg) present
indicative form of the lexeme talk is talks . By contrast, the sys-
tem of word formation (better: lexeme formation ) defnes com-
plex lexemes in terms of simpler lexemes. Te latter system itself
(3) Grammatical words (4) Phonological words
a. talk, {3sg present
indicative}
a.(3a)s realization: [tks]
b. dog , {plural} b.(3b)s realization: [dgz]
c. good, {comparative} c.(3c)s realization: [b]
Morphology
517
general account of such essentially paradigmatic phenomena as
syncretism (the relation among paradigm cells that are identi-
cal in their realization), deponency (the realization of one cell by
means of morphology appropriate to a diferent cell), heterocli-
sis (the realization of distinct cells within a paradigm according
to contrasting conjugational/declensional patterns), and defec-
tiveness (the existence of unrealized cells within a paradigm). For
discussion of these phenomena in paradigm-based frameworks,
see Baerman ( 2004 ), Baerman, Brown, and Corbett ( 2005 ), and
Stump ( 2001 , 2006 ) .
Te alternative, paradigm-based approach is instantiated
by such realizational theories as A-morphous morphology
(Anderson 1992 ), network morphology (Corbett and Fraser 1993 ;
Brown and Hippisley in press), and paradigm function morphol-
ogy (Stump 2001 ). Tese theories take paradigms rather than
morphemes as the primary object of morphological inquiry and
formulate morphology as an autonomous grammatical compo-
nent. Despite diferences of detail, they are alike in assuming that
a lexeme has a paradigm of grammatical words a set of pairings
such as those in (3) whose phonological realization is deter-
mined by a system of deductive rules, for example, the rule of
exponence in (5a) and the rule of referral in (5b).
(5) a. Where lexeme L has root R, L, {fnite past } is realized
as R ed .
b. L , {past participle} has the same realization as L , {fnite
past }.
By (5a), the lexeme walk has the past tense form walked ; by
(5b), this lexeme also has walked as its past participle. A cen-
tral assumption in paradigm-based theories is that rules act as
defaults and are therefore subject to override; in the infection
of verbal lexemes such as sing , the rules in (5) are overridden
by those in (6). An important concern in such theories is that of
establishing general principles regulating the default/override
relations among rules of morphology. In network morphology,
these relations are regulated by their position in default-inher-
itance hierarchies; in paradigm function morphology, they are
regulated by the Pinian determinism hypothesis (Stump 2001 ,
23), according to which Rule A overrides Rule B if and only if A is
narrower in application than B.
(6) Where L belongs to the sing class and has root R,
a. L , {fnite past } is realized through the substitution of
[] for [] in R.
b. L , {past participle} is realized through the substitution of
[] for [] in R.
Because they defne complex words by means of deductive
rules such as those in (5)/(6), paradigm-based theories aford
a parsimonious account of interactions between concatenative
and nonconcatenative morphology: Te fact that (5a) fails to
apply in the defnition of sing s past tense form can be directly
attributed to the override relation between (6a) and (5a) .
In DM, by contrast, the absence of -ed in sang must instead be
attributed to an overriding, phonologically empty suf x whose
presence triggers a rule of [] [] ablaut. Tis account of the
sing/sang (* singed ) alternation implies a parallel account of sing/
sung (* have singed ), mouse/mice ( *mouses ), thief/thieve (* thiefze ,
cf. burglarize ), and other alternations: In each case, a default af x
kulu should alone suf ce to mark this form for class 7 agree-
ment; the appearance of the additional prefx ca- not only seems
unnecessary but actually violates the anti-redundancy principle
(Kiparsky 1982 ; 136 f) purported to prevent the suf xation of plu-
ral -s to English men .
Te alternative to an incrementalist theory is a realizational
theory, according to which a words content determines its mor-
phological form (Matthews 1972 ; Zwicky 1985 ; Anderson 1992 ).
In a realizational theory, the paradigm of the verbal lexeme talk
includes the grammatical word in (3a), and the phonological
word that realizes this grammatical word arises from talk s root
through the application of any rules associated with the mor-
phosyntactic property set in (3a); there is one such rule, which
realizes the property set {3sg present indicative} through the suf-
fxation of s.
In a realizational theory, the fact that a words form may
underdetermine its content is unproblematic, since content is
not deduced from form in any event. Tus, the fact that the frst-
singular aorist form krd-o-x I stole in Bulgarian has no expo-
nent of frst-singular subject agreement is simply the efect of a
kind of poverty in the languages verb morpology: It happens not
to have any means of expressing the property 1sg in the reali-
zation of the grammatical word steal , {1sg aorist}. Syncretism
is likewise unproblematic: One need only assume that the reali-
zation of one word in a lexemes paradigm may pattern after the
realization of a diferent word in that paradigm. Rules of referral
(Zwicky 1985 ; Stump 1993 ) express this kind of relation between
cells in a paradigm; thus, in Sanskrit, a rule of referral specifes
that the realization of a neuter nouns nominative singular cell is
the same as that of its accusative singular cell. Finally, extended
exponence is unproblematic in a realizational theory; in the case
of Nyanja ca-ci-kulu large [class 7], one need only assume that
more than one rule of prefxation participates in the realization
of the grammatical word large , {class 7} .
Current Theories of Morphology
Two approaches to morphology dominate the theoretical land-
scape: the morpheme-based approach and the paradigm-based
approach. Distributed morphology (DM) is the main embodi-
ment of the morpheme-based approach (Halle and Marantz
1993 ). DM maintains the structuralist focus on morphemes as
the central unit of morphological analysis, but difers from ear-
lier morpheme-based approaches in its assumption that mor-
phemes are inserted into abstract grammatical structures in a
realizational fashion. (Here and throughout, I use morpheme in
the Bloomfeldian sense of minimal form-meaning pairing.)
Te verb in Tey talked instantiates the abstract grammatical
structure V- past-pl through the insertion of the verbal mor-
pheme talk and the past tense morpheme -ed ; the property of
third-plural agreement goes unrealized because there is no non-
zero morpheme available to realize it. (DM therefore accom-
modates cases of underdetermination such as that of krd-o-x I
stole.) From earlier incrementalist approaches, DM inherits the
assumption that morphological structures such as V-past-pl
are defned by rules of syntax. Tis assumption presents prob-
lems that have never been convincingly resolved in the some-
what hermetic DM literature: In rejecting rules that (like rules
of referral) are defned over paradigms, DM is left without any
Morphology
518
words are invariant in their semantic interpretation; yet there
are instances in which the semantics associated with a particular
morphosyntactic property is sensitive to its paradigmatic context
(Stump 2007) .
A fnal area of current interest is that of implicative theories
of morphology. Like realizational theories, implicative theories
depend on the postulation of paradigms, but unlike them, they
assume that the forms realizing a paradigms cells are defned
by implicative relations among these cells (Blevins 2005 , 2006 ).
Tus, in an implicative theory, certain words in a paradigm have
a privileged status because they serve as the basis for deducing
the paradigms other words. If a small number of such privi-
leged forms uniquely determine the entire paradigm (as the
forms laud , laudre , laudv , and laudtum suf ce to deter-
mine the paradigm of praise in Latin), they may be charac-
terized as principal parts ; but even words that arent principal
parts may carry specifc implications for the formation of cer-
tain other members of their paradigm. Reference to these rela-
tions among a paradigms cells seems central for an account
of the processes of morphological deduction upon which the
acquisition and use of language depend; moreover, implicative
relations among the cells in lexemes paradigms are a signif-
cant domain of typological contrast among languages (Finkel
and Stump 2009 ) .
Gregory Stump
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ackerman , Farrell , and Gregory Stump . 2004 . Paradigms and periphras-
tic expression: A study in realization-based lexicalism. In Projecting
Morphology , ed. Louisa Sadler and Andrew Spencer , 11157. Stanford,
CA : CSLI Publications .
Anderson , Stephen R. 1992 . A-morphous Morphology . Cambridge :
Cambridge University Press .
Aronof , Mark. 1994 . Morphology by Itself: Stems and Infectional Classes .
Cambridge, MA : MIT Press .
Baerman , Matthew . 2004 . Directionality and (un)natural classes in syn-
cretism . Language 80 : 807 27.
Baerman , Matthew , Dunstan Brown , and Greville G. Corbett .
2005 . Te Syntax-Morphology Interface: A Study of Syncretism.
Cambridge : Cambridge University Press .
Bauer , Laurie . 1997 . Derivational paradigms. In Yearbook of
Morphology 1996 , ed. Geert Booij and J. van Marle , 24356. Dordrecht,
the Netherlands: Kluwer .
Blevins , James P. 2005 . Word-based declensions in Estonian. In
Yearbook of Morphology 2005 , ed. Geert Booij and J. van Marle , 125.
Dordrecht, the Netherlands : Springer .
. 2006 . Word-based morphology . Journal of Linguistics
42 : 531 73.
Booij , Geert. 1997 . Autonomous morphology and paradigmatic rela-
tions. In Yearbook of Morphology 1996 , ed. Geert Booij and J. van
Marle , 3553. Dordrecht, the Netherlands: Kluwer .
. 2000 . Infection and derivation. In Morphology: An International
Handbook on Infection and WordFormation , ed. Geert Booij ,
C. Lehmann , and J. Mugdan , 3609. Berlin : Walter de Gruyter .
Brjars , Kersti, Nigel Vincent , and Carol Chapman . 1997 . Paradigms,
periphrases and pronominal infection: A feature-based account. In
Yearbook of Morphology 1996 , ed. Geert Booij and J. van Marle , 15580.
Dordrecht, the Netherlands : Kluwer .
Brown , D . and A. Hippisley . Network Morphology . Cambridge : Cambridge
University Press. In press .
must be seen as being overridden by a null af x, which by stip-
ulation triggers a rule of internal modifcation. What emerges is
a widely recurrent coincidence that is never explained: Again
and again, a zero af x is stipulated as the unoverridden over-
ride among a set of competing morphemes; over and over, this
unoverridden override by stipulation triggers a rule of internal
modifcation. Tere is, of course, no overt class of phonologically
identical af xes in any language that ever shows the kind of syn-
tagmatic and paradigmatic distribution that DM must stipulate
for the artifactual class of phonologically null af xes upon which
this approach depends .
Current Issues in Morphology
Te diferences between morpheme-based theories and par-
adigm-based theories have been most clearly articulated with
reference to infectional phenomena. But the question naturally
arises whether the principles of lexeme formation are morpheme
based or instead favor a paradigm-based approach. Signifcantly,
derivational morphology exhibits the same sort of default/over-
ride relations as infectional morphology: In much the same way
as the lexeme sing possesses an infectional paradigm in which
the appearance of the past tense form sang blocks that of * singed ,
the lexeme strong seems to possess a derivational paradigm
in which the appearance of the nominal derivative strength
blocks that of * strongness . For discussion of the evidence for
derivational paradigms, see Bauer ( 1997 ), Booij ( 1997 ).
Te nature of the interface between morphology and syntax
also requires further scrutiny. One question subsumed by this
broad issue is whether a languages periphrases are defned by its
morphology or by its syntax. A periphrase is a multiword realiza-
tion for a grammatical word; thus, while the grammatical word
smart , {comparative} has the synthetic realization smarter ,
the grammatical word intelligent , {comparative} has the
periphrastic realization more intelligent (* intelligenter ). For dis-
cussion of the evidence in favor of a morphological approach to
periphrasis, see Kersti Brjars, Vincent, and Chapman ( 1997 ) and
Ackerman and Stump ( 2004 ); the grammatical consequences of
this conclusion remain to be worked out in detail .
Another controversial aspect of the morphology/syntax inter-
face is the phenomenon of clisis. Because their morphology
resembles that of af xes while their syntax is wordlike, clitics
raise very specifc questions about the division of labor between
the components of morphology and syntax. Although recent
years have seen a vast amount of research into the properties of
clitics, there is, as yet, little consensus as regards their precise
theoretical status. Particularly urgent are the need to understand
the diferences between clitics and af xes (Zwicky and Pullum
1983 ) and the need to reconcile these diferences with the com-
plex interactions between clisis and af xation (Spencer and Lus
2005 ) .
Te principles of the morphology/semantics interface also
urgently require clarifcation. Phenomena such as syncretism,
deponency, extended exponence, and morphological underde-
termination are apparently incompatible with the assumption
(characteristic of morpheme-based theories) that a words mor-
phology is isomorphic to its semantic structure. Research in par-
adigm-based theories has tended to assume (often tacitly) that
the morphosyntactic properties associated with grammatical
Morphology
519
early childhood. Te acquisition of derivational morphology, on
the other hand, is concerned with the formation of new words and
is thus related to school-age language learning and reading .
Jean Berkos classic 1958 study found that children fve to
seven years of age are able to apply both infectional and deri-
vational suf xes to novel stems (e.g., the plural wugs from wug ,
or the adjective quirky from the noun quirks ). Tese results were
interpreted at the time as evidence against the predictions from
the prevailing theories of Behaviorism (Skinner 1957 ) and sup-
ported the cognitive revolution in psychology and linguistics.
Tis study has been replicated with children as young as two
in English and other languages (e.g., Kopcke 1998 ; Akhtar and
Tomasello 1997 ). Together with work examining childrens natu-
ral productions of morphology, this experimental work has been
taken as evidence that young children use morphological rules
to infect and to form parts of speech .
Infectional Morphology
FIRST LANGUAGE. One major set of works investigating the
acquisition of morphemes examined the order of acquisition of
infectional morphemes (Cazden 1968 ; Brown 1973 ; de Villiers
and de Villiers 1973 ), focusing on English. Tis work found a
consistent (though not identical) order of acquisition among
children. Cross-linguistic work demonstrated that there was no
universal order of morpheme acquisition between languages,
that the order and speed of acquisition depends on the target
language and the morphemes themselves (Slobin 1985 ; Clark
1998 ). Work on other languages has also found consistent, but
not identical, orders within a language. Several factors appear to
infuence the order of acquisition, including perceptual salience,
complexity of the morpheme either semantically (how many
concepts it encodes) or formally (how variable the af x is, how
many parts it contains), and frequency in the input. Eve V. Clark
( 1993 ) suggested that several principles transparency (how
easily the meaning is derived from the parts), simplicity (how
variable the forms are), and productivity govern the order of
acquisition of both infectional and derivational morphemes.
Morphemes that are consistent in form (have few allomorphs)
and semantically encode a single feature, such as plural or pro-
gressive -ing , tend to be acquired earlier than morphemes that
show more allophonic variation, such as with regular past tense
- ed , and/or encode multiple features, such as third person
singular -s (Brown 1973 ; Clark 1993 ). Tis preference to encode
single features/forms with a single morpheme holds across lan-
guages and even for children acquiring more than one language.
For example, Melanija Mikes ( 1967 ) discussed the acquisition of
locatives for bilingual Hungarian-Serbo-Croatian children. Te
Hungarian locative suf x, which is relatively transparent, was
acquired earlier than the semantically equivalent structure in
Serbo-Croatian, which required locative prepositions + agree-
ment, which varied by gender.
Infected forms appear early, from the earliest word use, espe-
cially in highly or consistently infected languages (Slobin 1985 ).
Languages such as English, with fewer infected forms, have bare
forms appearing frst and infected forms appearing later, often
concurrently with frst word combinations. In English, childrens
production of infectional morphology begins with an initial
period of some variability, gradually becoming more consistent
Corbett , Greville G. , and Norman M. Fraser . 1993 . Network morphol-
ogy: A DATR account of Russian nominal infection . Journal of
Linguistics 29 : 113 42.
Embick , David , and Rolf Noyer . 2007 . Distributed morphology and the
syntax-morphology interface. In Te Oxford Handbook of Linguistic
Interfaces , ed. G. Ramchand and C. Reiss , 289324. Oxford : Oxford
University Press .
Finkel , Raphael , and Gregory Stump . 2009 . Principal parts and degrees
of paradigmatic transparency. In Analogy in Grammar, ed. J. Blevins
and J. Blevins, 1353. Oxford: Oxford University Press.
Halle , Morris , and Alec Marantz . 1993 . Distributed morphology and
the pieces of infection. In Te View from Building 20 , ed. K. Hale and
S. Keyser , 11176. Cambridge, MA : MIT Press .
Kiparsky , Paul . 1982 . From cyclic phonology to lexical phonology. In
Te Structure of Phonological Representations (Part I), ed. H. van der
Hulst and N. Smith , 13175. Dordrecht, the Netherlands : Foris .
Lieber , Rochelle . 1992 . Deconstructing Morphology . Chicago : University
of Chicago Press .
Matthews , P. H. 1972 . Infectional Morphology . Cambridge : Cambridge
University Press .
. 1991 . Morphology. 2d ed. Cambridge : Cambridge University
Press .
Selkirk , Elisabeth O. 1982 . Te Syntax of Words . Cambridge, MA : MIT
Press .
Spencer , Andrew , and Ana Lus . 2005 . A paradigm function account of
mesoclisis in European Portuguese (EP). In Yearbook of Morphology
2004 , ed. Geert Booij and J. van Marle , 177228. Dordrecht, the
Netherlands : Springer .
Stump , Gregory T. 1993 . On rules of referral . Language 69 : 449 79.
. 2001 . Infectional Morphology . Cambridge : Cambridge University
Press .
. 2006 . Heteroclisis and paradigm linkage . Language 82 :
279 322 .
. 2007. A non-canonical pattern of deponency and its implica-
tions. In Deponency and Morphological Mismatches , ed. Matthew
Baerman , Greville G. Corbett, Dunstan Brown , and Andrew Hippisley ,
7196. Oxford : Oxford University Press .
Zwicky , Arnold M. 1985 . How to describe infection. In Proceedings
of the Eleventh Annual Meeting of the Berkeley Linguistics Society , ed.
M. Niepokuj , M. VanClay , V. Nikiforidou , and D. Feder , 37286.
Berkeley, CA : Berkeley Linguistics Society .
Zwicky , Arnold M. , and Geofrey K. Pullum . 1983 . Cliticization vs. infec-
tion: English nt . Language 59 : 502 13.
MORPHOLOGY, ACQUISITION OF
Te acquisition of morphology has played a central role in
exploring both the acquisition of syntax (see syntax, acquisi-
tion of) and lexical acquisition . Te study of acquisition
of morphology can be distinguished by research in frst language
acquisition (FLA) and second language acquisition (SLA)
and by research into infectional or derivational morphology.
Little research has attempted to link acquisition of infectional
and derivational morphology. Tere has been some infuence
of work in FLA on that in SLA, in particular in debates about
the order of acquisition of morphemes for second language
learners.
Te acquisition of infection is often examined as the acqui-
sition of morphosyntax, that is, structures that are governed by
both morphological and syntactic rules such as subject-verb
agreement. Te majority of literature focuses on children under
the age of seven, as infectional morphology is acquired during
Morphology, Acquisition of
520
SECOND LANGUAGE. Early research in second language acquisi-
tion of infectional morphology investigated whether there was
a consistent order of acquisition of morphemes. Initial studies
reported consistent orders of acquisition across both adult and
child second language learners of English (e.g., Dulay and Burt
1974 ). Later research, however, criticized these early studies for
their methodology, and found variability among learners of dif-
ferent backgrounds that seemed to belie a single order of acqui-
sition for second language learners (e.g., Hakuta 1974 ; Rosansky
1976 ). Te present consensus seems to be that morpheme lan-
guage studies provide strong evidence that ILs [interlanguages,
developing grammars] exhibit common accuracy/acquisition
orders (Larsen-Freeman and Long 1991 , 92). After this intense
period of debate about the acquisition order of morphemes, later
research in SLA has focused on one of several areas: research on
specifc structures such as tense or aspect, the role of context in
the acquisition of morphology, and the implications of missing
morphology for the developing grammars of second language
learners. Research has begun to look again at order of acquisition
in SLA, considering how models of acquisition and/or the func-
tor morphemes themselves can explain the order (Goldschneider
and DeKeyser 2001 ). Here again, the discussion parallels that of
frst language acquisition, with perceptual salience, morphopho-
nological regularity, semantic complexity, and frequency being
posited as factors contributing to the order of acquisition .
Derivational Morphology
FIRST LANGUAGE. Because derived forms are often used to fll
semantic gaps within a lexicon, the acquisition of derivational
morphology is studied within the domain of vocabulary learn-
ing. Derivational morphology is acquired somewhat later, and
with a less clear order of development, than infectional mor-
phology. Highly infected languages such as Turkish or Finnish
show evidence of early derivational morphology. Studies indi-
cate that the earliest derivations are zero-stem alternations (e.g.,
the noun knife used as a verb) and compounding (e.g., dog-book )
for languages where these processes are productive (Clark 1993 ).
Agentive suf xes (-er in particular) and diminutives ( dogg-y ) are
also acquired between ages two and three across many languages.
However, the bulk of derivational morphology is acquired during
middle childhood and adolescence (Tyler and Nagy 1989 ).
As with infectional morphology, the course of develop-
ment for derivational morphology varies from language to lan-
guage and depends on both the patterns within the language
and the properties of the derivations themselves. For example,
Hebrew-speaking children derive verbs from nouns as young as
age three, while derived nominals are acquired later, after age
eight and continuing into adolescence (Berman 2003 ; Ravid
and Avidor 1998 ). Awareness of and ability to decode deriva-
tional morphology continues to develop throughout the school
years. Comprehension of derivational morphology is correlated
with reading skills in a number of languages, including English,
Hebrew and Chinese (Tyler and Nagy 1990; Levin, Ravid, and
Rapaport 2001 ; Kuo and Anderson 2006 ). As with infectional
morphology, acquisition appears to be best explained by the
transparency of the morpheme, complexity (semantic and for-
mal), and productivity (Clark 1998 ). Later-learned morphemes
such as -tion have more allomorphs, make more changes to the
over time. Across languages, some morphemes (such as plurals)
are consistently produced accurately by the age of three or four,
whereas others, such as conditional marking of verbs, are often
not mastered until ages seven or eight. While little research has
been done on infectional prefxes, it has been suggested that
infectional prefxes are more dif cult and are acquired later
than suf xes (Slobin 1982 ; Clark 1998 ).
In all languages, children often go through a period during
which they overregularize irregular forms in their grammar,
for example, breaked or mans . Irregular forms are often acquired
early ( broke/men ) and then when regular forms are acquired,
overregularized forms coexist with irregular forms ( breaked/
broke ) for a time until the irregular forms and exceptions are
mastered (Marcus et al. 1992 ) .
Te majority of research on infectional morphology has
focused on the verbal and nominal domains, with relatively
little work on adjectives or adverbs. According to Clark ( 1998 ),
the earliest verb forms across languages tend to be imperative,
infnitive verb forms, and third person singular; singular forms
tend to be acquired before plural forms. Te characteristics of
individual languages determine which agreement markers
are learned earliest in any given language. In terms of tense
and aspect , present/nonpresent is the frst distinction children
make, and distinctions between past, present and future appear
to be in place by age three. In languages that distinguish aspect,
aspect also appears to be acquired around age three. However,
early aspect marking is also associated with the semantic char-
acteristics of the verb. For example, Li and Shirai (2000) argue
that early use of perfective aspect is more likely to occur with
telic (e.g., walked ) or resultative verbs (e.g., smashed ) than other
types of verbs. Compound or periphrastic tenses such as present
perfect are acquired later, and may not be in place until age fve
or older.
Within the nominal domain, number marking occurs early
in nouns and is one of the earliest nominal morphemes to be
seen. gender marking also appears early, just after the frst
nouns. Early gender/noun class marking appears to be based
primarily on the phonological shape of the noun and later
becomes associated with individual lexical items. Noun class
marking appears by age three, but adultlike acquisition of gen-
der or noun class marking, which requires attention to both
phonological form and semantics, does not appear until age
four or fve in many languages (Demuth 2003 ). In languages
with classifer systems, such Chinese, Japanese or Tai, general
classifer patterns appear early, and more fne-grained seman-
tic distinctions appear gradually. Case marking occurs early as
well, just after frst nouns are learned. In nominative-accusa-
tive languages, the frst distinction to be acquired appears to
be between nominative and accusative. Dative case is next,
followed by other oblique cases. Tere is also evidence from
ergative languages that the major case distinctions in these lan-
guages are acquired early as well. Languages in which case var-
ies by gender (e.g., Russian) take longer for contrasts to appear
than in languages that simply mark for case (e.g., Hungarian,
or Turkish).
Infectional morphology is generally mastered by the early
school years for all but the most infrequent or irregular construc-
tions across languages .
Morphology, Acquisition of
521
words, while regular forms are concatenated via rules (Pinker
and Ullman 2002 ). Te other approach, from those mainly work-
ing within connectionism , claims that learners are extracting
statistical regularities and patterns, without positing a mental
rule (e.g., Plunkett and Marchman 1993 ). Tese researchers sug-
gest that there is a single, associationist mechanism for forming
past tense for all verbs, regular or irregular, and thus the rep-
resentations of regular and irregular forms do not difer. While
rule-based accounts appear to have underestimated the human
ability to track statistical information about morphology, con-
nectionist models may not be able to generalize regular forms
based on the typical frequencies of those forms in the input
(Marcus 1995 ). Tus, the extent to which these networks can
model human language acquisition is unclear .
Summary
Te acquisition of morphology is central to both morphosyntac-
tic development and lexical development for frst and second
language learners. While the acquisition of infectional mor-
phology is largely complete for children by the time they enter
school, it is often problematic for adult learners of a language.
Derivational morphology, on the other hand, is an ongoing, life-
long process for both frst and second language learners. Little
research has attempted to link the two types of morphological
acquisition. More research on both derivational and infectional
morphology is needed on a wider variety of languages or varying
typologies, in particular on non-Indo-European languages. Tis
is particularly the case for SLA, where the bulk of the research
has focused on European languages .
Lynn Santelmann
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Akhtar , N. , and M. Tomasello . 1997 . Young childrens productivity
with word order and verb morphology . Developmental Psychology
33.6 : 952 65.
Berko , Jean . 1958 . Te childs learning of English morphology . Word
14 : 150 77.
Berman , Ruth A. 2003 . Childrens lexical innovations: Developmental
perspectives on Hebrew verb structure. In Language Processing and
Acquisition in Languages of Semitic, Root-Based, Morphology , ed.
J. Shimron. Amsterdam : John Benjamins .
Brown , Roger W. 1973 . A First Language: Te Early Stages .
Cambridge : Harvard University Press .
Cazden , Courtney B. 1968 . Te acquisition of noun and verb infections .
Child Development 39 : 433 48.
Clark , Eve V. 1993 . Te Lexicon in Acquisition . Cambridge : Cambridge
University Press .
. 1998 . Morphology in language acquisition. In Te Handbook of
Morphology , ed. A. Spencer and A. Zwicky . Oxford : Blackwell .
Demuth , K. 2003 . Te acquisition of the Bantu languages. In Te Bantu
Languages , ed. D. Nurse and G. Phillipson . Surrey : Curzon .
de Villiers , Jill G. , and Peter A. de Villiers . 1973 . A crosssectional study of
the acquisition of grammatical morphemes . Journal of Psycholinguistic
Research 2 : 267 78.
Dulay , Heidi, and Marina Burt. 1974 . Natural sequences in child second
language acquisition . Language Learning 24 : 37 53.
Goldschneider , Julie M. , and Robert M. DeKeyser . 2001 . Explaining
the natural order of L2 morpheme acquisition . Language Learning
51 : 1 50.
stem they attach to, encode more concepts, and/or are less pro-
ductive than early learned morphemes such as -er .
SECOND LANGUAGE. As with frst language (L1) acquisition, sec-
ond language acquisition of derivational morphology has been
studied together with general vocabulary growth and knowl-
edge of the lexical semantics of the language (Redouane 2004 ;
Montrul 2001 ; Lardiere 1997 ). Little research has examined this
area of second language acquisition in detail. Te research that
exists suggests that beginning learners tend to fll lexical gaps
with word formation strategies from their L1 or by extending
existing second language vocabulary. More advanced learn-
ers are more likely to use derivational morphology found in the
target language, although they are still likely to use non-target
forms. Evidence also suggests that derivational morphology pat-
terns that are substantially diferent from the L1 present consid-
erable problems to second language learners and are learned
only gradually through time .
Current Debates
Major areas of debate in the acquisition of morphology include
the status of missing infectional morphemes in learners gram-
mars, whether productive use of morphemes refects rules and
an innate capacity for rule formation, and whether regular and
irregular forms are acquired and represented similarly.
For frst language acquisition, one area of strong debate has
been the status of missing or omitted morphology by children.
Within a general nativist framework, explanations range from
maturation of grammatical structures, to prosodic and/or pho-
nological learning, to language-specifc lexical learning (e.g.,
Peters and Stromqvist 1996 ; Santelmann, Berk, and Lust 2000 ).
More empiricist approaches point to such issues as frequency
of morphemes, vocabulary size, and transparency of the mor-
phemes themselves to explain the course of acquisition (e.g.,
Rowland et al. 2003 ).
Parallel debates have taken place within the literature on
second language acquisition. Within the search for a consistent
order of morpheme acquisition, debates have focused on the
possible reasons for such an order: Is it due to an underlying uni-
versal grammar (Dulay and Burt 1974 ), characteristics of the
input, or general learning and characteristics of the morphemes
themselves (Goldschneider and DeKeyser 2001 ). In addition, the
issue concerning why adult learners, unlike child learners, con-
tinue to make mistakes with infectional morphology even after
years of exposure to a language has been a source of signifcant
debate. Explanations for this phenomenon range from a lack of
access to universal grammar for adult learners to issues with pro-
sodic diferences between languages, diferences in profciency,
or other factors .
Within the domains of both infectional and derivational
morphology, another major theoretical debate concerns the
nature of rules. Do children and adults learn rules, or do they
simply extract regularities from the speech stream? Tis debate
has been particularly strong in the area of regular versus irregu-
lar verbs. One side of the debate argues that children use rules for
regular verbs, and thus have distinct processes for forming and
representing past tense with regular versus irregular forms (e.g.
Marcus 1996 ). Tis view argues that irregular forms are stored as
Morphology, Acquisition of
522
Slobin , Dan I, ed. 1985 . Te Cross-Linguistic Study of Language Acquisition .
Hillsdale, NJ : Erlbaum .
Tyler , Andrea , and William Nagy . 1989 . Te acquisition of English
derivational morphology . Journal of Memory and Language
28.6 : 649 67.
MORPHOLOGY, EVOLUTION AND
For some aspects of language, possible evolutionary explana-
tions are not hard to imagine, even if establishing the truth of
any of them may be dif cult or impossible. For example, the
fact that utterances can be segmented into individual meaning-
ful elements ( words or morphemes ) has a clear functional
advantage in that combinations of these elements can be used
to express a huge range of complex meanings an advantage
exploitable in natural selection. Likewise, it is advantageous
to have a syntax, that is, a set of traf c rules for combining
words or morphemes into larger units that can be interpreted
reliably. Less immediately obvious but nevertheless vigorously
defended by some scholars in recent years is the possibility
that certain aspects of language are as they are for physical or
mathematical, rather than biological, reasons. Te hexagonal
shape of honeycomb cells is not due to a hexagonal-cell gene
inherited by bees but is a self-organizing outcome of cell con-
struction under particular spatial constraints. Conceivably,
nonbiological self-organizing factors infuence language,
too.
For the existence of morphology , however, no such expla-
nations seem immediately plausible. What functional advan-
tage is there in the availability of not one but two patterns of
grammatical organization for complex expressions: syntactic,
as in the French phrase tasse th (literally cup to tea) and
the English sentence Tey were being bitten , and nonsyn-
tactic, as in the English compound word teacup and the Latin
one-word sentence Mordebantur ? More puzzling still, what
functional advantage is there in the fact that the plural of pan is
pans while the plural of man is men or that the past tense forms
of bring , sting , and sing are brought , stung , and sang , respec-
tively? Te frst question concerns the relationship between
morphology and syntax, while the second concerns allomor-
phic variation. Wouldnt languages, in general, function better
if there were just one set of traf c rules, not two, to guide the
interpretation of complex expressions? And wouldnt English
function better if plurality and pastness were expressed in a
uniform fashion, as in the Newspeak of George Orwells 1984
or as in an artificial language such as Esperanto? Should
we then look for a physical or quasi-mathematical explanation
instead? Yet these phenomena do not display the sort of hon-
eycomb-like elegance that renders them obvious candidates
for that explanation.
For possible solutions to these puzzles, it is natural to con-
sult hypotheses specifcally concerning the evolutionary origin
of morphology as a component of grammar. Hypotheses of that
kind so far published are sketchy. Nevertheless, four trends can
be distinguished:
(a) an appeal to uninterpretable features in Noam
Chomskys minimalist syntax (one type of appeal to self-
organization);
Hakuta , Kenji . 1974 . A preliminary report on the development of gram-
matical morphemes in a Japanese girl learning English as a second lan-
guage . Working Papers on Bilingualism 3 : 18 38.
Kopcke , K. M. 1998 . Te acquisition of plural marking in English and
German revisited: Schemata versus rules . Journal of Child Language
25.2 : 293 319.
Kuo , Li-jen , and Richard C. Anderson . 2006 . Morphological awareness
and learning to read: A Cross-language perspective . Educational
Psychologist 41.3 : 161 80.
Lardiere , Donna . 1997 . On the transfer of morphological parameter
values in L2 acquisition . Proceedings of the Annual Boston University
Conference on Language Development, 1997 21.2 : 366 77.
Larsen-Freeman , Diane , and Michael H. Long . 1991 . An Introduction to
Second Language Acquisition Research . New York : Longman .
Levin , Iris, Dorit Ravid , and Sharon Rapaport . 2001 . Morphology and
spelling among Hebrew-speaking children: From kindergarten to frst
grade . Journal of Child Language 28.3 : 741 72.
Li , P. and Y. Shirai. 2000. Te Acquisition of Lexical and Grammatical
Aspect . Berlin: Mouton de Gruyter.
Marcus , Gary F. 1995 . Te acquisition of the English past tense in children
and multilayered connectionist networks . Cognition 56. 3: 271 9.
. 1996 . Why do children say breaked? Current Directions in
Psychological Science 5.3 : 81 5.
Marcus , Gary F., Steven Pinker, Michael Ullman, Michelle Hollander, T.
John Rosen , and Fei Xu . 1992 . Overregularization in language acqui-
sition. Monographs of the Society for Research in Child Development
57 . 4 : v 164.
Mikes , Melanija . 1967 . Acquisition des categories grammaticales dans
le langue de lenfant [Acquisition of grammatical categories in the lan-
guage of the child]. Enfance 3/4 : 289 98.
Montrul , Silvina . 2001 . Te acquisition of causative/inchoative verbs in
L2 Turkish . Language Acquisition 9 .1: 1 58.
Peters , Ann M. , and Sven Stromqvist . 1996 . Te role of prosody
in the acquisition of grammatical morphemes. In Signal to
Syntax: Bootstrapping from Speech to Grammar in Early Acquisition ,
ed. J. Morgan and K. Demuth . Mahwah, NJ : Lawrence Erlbaum .
Pinker , Steven , and M. T. Ullman . 2002 . Te past and future of the past
tense . Trends in Cognitive Sciences 6 : 456 63.
Plunkett , Kim , and Virginia A. Marchman . 1993 . From rote learning to
system building: Acquiring verb morphology in children and connec-
tionist nets . Cognition 48.1 : 21 69.
Ravid , Dorit , and Avraham Avidor . 1998 . Acquisition of derived nomi-
nals in Hebrew: Developmental and linguistic principles . Journal of
Child Language 25.2 : 229 66.
Redouane , Rabia . 2004 . Te acquisition of MSA word formation pro-
cesses: A case study of English-speaking L2 learners and native speak-
ers . ITL, Review of Applied Linguistics , 145/ 146 : 181 217.
Rosansky , E. 1976 . Methods and morphemes in second language acqui-
sition . Language Learning 26: 409 25.
Rowland , Caroline F., Julian M. Pine, Elena V. Lieven , and Anna L.
Teakston . 2003 . Determinants of acquisition order in wh-ques-
tions: Re-evaluating the role of caregiver speech . Journal of Child
Language 30.3 : 609 35.
Santelmann , Lynn, Stephanie Berk , and Barbara Lust . 2000 . Assessing
the strong continuity hypothesis in the development of English
infection: Arguments for the grammatical mapping paradigm .
Proceedings of the XIX West Coast Conference on Formal Linguistics
19 : 439 52.
Skinner , B. F. 1957 . Verbal Behavior . New York : AppletonCentury
Crofts .
Slobin , Dan I. 1982 . Universal and particular in the acquisition of lan-
guage. In Language Acquisition: Te State of the Art, ed. E. Wanner
and L. Gleitman , 12872. Cambridge: Cambridge University Press.
Morphology, Acquisition of Morphology, Evolution and
523
used by some minimalist theorists to explain the existence of
morphology .
Grammaticalization theory concerns itself with the process
whereby in language change , what were once free forms
with concrete, lexical meanings can change in three ways: gram-
matically, so as to become bound rather than free (as the free
form full has developed into a suf x in helpful ); semantically ,
so as to contribute grammatical rather than lexical information
(as the verb will in English has shifted from desire to future
tense); and phonologically, so as to merge with a neighboring
phonological word (as in Ill come , derived from I will come )
(Heine and Kuteva 2002 ). All three changes can be observed in
the history of Swedish, where what was once a free pronoun sik
meaning himself/herself has developed into a suf x -s with a
habitual passive meaning. Bernard Comrie ( 1992 ) has suggested
that not only individual af xes but also morphology overall origi-
nated this way. At an earlier, simpler stage of language, there was
syntax but no morphology. Subsequently, phonological reduc-
tion and meaning change in frequently occurring collocations
brought into being a new kind of structure, with bound items
alongside free ones, and phonologically reduced items alongside
phonologically full ones.
A drawback with this approach is that it privileges syntax over
morphology in an arbitrary way. Granted, all languages have
syntax while some languages today make little or no use of mor-
phology. Tat does not, however, constitute evidence that syntax
evolved earlier than morphology did. Implicitly, this approach
posits a sort of prehistoric linguistic Golden Age when forms
and meanings were neatly paired one-to-one, and when lan-
guage did indeed have only one set of grammatical traf c rules.
However, such a Golden Age would have no parallel elsewhere
in evolutionary biology; there is no reason to think that func-
tionality as an outcome of natural selection was more pervasive
at some time in the past than it is now .
Is there any reason to think, then, that the anomalies of mor-
phology have been around for as long as modern-style syntax
has, or even longer? Andrew Carstairs-McCarthy ( 2010 ) argues
just this. Individual meaningful units (morphemes), whether
spoken or signed, are not diamond-hard discrete entities, unaf-
fected in their shape by neighboring units. Tis would have been
just as true before fully modern syntax had evolved as subse-
quently. Tus, there would already then have been in existence
phonological processes that would give rise to allomorphy, and,
just as now, historical changes would sometimes have deprived
this allomorphy of its phonological motivation (just as the voic-
ing of the [v] in wives , plural as wife , now lacks the phonological
motivation it had in Old English and has thus acquired a gram-
matical function, as an exponent of plural). Let us assume
that this allomorphy was coupled with an expectation that for-
mal diferences should always be accompanied by diferences
in information content. One has already the seeds of a kind of
grammar in which the same item can be viewed as having more
than one form, provided that these forms are diferentiated
somehow. Te diferentiation could be semantic (e.g., rise versus
raise ), grammatical (e.g., sang versus sing ), or in terms of phono-
logical environment (e.g., in Italian udire to hear, the stem is
od- when stressed and ud- when unstressed). (Grammatical dif-
ferentiation in this hypothetical stage of development would ex
(b) the projection into prehistory of grammaticalization
processes such as are observed in historical linguistic
change;
(c) an appeal to phonological consequences of the fact
that the speech signal is continuous, not segmentable into dis-
crete chunks with clear boundaries (and gesture likewise, if
we suppose that language originated in that medium);
(d) a variant of (c) that also invokes the special circumstances
of our huntergatherer ancestors.
Chomskyan Minimalism explores rationales for apparent
imperfections in language. One such apparent imperfection
is syntactic displacement: for example, the kind of noun phrase
fronting exhibited in Beans I like and Who did you see? and per-
haps even in a simple clause such as John kissed Mary , if one
assumes that syntactic subjects originate internally to the verb
phrase. Such displacement may serve communicative purposes
( topicalization , for example). It still counts as a grammati-
cal imperfection, however, if there is nothing within grammar
itself to drive it. Tis is where morphology may come in (it is
claimed). Let us suppose that some constituents have fea-
tures that are uninterpretable and thus need to be erased
by moving those constituents to a location where these features
can be matched (Chomsky 2000 , 2004 ). So far as grammar is
concerned, this matching helps to ensure that all of the syntactic
requirements of the vocabulary items in the sentence are met,
while so far as purposes of language that lie outside grammar
are concerned (such as communication) , it may aid their
fulfllment by (for example) moving shared information to the
start of utterance. Te apparent imperfection thus disappears.
If, incidentally, some of the features that drive displacement
manifest themselves in overt morphology (for example, as case
infections), that is hardly surprising; it may facilitate the acqui-
sition of grammar, for example. Tus, the existence of morphol-
ogy helps to resolve tensions between the way that the grammar
is ideally structured and the extragrammatical uses to which
language is put.
Tis line of argument has at least three weaknesses, however.
Firstly, it says nothing about the allomorphy exhibited in pans
and men or in brought , fung , and sang . Secondly, it says noth-
ing about why derivational morphology exists (for example, why
we say writer and artist rather than, say, person write or person
art ). Tirdly, it relies too much on the intellectual appeal that
paradoxes can exert. Consider the orbits of the planets around
the sun. Tese orbits are not circular, which may be seen as an
imperfection, but the imperfection disappears in a paradoxical
yet satisfying way (one may think) in that, even though an orbit
is elliptical, the planets position on its orbit is correlated with
its speed, as Johannes Kepler demonstrated. But the enjoyment
of paradox can go too far. I carry a puncture repair kit when I go
cycling, which is an apparent imperfection because it adds to
the weight of my equipment. Am I then entitled to claim that this
imperfection disappears whenever I get a puncture because my
repair kit enables me to get on the road again? And can I even
argue that getting a puncture is paradoxically a positive event
because it justifes my carrying the repair kit? Flat tires thus con-
tribute to perfect cycling! Tis style of argument is strange, to
say the least. Yet it is uncomfortably close to a style of argument
Morphology, Evolution and
524
MORPHOLOGY, NEUROBIOLOGY OF
Te nature of word formation and word storage has long been
prominent in the disparate felds of linguistics, psychology, and
neurobiology, and recently the neurobiology of morphology has
emerged as its own distinct subfeld of study.
Two prevalent issues in formal linguistic and psycholinguistic
approaches to morphology are the distinctions drawn between
infectional versus derivational morphology and between regular
versus irregular morphology. Another line of inquiry has sought
to identify the formal structure of basic morphological represen-
tations (i.e., words, stems, af xes) and to determine the extent
to which complex words are either composed by a grammati-
cal process or stored as unanalyzed wholes. Te issue of com-
positionality interacts with infectional/derivational status and
with regularity; for example, various theorists have proposed
that irregular infection is not compositional (1a), that familiar
infected forms are not compositional (2a), or that derivational
morphology is not compositional (3a) in the way that they are
stored or processed.
For these issues, there is an immense and contradictory body
of linguistic evidence. As a consequence, some linguists have
looked to new methods, such as neuropsychology and neuroim-
aging, as alternative sources of evidence .
Psychological research in the lexicon and morphology has
largely focused on the roles of frequency, familiarity, and simi-
larity in word storage and identifcation. A critical fnding in
this tradition concerns the efect of lexical frequency. Words
that occur more frequently are more quickly and successfully
recalled in a wide range of experimental settings. Tis fnding
has enabled psychologists to pose deeper questions about the
nature of lexical representations in terms of which aspects of
those representations are crucial to the frequency efect. Te
fact that processing morphologically complex words may be
afected by the frequency of the entire form has led some to
argue that familiar af xed words, including regularly infected
forms like walked , are stored as wholes in the mental lexicon
(e.g., Baayen, Dijkstra and Schreuder 1997). In essence, this
whole-word approach to representation and processing treats
regular forms like walked and irregular forms like ran as equiv-
alent: Both are past tense forms, and, by hypothesis, access to
that infectional information is mediated in both cases by the
whole-word recognition.
Neurobiological methods frst made an impact on our under-
standing of morphology through studies of individuals with
acquired impairments to their morphological systems. Tese
studies typically tried to establish the loci of morphological func-
tions in the brain (in either anatomical structures or functional
architectures). Te progenitors of these studies are the works of
French neurologist Paul Pierre Broca and German neurologist
Karl Wernicke during the late nineteenth century. Broca reported
on a patient whose production abilities were severely impaired
hypothesi not involve syntax, but could conceivably involve sys-
tematic expression of categories such as number, tense, or def-
niteness.) And, where formal diferences involved extra segments
at the beginning or the end of an item, the seeds were sown for
what we now call af xes , arising by a process quite distinct from
grammaticalization .
A variant of Carstairs-McCarthys approach has been devel-
oped by Dieter Wunderlich ( 2006a , 2006b ), linking grammatical
evolution with cultural and economic change. Te sort of syntax
that many modern languages have, with lavish opportunities for
long-distance syntactic movement, would not have had signif-
cant evolutionary advantages (Wunderlich suggests) until after
the emergence of large speech communities whose members
did not all know one another, that is, until the Neolithic period.
Until then, he suggests, that is, as long as all humans were hunt-
ergatherers living in small groups, elaborate morphology would
have preponderated over syntax .
It will be seen that, as regards morphological evolution,
widely divergent suggestions have been made about the balance
between cultural and noncultural factors. Carstairs-McCarthys
and Chomskys approaches, though diferent in many ways,
agree in emphasizing noncultural reasons for the existence of
morphology as a component of grammar. For Comrie, on the
other hand, cultural change is at least as important as biological
or self-organizational factors, while Wunderlich revives the view
that fully modern syntax came late as a cultural by-product of
population expansion and the transition to agriculture, with the
added twist that an elaborate morphological component was
already in existence early. Time will tell which viewpoint pre-
vails or which combination of viewpoints .
Andrew Carstairs-McCarthy
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Carstairs-McCarthy , Andrew . 2010 . Te Evolution of Morphology. Oxford :
Oxford University Press .
Chomsky , No am. 2000. New Horizons in the Study of Language and
Mind . Cambridge : Cambridge University Press . Chapter 1 is especially
relevant.
. 2004 . Language and mind: Current thoughts on ancient prob-
lems. In Variations and Universals in Biolinguistics , ed. L. Jenkins ,
379405. Amsterdam : Elsevier .
Comrie , Bernard . 1992 . Before complexity. In Te Evolution of Human
Languages , ed. J. Hawkins and M. Gell-Mann , 193211. Reading,
MA : Addison-Wesley .
Heine , Bernd , and Tania Kuteva . 2002 . World Lexicon of
Grammaticalization . Cambridge : Cambridge University Press .
Hinzen , Wolfram . 2006 . Mind Design and Minimal Syntax . Oxford : Oxford
University Press . Tis book confronts frankly, from a Chomskyan per-
spective, some of the dif culties that morphology poses for the min-
imalist program.
Wunderlich , Dieter . 2006 a. What forced syntax to emerge? In Between
40 and 60 Puzzles for Krifka , ed. Hans-Martin Grtner , Sigrid Beck ,
Regine Eckardt , Renate Musan , and Barbara Stiebels . Berlin : Zentrum
fr Allgemeine Sprachwissenschaft . Available online at: http://www.
zas.gwz-berlin.de/fileadmin/material/40 60-puzzles-for-krifka/
index.html.
. 2006 b. Why is there morphology? Abstract of paper presented at
Workshop on Teoretical Morphology, Leipzig, June. Available online
at: http://www.uni-leipzig.de/ ~jungslav/rmag/Wunderlich.pdf.
Noncompositional Compositional
(1) a. ran = [ran] b. ran = [run]+[ past]
(2) a. walked = [walked] b. walked = [walk]+[ past]
(3) a. hopeless = [hopeless] b. hopeless = [hope]+[less]
Morphology, Evolution and Morphology, Neurobiology of
525
produced as * la mia studia [det.Fs my.Fs of ce.Fs]) and between
subjects and verbs ( io vivo solo [I live.1s alone.Ms] was produced
as * io vive solo [I live.3s alone]). In repetition tasks, FS was 98 per-
cent correct in repeating derived words with their derivational
morphology intact, but only 40 percent correct in repeating the
infection.
One issue not expressly examined in the case of this patient
was whether the dif culty that FS had in repeating infected
forms was modulated by the regularity of their morphology.
Te relevance of this point relates to the importance of identi-
fying the particular level of morphological representation that is
implicated in the defcit. In some cases of acquired morpholog-
ical defcit, performance on regular and irregular morphology
dissociates. For example, patient SJD presented with a defcit
that disrupted the production (but not the comprehension) of
regularly infected forms like walked (which was read as walk
and as walking on diferent occasions), as well as morphologi-
cally derived words like publisher (which she read as publishing )
(Badecker and Caramazza 1991 ). In comparison, that patients
performance on irregularly infected words was equal to her rel-
atively intact production of uninfected forms. A complementary
dissociation of regular versus irregular infection has also been
reported. For example, patient AW exhibited poor performance
for irregularly infected forms in comparison to nearly intact per-
formance with regularly infected words (Miozzo 2003 ; see also
cases reported in Laiacona and Caramazza 2004 ; Shapiro and
Caramazza 2003 ).
Some single-case studies have reported patients who present
with impaired comprehension and production for both regular
and irregular infection in both spoken and written modalities
though not always in equal proportion. Tis defcit has been
construed as resulting from an abstract, morphosyntactic level
of defcit (i.e., one where walked and ran are both represented as
morphologically complex) (Badecker 1997 ).
Most often, the method of neuropsychological studies is to
establish dissociations between distinct morphological subsys-
tems, but sometimes the content of patients errors themselves
provides insight into the nature of morphological grammar. For
example, patient SJDs af x selection errors were not always
grammatically licensed (e.g., she read poorest as poorless ,
along with an elaborative comment that indicated comprehen-
sion of the superlative af x: the most poorless Indians have
very little money ). Tese performance features suggest that the
mechanisms for producing productively af xed words exploit
compositional procedures (Badecker and Caramazza 1991 ) .
Hemodynamic Methods
In contrast to neuropsychological methods, hemodynamic
neuroimaging methods have made it possible to directly observe
areas of the brain involved in normal (intact) morphological
processing. Tese methods include positron emission tomography
(PET) and functional magnetic resonance imaging (fMRI). Tese
methods compare levels of blood fow and blood oxygenation in
diferent areas of the brain as subjects perform cognitive tasks.
One of the few PET studies focused narrowly on morphology
sought to identify a neural correlate of the use of overt infec-
tional verbal morphology in German (Gnther et al. 2001 ).
Tis study contrasted German verbs with and without overt
due to a lesion in the inferior frontal gyrus of the frontal lobe,
a lesion/defcit pattern now know as Brocas aphasia. Wernicke
reported that patients with lesions in the left posterior section
of the superior temporal gyrus sufered from a severe compre-
hension defcit; their speech was fuid and natural sounding but
their word selection seemed divorced from meaning. Numerous
researchers have since argued that lexical storage and retrieval
functions are located in wernickes area , and that morpho-
logical grammar functions are housed in brocas area .
A quite diferent approach to the study of morphology, but one
that is also grounded in neurobiology, is connectionist mod-
eling. Work in connectionist modeling has called into question
some of the most fundamental tenets of morphological theory.
D. E. Rumelhart and J. L. McClelland ( 1986 ) presented argu-
ments from modeling against the distinction between regular and
irregular morphology. M. S. Seidenberg and L. M. Gonnerman
( 2000 ) challenged the very existence of morphological represen-
tations, arguing instead that what linguists call morphemes are
merely the points of convergence between sound and meaning
codes, and are not a distinct type of entity of their own. Tese
studies have drawn ferce criticism, stimulated vigorous debate,
and played a major role in shifting the standards of mainstream
morphology to a consideration of brains and simulated brains as
viable data sources .
In each of the disciplines concerned with the neurobiology
of morphology, many central issues remain unresolved and
hotly debated. Furthermore, there had been little occasion until
recently for these methodologically distinct disciplines to com-
municate, despite their concern with fundamentally the same
topic. However, recent technological advances have provided
each of these disciplinary perspectives with new methods of
study, and thus new insights into perennial questions. Tese
advances, combined with shifting disciplinary boundaries, have
enabled the neurobiology of morphology to become a largely
coherent line of inquiry into the neural underpinnings of word
storage and processing. We may divide recent approaches into
neuropsychological, hemodynamic, and neurophysiological
methods.
Neuropsychological Methods
Patients with brain damage and resulting impairments have
long been a valuable source of evidence about the neurobiology
of morphology. Modern neuropsychological studies of morphol-
ogy evaluate patients who sufer impairments that selectively
afect (or selectively spare) morphological functions as the
result of brain damage. Typically, the rationale of these studies
is to identify dissociations between patterns of impaired and
preserved capacities in order to establish whether certain mor-
phological functions are distinct from other components of the
lexical system.
One such dissociation is a morphological defcit that disrupts
the processing of infectional morphology in the context of a
relatively spared ability to process derivational morphology. Te
Italian-speaking patient FS reported in Miceli and Caramazza
( 1988 ) presented with one of the clearest instances of this per-
formance pattern. In spontaneous speech, FS made frequent
errors of agreement between nouns and attributive modifers
(e.g., the target phrase il mio studio [det.Ms my.Ms of ce.Ms] was
Morphology, Neurobiology of
526
the orthographical and semantic levels (e.g., boil broil , screech-
scream ). Te former pairs exhibit a facilitatory priming efect on
the M350 latency, while the latter pairs do not. However, as in
many MEG studies, efects that were visible in the M350 peak
latency were obscured in the behavioral response latency, pre-
sumably by a diferent and opposite efect that arose later in the
time course of processing .
Many other neurophysical studies have found convergent evi-
dence that morphological constituents are actively recognized
in the early stages of lexical processing. Repetition priming has
been found to attenuate the N400 response component to iso-
lated words in lexical decision tasks (Rugg 1985 ). Tis efect on
ERP has also been observed with priming by morphological rela-
tives: Regularly infected primes elicit a weaker N400 response
for uninfected verb targets in comparison to unrelated primes,
whereas irregularly infected primes do not produce a compa-
rable reduction (Mnte et al. 1999 ). Tese contrasting priming
efects have been taken as evidence for morphological parsing
of regularly infected forms. Other studies have found evidence
that the recognition of regularly infected words is supported by
morphological decomposition in brain responses to morpho-
logically illegal combinations like bringed (Morris and Holcomb
2005 ; Lck, Hahne, and Clahsen 2006 ; see also McKinnon, Allen,
and Osterhout 2003 for evidence of bound-stem parsing) .
Evidence for decomposition is also observed in the efects
of regularity and lexical frequency on the P600 response to
infected words that are ungrammatical for their context. In
a study that manipulated lexical frequency, morphological
regularity, and grammatical ft, high-frequency irregularly
infected verbs in ungrammatical contexts (e.g., the boy couldnt
*ran / *walked fast enough ) showed an earlier onset of the P600
response than did low-frequency irregularly infected verbs,
in comparison to their grammatical counterparts (e.g., the boy
couldnt run / walk fast enough ); but the onset of this response
was unafected by lexical frequency for regularly infected verbs
(Allen, Badecker, and Osterhout 2003 ). Te pattern suggests
that for irregular verbs, surface frequency afects the speed with
which lexical recognition mechanisms can gain access to (and
exploit) infectional content, but that this is not so for regularly
infected forms .
Recent studies have used MEG to explore how morphology
shapes the recognition and interpretation of compounds and
morphologically derived words (Fiorentino and Poeppel 2007 ;
Pylkknen et al. 2004 ). Tese studies provide further support for
the view that the detection and exploitation of morphological
structure play a major part in the early and subsequent stages of
lexical recognition.
Neural methods provide a distinctive and potentially compel-
ling source of data for the study of morphology. Still, the set of
compelling neural studies of morphology remains quite small,
relative to other research methods, and the coherence among
these studies remains low. Tis is due in part to the limited avail-
ability of costly neuroimaging equipment and to the limited num-
ber of scholars who have expertise in both the technical issues of
morphology and the technical methods of cognitive neurosci-
ence. However, as the equipment proliferates and the methods
gain a stronger foothold in the feld, we can expect that more rig-
orous methodological conventions will develop, and that studies
infectional suf xes, revealing a diference in activation in and
around Brocas area. Te researchers interpret these results as
evidence that Brocas area subserves morphological and/or
morphosyntactic functions.
Increasingly, fMRI is the preferred method for hemodynamic
investigations of morphology. Numerous fMRI studies have
pursued a functional localization for morphology by contrast-
ing conditions with and without af xes, but they have often
returned inconclusive results (see, e.g., Davis, Meunier, and
Marslen-Wilson 2004 ).
A few fMRI studies of morphology and the lexicon, however,
have taken on more refned questions, producing more robust
results. One such study is the investigation by A. Beretta and col-
leagues ( 2003 ) of German regular and irregular infection. Tey
fnd signifcant overall diferences in neural activation between
the processing of regularly infected and irregularly infected
nouns and verbs. Tey interpret this as evidence that regular and
irregular morphological functions are subserved by distinct neu-
ral systems, though their fndings do not address where irregulars
are processed or where regulars are processed, if such distinct
locations were to exist .
Neurophysiological Methods
In order to examine how specifc types of morphologically com-
plex words are processed (e.g., regularly vs. irregularly infected
words) over the time course of lexical processing, researchers
have increasingly turned to neurophysical recording techniques
whose temporal resolution is well suited to the rapid changes in
brain response to linguistic materials. Tese imaging methods
include electroencephalography (EEG) also known as event-
related brain potentials (ERPs) which measures electrical cur-
rents caused by neural activity, and magnetoencephalography
(MEG), which measures the magnetic felds that result from this
neuroelectrical activity.
For the most part, neurophysical studies of morphology
exploit well-studied event-related response components under
a variety of stimulus conditions (including lexical priming ,
manipulations of lexical properties such as surface or stem fre-
quency, and contextual ft). In EEG/ERP studies, there are two
response components that have been exploited to some advan-
tage: the N400 a negative defection peaking around 400 ms
that increases in amplitude after a novel or unexpected lexi-
cal stimulus and the P600 a positive current shift following
syntactic anomalies (Kutas and Hilyard 1980 ; Osterhout and
Holcomb 1992 ; Osterhout and Nicol 1999 ). In MEG, most mor-
phology studies have focused on the M350 response compo-
nent believed to refect some of the currents underlying the
N400 ERP which peaks approximately 350 ms after the presen-
tation of a word stimulus and is sensitive to stimulus factors such
as lexical frequency (Embick et al. 2001 ) .
Several MEG studies have engaged the connectionist litera-
ture on morphology, addressing the issue of whether there exists
a distinctly morphological level of representation, one that dif-
fers from the representation of meaning and spoken/written
form. In a MEG priming study, L. Stockall and A. Marantz ( 2006 )
found that genuine morphological relatives (e.g., give gave ,
teach taught ) pattern diferently at the M350 response than
morphologically nonrelated word pairs that are merely similar at
Morphology, Neurobiology of
527
Pylkknen , L., S. Feintuch, E. Hopkins , and A. Marantz . 2004 . Neural
correlates of the efects of morphological family frequency and family
size: An MEG study . Cognition 91 : B35 45.
Pylkknen , L. , and A. Marantz . 2003 . Tracking the time course of word
recognition with MEG . Trends in Cognitive Science 7 : 187 9.
Pylkknen , L., A. Stringfellow , and A. Marantz . 2002 . Neuromagnetic
evidence for the timing of lexical activation: An MEG component sen-
sitive to phonotactic probability but not to neighborhood density .
Brain and Language 81 : 666 78.
Rugg , M. D. 1985 . Te efects of semantic priming and word repetition
on event-related potentials . Psychophysiology 22 : 642 7.
Rumelhart , D. E. , and J. L. McClelland . 1986 . On learning the past tenses
of English verbs. In Parallel Distributed Processing:Explorations in the
Microstructure of Cognition, ed. D. E. Rumelhart , J. L. McClelland, and
the PDP Research Group , 21671. Cambridge, MA : MIT Press .
Seidenberg , M. S. , and L. M. Gonnerman . 2000 . Explaining derivational
morphology as the convergence of codes . Trends in Cognitive Sciences
4: 353 61.
Shapiro , K. , and A. Caramazza . 2003 . Grammatical processing of nouns
and verbs in the left frontal cortex? Neuropsychologia 41 : 1189 98.
Stockall , L. , and A. Marantz. 2006 . A single route, full decomposition
model of morphological complexity . Mental Lexicon 1 : 85 123.
MORPHOLOGY, UNIVERSALS OF
Te topic of universals has not been as prominent in the area of
morphology as it has been in some other areas of linguistics.
Many linguists share the impression that morphology is pre-
dominantly a domain of the language particular, rather than the
general and universal. Whereas all languages compose words to
make sentences in one way or another ( syntax ), it is not cer-
tain that all languages compose morphemes to make words .
Te Chinese languages, for example, have almost no morphol-
ogy apart from the possibility of compounding two roots to make
a word.
While it is probably too strong to say that some languages have
no morphological system at all, at least it seems clear that there
are no universal morphological categories notions that are
expressed by morphological means in all languages. For exam-
ple, English requires that past tense be expressed as a suf x on
verbs ( stun vs. stunned ) and that plural number be expressed as
a suf x on nouns ( box vs. boxes ), but there are many languages
in which these notions are not expressed morphologically (e.g.,
Yoruba); they are either expressed by syntactic constructions or
not expressed at all. Moreover, morphology is a notorious reposi-
tory of many historical relics, irregularities, exceptions, and idio-
syncrasies. For example, the plural of box in English is boxes , but
the plural of ox is oxen ; similarly, the past tense of stun is stunned
but the past tense of run is ran . Irregularities of this sort are tol-
erated in morphology in a way that they may not be (or not as
much) in other linguistic domains. What universals we can hope
to fnd in morphology, then, are statistical and implicational
universals, rather than absolute universals (see absolute and
statistical universals ).
Despite the inherent noisiness of morphology, some uni-
versals of these sorts are discernable. Perhaps the best known
and best understood are universals of markedness . Tese uni-
versals have the form of statements saying that no language will
have an af x that expresses a marked category Y unless it also
has an af x that expresses a less marked category X within the
in this emerging area will have an even greater impact on our
understanding of language and morphology .
Ehren Reilly and William Badecker
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Allen , M ., and W. Badecker . 2000 . Morphology: Te internal struc-
ture of words. In What Defcits Reveal about the Human Mind/
Brain: Handbook of Cognitive Neuropsychology , ed. B. Rapp , 21132.
London : Psychology Press .
Allen , M., W. Badecker , and L. Osterhout . 2003 . Morphological analy-
sis in sentence processing: An ERP study . Language and Cognitive
Processes 18 : 405 30.
Baayen , R. H., T. Dijkstra, and R. Schreuder. 1997. Singulars and plurals
in Dutch: Evidence for a parallel dual-route model. Journal of Memory
and Language 37: 94117.
Badecker , W . 1997 . Levels of morphological defcit: Indications from
infectional regularity . Brain and Language 60 : 360 80.
Badecker , W. , and A. Caramazza . 1991 . Morphological composition in
the lexical output system . Cognitive Neuropsychology 8 : 335 67.
Beretta , A., C. Campbell, T. H. Carr, J. Huang, L. M. Schmitt,
K. Christianson , and Y. Cao . 2003 . An ER-fMRI investigation of morpho-
logical infection in German reveals that the brain makes a distinction
between regular and irregular forms . Brain and Language 85 : 67 92.
Davis , M., F. Meunier , and W. Marslen-Wilson . 2004 . Neural responses to
morphological, syntactic, and semantic properties of single words: An
fMRI study . Brain and Language 89 : 439 49.
Embick , D., M. Hackl, J. Schaefer, M. Kelepir , and A. Marant z. 2001. A
magnetoencephalographic component whose latency refects lexical
frequency . Cognitive Brain Research 10 : 345 8.
Fiorentino , R. , and D. Poeppel . 2007 . Compound words and structure in
the lexicon. Language and Cognitive Processes 22 .7: 9531000 .
Gnther , T., F. Longoni, O. Sabri, L. Sturz, K. Setani , and W. Huber .
2001 . PET study of basic syntax and verb morphology . NeuroImage
13 : 538 .
Kutas , M., and S. A. Hillyard . 1980 . Reading senseless sentences: Brain
potentials refect semantic incongruity . Science 207 : 203 5.
Laiacona , M. , and A. Caramazza . 2004 . Te noun/verb dissociation in
language production: Varieties of causes . Cognitive Neuropsychology
21 : 103 23.
Lck , M., A. Hahne , and H. Clahsen . 2006 . Brain potentials to morpholog-
ically complex words during listening . Brain Research 1077 : 144 52.
McKinnon , R., M. Allen , and L. Osterhout . 2003 . Morphological decom-
position involving non-productive morphemes: ERP evidence .
NeuroReport 14 : 883 6.
Miceli , G. , and A. Caramazza . 1988 . Dissociation of infectional and deri-
vational morphology . Brain and Language 35 : 24 65.
Miozzo , M. 2003 . On the processing of regular and irregular forms of verbs
and nouns: Evidence from neuropsychology . Cognition 87 : 101 27.
Morris , J. , and P. Holcomb . 2005 . Event-related potentials to violations
of infectional verb morphology in English . Cognitive Brain Research
25 : 963 81.
Mnte , T., S. Tessa, H. Clahsen, K. Schlitz , and M. Kutas . 1999 .
Decomposition of morphologically complex words in
English: Evidence from event-related brain potentials . Cognitive
Brain Research 7 : 241 53.
Osterhout , L. , and P. J. Holcomb . 1992 . Event-related brain potentials
elicited by syntactic anomaly . Journal of Memory and Language
31 : 785 806.
Osterhout , L. , and J. Nicol . 1999 . On the distinctiveness, independence,
and time course of brain responses to syntactic and semantic anomal-
ies . Language and Cognitive Processes 14 : 283 317.
Pinker , S. , and M. Ullman . 2002 . Te past-tense debate: Te past and
future of the past tense . Trends in Cognitive Science 6 : 456 63.
Morphology, Neurobiology of Morphology, Universals of
528
af x of the language is indexed as being able to appear (ideally)
in one and only one of these slots .
Another putative universal of morphology is that infectional
morphology can only appear outside of (further from the root
than) derivational morphology. Roughly speaking, derivational
morphology creates new words by adding af xes to existing
stems, whereas infectional morphology creates the forms of a
word that a specifc syntactic context might require. Examples
of derivational af xes in English include - ize ( crystal crystallize ),
- ship ( friend friendship ), - less ( care careless ), - able ( lift liftable ),
and - ing ( clip clipping ). Examples of infectional af xes in
English include the plural af x that attaches to nouns ( crystal -
crystals , friend friends ) and the past tense af x that attaches to
verbs ( lift lifted , clip clipped ). Now, there is no problem adding
infectional morphology to the output of derivational morphol-
ogy: Words like fossil-ize-d , friend-ship-s , and clip-ping-s are per-
fectly possible. But the reverse order is not allowed: Te process
of having a solution turn into more than one crystal is not to
* crystal-s-ize and the state of having many friends is not * friend-s-
ship , nor is a * clip-ped-ing something that was clipped out in the
past. Also bad are words like * lift-ed-able and * care-s-ful (having
many cares). A similar constraint says that infectional morphol-
ogy cannot be found inside a compound word: One can have
doghouse but not * dogshouse (a large house intended for more
than one dog); one can have a pickpocket but not a * pick-ed-
pocket (a thief who has already done his dirty work). One can cal-
culate what these words should mean, and in some cases one can
imagine uses for the word; nevertheless, the examples are at best
highly marked and unlikely to be used. Similar restrictions can
be observed in many other languages, although a limited range
of counterexamples has occasionally been pointed out. Tere are
also some unresolved questions about how exactly to defne the
diference between derivational and infectional morphology,
which need to be clarifed to make this generalization meaning-
ful and applicable in all cases (see, for example, Anderson 1982 ).
Nevertheless, there is little doubt that an important universal
characteristic of morphological systems lurks here .
Some fner-grained universal restrictions have been discov-
ered. One idea, supported in many studies, is that the order of
morphemes in a complex word refects the scope of those mor-
phemes the order in which they were composed for syntax and
semantics (Baker 1985 , 1988 ; Bybee 1985 ; Cinque 1999 ; Rice
2000 ). An example from derivational morphology is the follow-
ing pair from Quechua (Baker 1988 ):
(2) a. Mikhu-naya-chi-wa-n.
eat-want-make-1sO-3sS
It makes me feel like eating.
b. Mikhu-chi-naya-wa-n.
eat-make-want-1sO-3sS
I feel like making someone eat.
Te suf xes chi to make and naya to want can attach to a verb
stem in either order, but there is a systematic diference in mean-
ing. If make attaches before want, the combination means
to want to make someone eat, whereas if want attaches
before make, the combination means to make someone want
to eat. Te order in which the af xes attach in Quechua matches
the order in which the words are combined syntactically in the
same semantic domain. For example, many languages (includ-
ing English) have a plural form for nouns but no dual form that
means exactly two, whereas there are few or no languages that
have morphological marking for the category dual without also
marking the category plural. Similarly, verbs in a given language
do not express the distinction between inclusive frst person
plural (we including you) and exclusive frst person plural (we
not including you) without also expressing the more basic dis-
tinction between frst person and second person (we versus
you). Along the same lines, languages do not have special af xes
for remote past and remote future without also having af xes
for simple past and simple future, and if a language makes any
aspect distinctions in its verbal morphology, it will make a dis-
tinction between imperfective aspect and perfective aspect.
A plausible reason why universals of this sort hold has to do
with the logic of features. It is assumed that more marked, seman-
tically complex categories are built up out of simpler categories.
For example, the category dual shares a semantic feature [Group]
with the category plural but adds a feature such as [Minimal],
which it shares with singular (Harley and Ritter 2002 ). It stands
to reason, then, that a language will not have morphemes that
realize a more complex feature bundle like [Group, Minimal]
(dual) without also having morphemes that realize the simpler
feature bundles [Group] (plural) and [Minimal] (singular) that
this bundle properly contains. It seems likely that this vision can
be extended to the full range of nominal and verbal infectional
categories, although many details remain to be worked out .
One of the most general universals of morphology is that the
order of morphemes in a complex word is almost always rigidly
fxed. Almost all languages allow the words of a sentence to be
rearranged to some degree for stylistic or pragmatic efect.
Tis is particularly true of a language such as Mohawk, in which
any word order is usually possible. In contrast, no language
allows the morphemes in a complex word to be freely rearranged
in this way. For example, the following Mohawk word consists
of 11 distinct morphemes, but any other ordering of these mor-
phemes is ungrammatical:
(1) Wa-sha-ko-t-yat-awi-tsher-ahetkv-t-v
FACT-he-her-self-body-put.on-NOML-be.ugly-make-for-
PUNC
He made the thing you put on the torso [i.e., a shirt or dress]
ugly for her.
Tere are occasional examples that might seem like exceptions to
this rule, when two af xes can come in diferent orders to express
diferent semantic scopes (as will be seen), or when a morpheme
is displaced from its expected position in order to respect con-
ditions of phonological well-formedness. But even such devia-
tions, which are strongly motivated by semantic or phonological
concerns, stand out as being rather unusual. A striking univer-
sal of morphology, then, is that morpheme order is fxed for a
language, and is not permitted to vary for pragmatic or stylistic
reasons. structuralist linguists and descriptive linguists
commonly capitalize on this property of morphology by using
the device of a template to describe the morphological structure
of words in a given language: A set number of morphological slots
(position classes) are identifed for each word class , and every
Morphology, Universals of
529
Cinque ( 1999 ) presents a more fne-grained approach of this
kind, in which some 30 distinct infectional categories are identi-
fed, each of which is shown to attach to a verb in a set order rel-
ative to all the others. (It is possible that the ban on infectional
morphology coming inside of derivational morphology is a spe-
cial case of this mirror principle/relevance principle, though that
is not obvious in all cases.)
In some of the more recent literature, however, there have
been hints that the fxedness of morpheme order might be even
more restricted than one would expect, given considerations of
semantic compositionality alone. For example, Larry Hyman
( 2003 ) discusses examples like (5) from Chichewa:
(5) Alenj a-ku-tks-its-il-a mkz mthko. (*a-ku-tks-il-its-a)
Hunters 3.PL-PROG-stir-make-APPL-FV woman spoon
Te hunters are making the woman stir with a spoon.
At issue here are the suf xes its causative and il applicative,
which (in this case) adds the meaning of doing an action with
a particular instrument. From the semantic point of view, one
would think that these af xes should be able to attach to the verb
in either order, giving diferent compositional meanings. One
could start with the base verb stir, add the applicative af x il to
get stir with a spoon, and then add the causative to get make
someone [stir with a spoon]. Alternatively, one could add the
causative af x to the base verb frst to get a causative action
make someone stir and then add the applicative af x to create
[make someone stir] with a spoon. In the frst case, it would be
the woman who is using the spoon to stir; in the second case, it
would be the hunters who are using the spoon to impose their
will on the woman. Yet only the second af x order is possible,
and this form is ambiguous (or perhaps vague ) concerning
the two imaginable meanings.
Hyman himself points to a historical explanation of the
absence of a second form in (5). He shows that the same restricted
ordering holds true for a wide range of Bantu languages spoken
in sub-Saharan Africa and that it is a special case of a more far-
reaching template, which stipulates not only that applicative
must follow causative but also that the reciprocal suf x can only
follow both of these, and the passive suf x can only follow all of
the others. He claims that this particular template was inherited
from Proto-Bantu by most of the descendant languages. But this
sort of historical explanation might not be general enough. First,
it begs the question of why the relevant af xes had to attach in
this particular order in the ancestor language. Second, it turns
out that most non-Bantu languages also allow the causative af x
to attach before applicative but not vice versa as well. Example
(6) shows that this is true for classical Nahuatl (spoken in Mexico
[Launey 1981 , 197]); it also holds for Mohawk (northeastern
United States), Hiaki (southwestern United States), Shipibo
(Peru), Mapudungun (Chile), and many others.
(6) Ti-ne:ch-in-tlacua-l-ti:-li-a in no-pil-hua:ntoto:n.
2sS-1sO-PL-eat-caus-appl the my-chidren
You made my children eat for me. (*Ti-ne:ch-in-tlacua-li-tia)
Something more general seems to be at work here.
In a similar vein, Gabriella Caballero and her colleagues
( 2006 ) have recently argued that the morpheme order noun-
verb is universally preferred to the order verb-noun whenever a
corresponding English paraphrases, and this in turn refects how
the meanings are composed semantically in both languages. Not
all languages that have similar af xes allow both of the orders
shown in (2), but it is generally true that the orders that are used
correspond systematically to the order of interpretation for pur-
poses of syntax and semantics.
Compositional ordering efects of this kind are rather wide-
spread and apply to diferent types of morphology. One can see
something similar in English in the domain of compounding. For
example, ethics committee proposal refers to a proposal by or for
the ethics committee, whereas ethics proposal committee refers
to a committee in charge of formulating an ethics proposal. As in
Quechua, the diferent morpheme orders refect diferent orders
of semantic composition. Te only diference between the two
cases is that chi and naya are bound af xes that must attach to
verb roots, whereas proposal and committee are roots that can be
used as words in their own right in English. Te observation that
morpheme orders must directly refect the order of syntactic/
semantic composition is sometimes called the mirror principle
(Baker 1985 ); Bybee ( 1985 ) refers to a similar idea as the principle
of relevance .
Much the same constraint seems to hold of infectional mor-
phology as well, except that in this domain, there are very few
cases in which the order of morphemes can be reversed to give a
semantic efect. For example, Joseph Greenberg ( 1963 ) showed
that when both number marking and case marking are attached
to a noun root, the number marking almost always attaches
before the case marker does:
(3) a. adam-lar-a (man-PL-DAT) to the men (Turkish)
b. *adam-a-lar
Te reason is presumably because the plural operator is defned
semantically over the meaning of the noun itself, whereas the
function of the case marker is to relate the noun phrase as a
whole to the rest of the sentence in which it appears. Tus, in
(3a) the order of the af xes refects the natural order of the
semantic composition, just as in the examples in (2). In contrast,
(3b) is bad because the relevant semantic operators do not
combine that way. Te morphological universal in (3) can thus
be related to the fact that in languages like English, the plural
marker must attach directly to the noun, not to the prepositional
phrase that contains the noun ( to the boy+s , not * to+s the boy ) .
Joan Bybee ( 1985 ) applies the same kind of reasoning to verbs
(see also Bybee, Perkins, and Pagliuca 1994 ). When verbs bear
multiple infectional af xes, they almost always come in a fxed
order: An aspect marker attaches frst, a tense marker attaches
outside an aspect marker, mood markers attach outside both
tense and aspect, and subject agreement markers attach last of
all (though the position of agreement is a bit more variable than
the others). Example (4) thus shows a typical morpheme order;
other orders are rare or nonexistent.
(4) aku-wye-a-y-mi. (Mapudungun)
arrive-PERF(aspect)-FUT(tense)-IND(mood)-2sS(agreement)
You will have arrived.
Notice that the English auxiliary verbs appear in essentially the
same relative order, suggesting that this, too, can be attributed
to domain-general facts about semantic composition. Guglielmo
Morphology, Universals of
530
MOTIF
Motif is a unit of measurement and content analysis. It is applied
to expressive culture, especially literatures and certain branches
of the arts, such as painting, sculpture, and music. Te system-
atic application of the term was established in the early 1900s
through the work of the Finnish School and its attempts to use
an international folktales history (time) and dispersal in social
and physical space (geography) to reconstruct the tales orig-
inal form ( urform , archetype ), place of birth, and other related
matters of diffusion . Tis approach is also known as the his-
toric-geographic method for its reliance on objective verifable
criteria, rather than speculative hypotheses . For the Finnish
School, two key concepts became indispensable research instru-
ments: tale-type , a problematic term signifying a full folktale
known cross-culturally, and motif, a smaller unit designating a
detail contributing to the formation of the plot. Motif complex/
cluster/sequence and episode are other related measurement
units. Tough arising from the works of the Finnish School,
perceived by many as useful only in comparative studies, and
shackled by problems of name interpretation and linkage to the
currently unfashionable quest for origins, the usefulness of these
terms as tools of data identifcation and objective analysis tran-
scends these limitations (El-Shamy 1997 , 235) .
Te most salient attributes of a motif are its endurance (con-
tinuity in time) and recurrence within a community (continuity
in space). Continuity in time and space are basic requirements
for traditionality. In 1925, Arthur Christensen argued that a motif
persists in tradition according to a psychological law which is
not easily explicable. Tis empirical characteristic was left unex-
plained (Bdker 1965 , 2013). From the perspective of the bearer
of lore and the author or composer of elite art, certain traditional
themes possess cognitive salience (impressiveness) that make
them stand out and grab a persons attention. Tis salience (log-
ical or afective) may be due to frequency of occurrence (repeti-
tion), meaningfulness, structure, uniqueness, ego involvement,
and so on, properties that make such themes easily perceived,
learned, retained, and recalled (El-Shamy 1997 ) .
Te concept of motif is a close parallel to that of theme . Te
two terms are often used interchangeably. However, theme has
dominated in the study of elite literature, whereas motif has
been more common in the study of folklore. Research in literary
themes has commonly been pursued because of its interpret-
ive potentialities and its intrinsic congruency with the history
of ideas (Jost 1988 , xv). Additionally, literary authorities have
considered it an ef cient counteragent against primarily aes-
thetic movements, such as progressive Universalpoesie. It is
also argued that the motif is intellectual by nature. It expresses
a process of reasoning about mens conduct of life and, as a con-
sequence, does not concern itself with the analysis of individual
characters or extraordinary happenings (Jost 1988 , xvii) .
Te viewpoint outlined here addressing elite literature stands
at some variance with the folkloristic usage of the term motif and
the perceived scope of its applicability. Introducing his Motif-
Index , Stith Tompson explained that his system is built around
the interest of students of the traditional narrative and would
address a certain type of character, action, as well as atten-
dant circumstances of the action (1955, I.11). (See, respectively,
noun and a verb combine to form a single verb. Tis holds true
regardless of the specifc mode of combination, whether it is
the result of syntactic noun incorporation, productive morpho-
logical compounding, or idiosyncratic lexical combination. Te
order of morphemes in the Mohawk example in (7) is thus typical
in this respect:
(7) Wa-ke- nakt -a- hninu - (not: hninu-nakt buy-bed)
Fact-I-bed-0-buy-PUNC
I bought a bed.
Tis universal order is occasionally overridden by syntactic
ordering principles in particular languages (like Mapudungun).
But noun-verb order is always preferred by the morphology, and
it emerges more strongly in the more lexicalized, purely morpho-
logical constructions, where contamination from syntactic fac-
tors is minimal.
Taken together, studies like these hint that there might be a
universal morphological template, roughly of the form noun-
verb-causative-applicative-passive. Tis template appears to be
a force (though not an irresistible one) that is at work infuencing
the morpheme orders of all languages. Moreover, this order does
not seem to reduce to historical factors or semantic composition.
Why this should be is unknown at this point. However, it seems
likely that more morphological universals of this sort will be dis-
covered in the future as linguists recover from their impression
that morphology is primarily the domain of the idiosyncratic and
the language particular .
Mark C. Baker
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Stephen. 1982 . Wheres morphology ? Linguistic Inquiry
13 : 571 612.
Baker , Mark . 1985 . Te mirror principle and morphosyntactic explan-
ation . Linguistic Inquiry 16 : 373 415.
. 1988 . Incorporation: A Teory of Grammatical Function Changing .
Chicago : University of Chicago Press .
Bybee , Joan . 1985 . Morphology: A Study of the Relation Between Meaning
and Form . Amsterdam: John Benjamins .
Bybee , Joan, R. Perkins , and W. Pagliuca . 1994 . Te Evolution of
Grammar: Tense, Aspect, and Modality in the Languages of the World .
Chicago : University of Chicago Press .
Caballero , Gabriella, Michael Houser, Nicole Marcus, Teresa
McFarland, Anne Pycha, Maziar Toosarvandani, Suzanne Wilhite ,
and Johanna Nichols . 2006 . Nonsyntactic ordering efects in syn-
tactic noun incorporation. Manuscript, University of California at
Berkeley.
Cinque , Guglielmo . 1999 . Adverbs and Functional Heads: A Cross-
Linguistic Perspective . New York : Oxford .
Greenberg , Joseph . 1963 . Universals of Language . Cambridge, MA : MIT
Press .
Harley , Heidi, and Elizabeth Ritter . 2002 . A feature-geometric analysis of
person and number . Language 78 : 482 526.
Hyman , Larry . 2003 . Suf x ordering in Bantu: A morphocentric
approach. In Yearbook of Morphology 2002 , ed. Geert Booij and Jaap
van Marle , 24582. Dordrecht, the Netherlands : Kluwer Academic
Publishers .
Launey , Michel . 1981 . Introduction la Langue et la Littrature
Azteques . Vol. 1. Paris: LHarmattan.
Rice , Keren . 2000 . Morpheme Order and Semantic Scope . Cambridge:
Cambridge University Press .
Morphology, Universals of Motif
531
Dundes , Alan . 1964 . Te Morphology of North American Indian Folktales .
Folklore Fellows Communications, no. 195. Helsinki : Academia
Scientiarum Fennica .
El-Shamy , Hasan . 1995 . Folk Traditions of the Arab World: A Guide to
Motif Classifcation. 2 vols. Bloomington : Indiana University Press.
. 1997 . Psychologically-based Criteria for Classifcation by Motif
and Tale-Type . Journal of Folklore Research 34.3 : 233 43.
Garry , Jane , and Hasan El-Shamy , eds. 2005. Archetypes and Motifs in
Folklore and Literature: A Handbook , Armonk, NY : M. E. Sharpe .
Jost , Franois . 1988 . Introduction. In Dictionary of Literary Temes
and Motifs , ed. Jean-Charles Seigneuret , xv-xxiii. Westport,
CT : Greenwood .
Murdock , G. P., Clelland S. Ford , and Alfred E. Hudson . 1938 . Outline of
Cultural Mater ials . New Haven, CT : Institute of Human Relations, Yale
University .
Sperber , Hans , and Leo Spitzer . 1918 . Motiv und Wort, Studien zur
Literatur-und Sprachpsychologie . Leipzig : O. R. Reisland .
Tompson , Stith . 1955 8. MotifIndex of Folk Literature . 6 vols. Revised
ed. Bloomington : Indiana University Press .
MOVEMENT
Movement is an operation posited in theoretical syntax, in which
words, phrases, and perhaps also morphemes are relocated from
one part of a sentence to another. Movement is generally invoked
in cases in which a phrase has a combination of properties asso-
ciated with distinct positions in the sentence. Consider (1):
(1) Whom did you see?
Approaches to syntax that posit movement typically claim that
the word whom in (1) has moved from the end of the sentence,
where the direct object would ordinarily be, to the beginning.
Tis explains why, for instance, whom bears the accusative case;
whom has properties associated with direct objects, because it
has occupied the direct objects position.
One type of argument for movement tries to establish that
an apparently empty position must be flled, and that the logical
fller is a moved phrase. For example, a verb like fx must take a
direct object, which is typically in immediately postverbal posi-
tion in English:
(2) a. You fxed the car.
b.*You fxed.
c.*You fxed yesterday the car.
However, a question like (3a) is well formed (unlike (3b):
(3) a. What did you fx?
b. *When did you fx?
Te ill-formedness of (3b) shows that the required transitivity of
fx is not suspended in questions; fx must have a direct object in
(3a). Te contrast in (3) suggests that the direct object is what ,
since what is present just in the well-formed (3a). Example (2c)
shows that the direct object in English must typically be imme-
diately postverbal; we can preserve this generalization in (3a)
by positing movement of what from postverbal position to the
beginning of the sentence.
Other arguments for movement try to show that a moved
phrase has occupied positions in the sentence that it no longer
occupies. For instance, James McCloskey ( 2000 ) discusses a
for examples: W10, Kindness; T72.2.1, Prince marries scorn-
ful girl and punishes her; R216, Escape from ship while captors
quarrel).
Infuenced by anthropology, Tompsons system may be
compared to the analytical-classifcatory devices of culture ele-
ment , culture complex , and culture institution with culture ele-
ment being the smallest identifable component of culture. In
congruence with his companions in the Historic-Geographical
School, Tompson saw folk literature (especially narratives) as
analyzable in terms of motifs, episodes (or motif complexes/
sequences), and full narrative plots constituting tale-types. A
motif, though considerably more intricate, is comparable to
culture element; culture complex is comparable to episode; and
culture institution is comparable to tale-type. For Tompson,
motifs are those details out of which full-fedged narratives are
composed (1955, I.10).
Explaining the rationale for his motif system and its main
objective, Tompson declares that it emulates what the sci-
entists have done with the worldwide phenomena of biology
(1955, I.1011). In this respect, the underlying principle for
motif identifcation and indexing is comparable to that devised
by anthropologists at Yale for categorizing culture materials
in terms of 78 macrounits (1088) and 629 subdivisions thereof
used to establish Te Human Relations Area Files (HRAF);
these fles, begun almost contemporaneously with the frst pub-
lication in the 1930s of Tompsons Motif-Index , may be viewed
as an unprinted index. Twenty-three divisions make up the spec-
trum of sociocultural materials covered in Tompsons Motif-
Index , each treated in an independent chapter (e.g., B. ANIMALS;
C. TABU; F. MARVELS; X. HUMOR). Tese cardinal themes are
divided into 1,730 subdivisions (El-Shamy 1995 ).
Because Tompsons Motif-Index seeks global coverage,
numerous geographic regions and national entities did not
receive adequate attention. Consequently, signifcant felds of
human experience are missing or sketchily presented. Major
expansions are ofered in ensuing works, for example ( note : the
sign indicates addition to Tompsons system): F70, Ascent
to other planets (worlds) by space ship (fying saucer); J70,
Teaching (training) by cruel example; P610, Homosociality
[]; P770, Markets: buying, selling, trading; X580, Humor
concerning misers and miserliness (El-Shamy 1995 ; cf. Birkhan,
Lichtblau, and Tuczay 2005 ) .
An ofshoot of the motif system is the concept of motifeme,
a hybrid of folklore and linguistics modeled after morphological
analyses of folktales. In this system, an abstract unit of action or
state is labeled motifeme , and its manifestations are motifs. Te
variety in which a given motifeme is manifested is termed allo-
motif (Dundes 1964 ) .
Hasan El-Shamy
WORKS CITED AND SUGGESTED FURTHER READING
Birkhan , Helmut, Karin Lichtblau , and Christa Tuczay . 2005 . Motif-Index
of German Secular Narratives from the Beginning to 1400 . 6 vols. Berlin
and New York : W. de Gruyter .
Bdker , Laurits . 1965 . Folk Literature (Germanic) . Vol. 2 of International
Dictionary of Regional European Ethnology and Folklore .
Copenhagen : Rosenkilde and Bagger .
Motif Movement
532
(9) a. What did he say that he wanted all ?
b. What did he say all that he wanted?
c. What all did he say that he wanted?
Te examples in (9) difer with respect to how far all travels with
what : not at all (9a), only to the intermediate site (9b), or to the
beginning of the clause (9c).
Another controversial type of movement is involved in the
Chinese and Japanese wh-questions in (10a) and (10b), both of
which have the same meaning as the English (10c):
(10) a. John mai-le sheme ?
John buy Perf what
b. John-wa nani -o kaimasita ka?
John TOP what ACC bought Q
c. What did John buy?
In English, what is moved to the beginning of the clause, but in
Japanese and Chinese, the corresponding words can appear in
noninitial positions, just where they would be if they were not
wh-phrases ( wh-in-situ ). On one approach to these data, Chinese
sheme and Japanese nani , like their English translation what ,
undergo wh-movement to the beginning of the clause; unlike
what , however, these words move in a way that does not afect
where they are pronounced ( covert movement ) .
Some arguments for covert movement center on the inter-
action of wh-in-situ with established constraints on movement
(cf. Huang 1982 ; Richards 2009). For instance, wh-movement
in many languages is unable to pass out of an embedded inter-
rogative clause; we say that interrogative clauses are islands for
wh-movement. Tus, (11a) is well-formed, but (11b), with wh-
movement out of an interrogative clause, is not:
Japanese wh-in-situ exhibits a similar constraint; the sentences
in (11) have the Japanese translations in (12) with a similar con-
trast in grammaticality:
(12) a. Mary-wa John-ga nani-o katta to omoimasu ka?
Mary TOP John NOM what ACC bought that thinks Q
b. * Mary-wa [dare-ga nani-o katta ka] siritagattei-
masu ka?
Mary TOP who NOM what ACC bought Q wonders Q
Norvin Richards
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Fox , Danny . 2000 . Economy and Semantic Interpretation . Cambridge,
MA : MIT Press .
Huang , James . 1982 . Logical relations in Chinese and the theory of
grammar. Ph.D. diss., Massachusetts Institute of Technology.
McCloskey , James . 2000 . Quantifer foat and wh-movement in an Irish
English . Linguistic Inquiry 31 : 57 84.
Richards , Norvin . 2009 . Wh-questions. In Te Oxford Handbook
of Japanese Linguistics , ed. Shigeru Miyagawa and Mamoru Saito.
Oxford: Oxford University Press .
West Ulster dialect of English in which the questions in (4a) and
(4b) are synonymous:
(4) a. What all did you buy?
b. What did you buy all ?
c. *What did all you buy?
The position of all is not completely free in this dialect, as
the ill-formed (4c) shows. In fact, all can only appear in posi-
tions that what has occupied. In (4b), all appears in the direct
object position; this is the position that what occupies before
moving to the beginning of the sentence. Theories that posit
movement can account for the distribution of all in this dia-
lect; the phrase what all can leave the word all behind when
it moves.
Another argument for movement is based on the phenom-
enon of reconstruction , in which moved phrases are treated by
certain semantic dependencies as though they had not moved.
Space constraints prevent further discussion here (cf. Romero
1997 ; Fox 2000 ) .
Syntacticians distinguish several subtypes of movement. Te
examples thus far have all involved wh-movement , which forms
questions by moving certain phrases to the beginning of the
clause . Another type, head-movement , derives (5b) from (5a), via
movement of is :
A third type is sometimes called NP-movement ; examples
include movement of John in (6a) (compare the roughly syn-
onymous [6b]), and movement of Mary from object to subject
position in (7):
Other types of movement are more dif cult to detect. For
instance, much work argues that wh-movement in (8) is not a
single move from the end of the sentence to the beginning, but
stops in at least one intermediate landing site ( successive-cyclic
movement ):
One argument for this type of movement comes from West Ulster
English. We saw that this dialect allows all to be stranded in posi-
tions formerly occupied by a wh-moved phrase. Te all -strand-
ing facts for this dialect in (9) show that wh-movement has an
intermediate landing site:
(5) a. He is leaving.
b. Is he ___ leaving?
(6) a. John seems __ to be happy.
b. It seems that John is happy.
(7) Mary was promoted __.
(8) What did he say that he wanted __?
(11) a. What does Mary think that John bought __ ?
b. *What does Mary wonder [who bought __ ]?
Movement
533
yield, as Brown notes, analogies between music and language,
helpful and suggestive analogies, to be sure, but which do not
constitute arguments for a shared cognitive basis.
Musicalist Representation of Linguistic Structure
Te studies just alluded to are representative of recent scholar-
ship in that they attempt to ground subjective judgments with
respect to musical structure on the hard foundation provided by
linguistic science. While this has been the prevailing direction
of infuence, it has at times extended in the opposite direction.
Most notably, musical notation for several centuries constituted
the only efective means for visually representing the structure
of audible sound. Among the acoustical phenomena rendered
visible and thereby amenable to a structural analysis were speech
sounds of English carefully transcribed into a modifed form of
musical notation by Joshua Steele in his 1775 Essay Towards
Establishing the Melody and Measure of Speech .
In a recent review of the work, Jamie Kassler ( 2005 ) credits
Steele as being among the frst to identify linguistic supraseg-
mentals the tier of linguistic structure computed indepen-
dently of and mapped onto phonemic segments. tone , the
hierarchically related sequence of pitch locations assigned to
voiced segments, is one such suprasegmental and is relatively
naturally represented in musical notation. Steele also recog-
nized that unlike musical pitch, which tends to be discrete, the
target pitches of speech are consistently connected by contin-
uous glissandi or slides, represented in his scores by diagonal
line segments of various types attached to note stems, shown in
Figure 1.
Te other linguistic suprasegmental identifed by Steele,
stress , emerges somewhat obliquely from his transcriptions.
One of Steeles important insights was to have recognized that a
particular type of musical accent, the metrical accent, is associ-
ated with linguistic stress. Tus, for example, the initial beat of
the musical measure is metrically strong, and the most stressed
syllables of a text assigned to a tune (generally the stressed
syllables of polysyllabic words Peter , going , mistake , and coming
in Figure 1) are assigned to what he refers to as the ictus position .
Finally, and most signifcantly, Steele recognized that metrical
accent is not an objective feature of the musical event but is a
psychological attribute inherited from its temporal location. A
strong position will be perceived as such regardless of whether the
event occupying the position is objectively accented in the form
of higher pitch, amplitude, or length. Indeed, it may be heard as
strong even when it is vacant occupied by a rest. Metrical accent
Romero , Maribel . 1997 . Te correlation between scope reconstruc-
tion and connectivity efects. In Proceedings of WCCFL 16 , 35166.
Stanford, CA : CSLI.
MUSIC, LANGUAGE AND
Connections between music and language have been a peren-
nial concern of scholars, poets, music and literary theorists, and
musicians going back to antiquity. Te basis of this interest lies
in certain commonalities that are intuitively understood to lie at
the heart of the two capacities, but which become complex and
problematic when one attempts to elucidate their precise nature.
Confusion on the question of the interaction of music and lan-
guage is not surprising since the underlying basis of music and
language as independent objects has been poorly understood
until recently. Nor is it surprising, given traditional disciplin-
ary divisions, that scholars have tended to focus on developing
descriptive and explanatory frameworks (see descriptive,
observational, and explanatory adequacy ) for each,
which take for granted their status as independent rather
than common faculties. With the exception of Jean-Jacques
Rousseaus ([1763] 1997 ) Essai sur lorigine des langues, which
claims an ancestral proto-language from which language and
music both derive, most attempts at engaging the question have
stressed that music and language not only serve distinct ends,
one mainly aesthetic and the other mainly communicative, but
also access distinct underlying psychological means.
Te umbrella of cognitive science has provided the context
for a renewed discussion of some of these points of comparison.
Perhaps most striking is the reemergence of arguments for an
evolutionary precursor in the form of what Stephen Brown ( 2000 )
refers to as musilanguage whose essential characteristics are
identifed in Stephen Mithen ( 2005 ) by the acronym hmmmm
(holistic, manipulative, multimodal, musical, and mimetic).
Tese approaches are somewhat controversial not only in their
endorsement of what might be called a neo-Rousseauvian per-
spective but also in their assumption of a signifcant overlap in
some of the cognitive structures that underlie both music and
language. Te best-known explorations of the common ground,
Leonard Bernsteins Te Unanswered Question , Deryk Cookes
Te Language of Music , and Joseph Swains Musical Languages
have not suggested any specifc shared mechanisms. Rather,
they and others have applied certain aspects of the descriptive
apparatus and general methodologies of linguistic theory to
Figure 1.
Music, Language and
534
the same representation, namely, the grid, would emerge as the
optimal means for representing linguistics stress. Indeed, met-
rical stress theory, the dominant explanatory framework within
the generativist paradigm would be defned by the grid represen-
tation, one variant of which is shown in example 3 (from Halle
and Vergnaud 1987 ) (Figure 3).
A comparison of the grids in examples 2 and 3 reveals two
essential diferences between linguistic and musical structure.
First, while the stress grid projects syllables onto higher metri-
cal levels, the bottom level of the musical grid indicates not
actual musical events (i.e., notes) but, rather, temporal loca-
tions. As a consequence, empty metrical locations, such as those
in example 2, which are a necessary component of any reason-
able description of musical meter, are excluded from the stress
grid. Secondly, as Jackendof and Lerdahl ( 1983 ) show, musical
structure imposes strict requirements on the geometric form
that grids may assume, limited to what they refer to as a small
class of well-formed structures. In contrast, there are no a priori
constraints on the form taken by the stress grid. Te successive
positions projected onto line 1 of example 3 would be ruled out
as a potential metrical structure in music, where strong positions
need to be separated by at least one position on level 0. Tis vio-
lation of musicalist well-formedness does not, however, prevent
example 3 from accurately characterizing the pattern of second-
ary and primary stress for the word in question.
Te asymmetries in the two forms of representation are, it
would seem, necessary for a description of the output of each
system: As mentioned, the projection of musical meter requires
an underlying temporal periodicity that is neither intuitively
obvious nor empirically demonstrable in language except as a
statistical regularity (see Patel and Daniele 2003 for discussion).
In addition, the asymmetry refects essential diferences in the
character of the basic elements of musical versus linguistic struc-
ture. Te assignment of stress is a formal computation efected
on syllables from the rich phonemic inventory of particular
languages. In contrast, the computation of musical meter what
is known in the music perception literature as beat induction
can be efected on a highly impoverished musical input. As has
been shown repeatedly, a listener will unproblematically assign
a metrical structure, even when the events to which it is assigned
appear as series of pitchless claps, clicks, or drumbeats. Te sorts
of subtle variation in timbre and pitch characteristic of the pho-
nemic repertoire may tip the balance between competing met-
rical interpretations when these appear in a musical context;
however, they are in themselves insuf cient for the inference of
meter .
Rhythmic Structure in Language and Music
It is worth noting that the uncoupling of a musicalist interpre-
tation from metrical grids in their application in most variants
of metrical stress theory is, in some respects, inconsistent with
the grid notation as it was proposed in work by Mark Liberman
([ 1975 ] 1979), undertaken concurrently with Jackendof and
Lerdahl ( 1983 ). Here, the intention was explicitly musicalist,
namely, to relate the metrical structure of simple childrens
songs and chants to the syntactic and phonological structure
of the words and phrases assigned to them. In contrast to most
approaches, cognitivist and traditional, which view musical and
is therefore, in Steeles words, a subjective mental sensation
deriving from a sense of pulsation giv[ing] the mind an idea
of emphasis independent of any actual increment of sound or
even of any sound at all (1775, 117) . In recognizing the abstract
character of meter, he anticipated twentieth-century cognitiv-
ist approaches that view linguistic stress, along with most other
salient characteristics of language, as mental constructs, pho-
nological rather than phonetic , psychologically real but
only obliquely related to the acoustical or physiological surface
form .
The Grid Representation
Te measure of speech referred to in Steeles title the pat-
terned occurrence of strong and weak metrical positions is
represented in his transcriptions by a three-level hierarchy
appearing below the staf in example 1: Heavy, light, and light-
est locations within each measure are assigned a triangle, three
dots, and two dots, respectively. Tis would be the frst, and for
many years one of the few, attempts to make explicit the under-
lying form corresponding to the way that meter is mentally con-
structed by listeners. When this objective would be reinitiated
in the 1970s, most notably within the generative theory of Ray
Jackendof and Fred Lerdahl ( 1983 ), the representation would
take the form of the metrical grid shown in example 2 from
Mozarts Symphony 40 (Figure 2). It will be noticed that this
example omits the conventional notational means for indicat-
ing the metrical hierarchy barlines, the beaming of eighth notes,
and the time signature. It can do so since these are indicated
with greater precision by the grid, which identifes the relative
prominence of particular locations by their inclusion at suc-
cessive horizontal tiers, referred to as higher levels of the grid.
Relatively strong positions at the measure, half note, and quar-
ter notes are represented by columns appearing above these
locations, while weak positions at the eighth-note level appear
only at the lowest level of the grid.
Tat metrical structure is a fundamental component of
music or, to put it informally, that music frequently has a
beat is, of course, self-evident to most listeners. Tat normal
linguistic utterances are rhythmic in anything like the same
sense is less apparent and remains a subject of some controversy
within linguistics. For this reason, it might appear surprising
that as phonologists confronted a range of data provided by a
cross section of the worlds languages, it became apparent that
Figure 2.
* level 3
* level 2
level 1
* level 0
Ticonderoga
* * *
* * *
*
Figure 3.
Music, Language and
535
and language used to be intertwined, they parted ways long ago.
Musical meter and meter of verse texts cannot be equated; musi-
cal theories of meter need no resurrection (2002, 39).
Tarlinskaya probably represents a consensus position among
scholars of metrics in doubting that there is signifcant evidence
for the infuence of musicalist rhythm on the structure of literary
verse. An ofshoot of generative metrics has been able to avoid
this problem by taking as its primary empirical domain texts
that are unambiguously intended as functioning with a musical
context, namely, lyrics of familiar strophic song forms. Te basis
of this work is the recognition that average listeners encounter-
ing unfamiliar texts for a familiar strophic song will sometimes
efect considerable modifcations in the structure of the original
to accommodate the text. Tus, as noted in Halle and Lerdahl
( 1993 ), those minimally competent in the relevant linguistic and
musical idiom will delete 3 of the original 10 notes of the song
Te Drunken Sailor when they encounter the 7-syllable text
Keel haul him til hes sober, while augmenting the original
melody with 3 additional notes when confronted with the 13 syl-
lables Scrape the hail of his chest with a hoop iron razor. Tese
strikingly uniform intuitions constitute the core data of what
Bruce Hayes (in press) designates as the textsetting problem,
for which he proposes an optimality theoretic solution. It
remains to be seen whether this work will validate Liebermans
initial insight of the deep connection between linguistic and
musical structure, or whether the relevant intuitions will apply
solely to the narrow artistic domain with which these analyses
are concerned .
Conclusion: An Internalist Perspective on Language
and Music
One possible explanation for the discrepancies in the forms
taken by linguistic and musical representations, even when they
are superfcially similar, is that this discontinuity simply refects
the fact of the matter. Tat is, no signifcant overlap in the empir-
ical domains of music and language exists beyond the fact that
both are, in an important sense, products of our minds, which
make use of our psychological capacities for structuring the
external world. But it should be recognized that however close or
distant the ultimate relationship, granting a signifcant psycho-
logical basis to musical structure is itself testimony to the infu-
ence of linguistics, namely, the recognition by modern linguists
of the status of language as grounded not in the external reality of
speech its acoustic and physiological structure (E-language)
but in the underlying psychological mechanisms that give rise to
linguistic behavior ( i-language ).
In contrast, musical scholarship has remained largely a
structuralist enterprise, devoted primarily to describing
the external tokens of music, most commonly musical scores.
Approaches that take as primary the unconscious knowledge
that listeners (and composers in their capacity as listeners)
access in making sense of what they hear and compose are decid-
edly peripheral within the feld. Consequently, confusions as to
what a theory of music is a theory of arise more or less routinely.
Insofar as traditional structuralist theories are seen as ofering
the only empirically sound and intellectually satisfying accounts
of musical form, then linguistic and musical scholarship, aside
from occasional points of convergence, are likely to continue on
linguistic computations as independent and self-contained,
Libermans objective was to establish an equivalence between
the underlying representation of song and speech, a connection
that was understood by Liberman to be in some ways a very
deep one (1975, 81). Tis hypothesis, whatever its ultimate heu-
ristic or conceptual value, has not been infuential within the
feld of linguistics or in music theory.
Two partial exceptions should, however, be mentioned.
While word-level stress provides evidence for the disassociation
of metrical structure in language and music, higher levels of lin-
guistic structure provide some evidence for a musicalist inter-
pretation of linguistic performance. In particular, phrasal stress,
unlike word stress, is not phonologically austere and requires
for its computation, in addition to morphological, syntac-
tic and pragmatic factors, the quasi-musical considerations
of what is referred to as phrasal euphony . Most conspicuous
among these is the stress clash resulting from stressed syllables
from two words appearing adjacent to each other within the
same phrase . Te unacceptable form in a) triggers the applica-
tion of the rhythm rule, which achieves euphony by retracting
leftward the frst of the two syllables involved in the stress clash
to produce the acceptable form b). While not a validation of the
musicalist view, the terminology that is adopted by linguists, as
well as the mechanisms by which this particular phenomenon is
explained, is suggestive of a shared basis underlying the compu-
tation of phrasal stress and the assignment of musical meter.
A second point of contact, as noted by Jamie Kassler ( 2005 ),
is the musicalist interpretation of metrical structure incor-
porated into certain approaches to formal prosody and in its
transformational variant, generative metrics. Te stated goal
of these approaches is to defne the abstract structure of lines
of texts composed in a poetic meter. But even here, the adop-
tion of a musicalist view as representing a signifcant aspect of
the relevant empirical domain remains controversial. Although
poets routinely invoke the music of poetry and the rhythms
of verse and appeal to explicitly musical terms such as phras-
ing , staccato , harmony , and so on, it remains an open question
whether poetic rhythm has any relationship to musical rhythm
as musicians understand as the term . Te reaction to Derek
Attridges Rhythms of English Poetry ( 1982 ) sheds some light on
these questions. Attridge posits a scansion that assigns syllables
to alternating beat and ofbeat positions (indicated by b and o,
respectively), some of which can remain vacant, most notably at
the end of lines (see verse line ). Te general approach and the
representation of empty positions, in particular, is criticized by
Marina Tarlinskaya for failing to recognize that though music
L (2)
L (1)
L (0)
a) * Thirteen men
L (2)
L (1)
L (0)
b) Thirteen men
Figure 4.
Music, Language and
536
a principled extension or, rather, a more or less metaphorical
extrapolation of concepts of grammar developed within the
language sciences.) Such higher-order grammars take the form
of rule systems designed to capture the basic units of narrative
and specify their distributional patterns in a more or less clearly
defned corpus of narrative texts (see corpus linguistics ).
Tis tradition of research can be traced back to the early prece-
dent set by Vladimir Propp ([1928] 1968), who analyzed a corpus
of 100 folktales into a fnite number of structural constituents
that he termed functions (or character actions defned in terms
of their sequential position within an unfolding plot) and identi-
fed rules for their patterning in the corpus he studied.
In the 1970s and 1980s, spurred by the (re)discovery of Propp
by structuralist narratologists, as well as by the attempt to develop
automated systems for story understanding and story generation,
theorists tried to create story grammars with the widest possible
scope. Gerald Prince (1973) drew on Chomskyean transforma-
tional generative grammar in an efort to develop a grammar of
stories with greater descriptive adequacy than the one pre-
sented by Tzvetan Todorov in a 1969 book entitled Grammaire
du Dcamron. Another narratologist, Tomas G. Pavel (1985),
proposed a move grammar that drew not on Chomskyean the-
ory but rather on Propps foundational work, analyzing narra-
tives into problems and moves performed by characters seeking
to bring about their solution . Meanwhile, in a contribution to the
cognitive-psychological strand of story grammar research, J. M.
Mandler (1984, 22) argued that stories have an underlying, or
base, structure that remains relatively invariant in spite of gross
diferences in content from story to story. Tis structure consists
of a number of ordered constituents that include a setting and
an episode, which is in turn decomposable into a beginning that
causes a development that causes an ending.
Given that another entry in this encyclopedia covers story
grammars in greater detail, the remainder of the present dis-
cussion will focus on the second broad approach to studying
the relations between narrative and grammar. Te aim of this
approach is to provide not a grammar of stories but, rather, a
principled account of how stories exploit grammatical resources
in narrative-pertinent (or even narrative-specifc) ways. Here,
the emphasis shifts from a grammar of narrative to the functions
of grammar in narrative or, put another way, to narrative uses
of grammatical structures. Tis second approach is arguably
the predominant one in current linguistically oriented research
on narrative, in part because of critiques of the story grammar
enterprise by theorists such as P. N. Johnson-Laird (1983), who
contends that settings, reactions, moves, and other basic units
posited by story grammarians are not specifed clearly enough
to be construed as elements of a grammar, strictly defned. Tey
are, rather, heuristic constructs based on a prior, unstated gloss
or interpretation of the narrative.
Grard Genettes ([1972] 1980) tripartite model of narrative
structure, which encompasses 1) the story (= the basic sequence
of states, actions, and events recounted), 2) the text on the basis
of which interpreters reconstruct that story, and 3) the act of nar-
ration that produces the text, provides a framework for study-
ing grammar and narrative. Specifcally, analysts can focus on
the role of grammar in narrative viewed under profles 2 and
3 story grammars being a perhaps quixotic attempt to model
their separate paths. If, on the other hand, the posing of interest-
ing questions and a viable theoretical framework relies crucially
on viewing musical works as psychologically based natural
objects in the Chomskyan sense, then in this respect, what we
understand about language has a great deal to ofer our under-
standing of music .
John Halle
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Attridge , Derek. 1982 . Te Rhythms of English Poetry . New York :
Longman .
Bernstein , Leonard . 1976 . Te Unanswered Question: Six Talks at
Harvard . Cambridge : Harvard University Press .
Brown , Stephen . 2000 . Te musilanguage model of music evolution.
In Te Origins of Music , ed. N. L. Wallin , B. Merker , and S. Brown ,
271300 . Cambridge, MA : MIT Press .
Cooke , Deryk . 1959 . Te Language of Music . Oxford : Clarendon .
Halle , John , and Fred Lerdahl . 1993 . A generative textsetting model .
Current Musicology 55 : 3 26.
Halle , Morris , and Jean-Roget Vergnaud . 1987 . An Essay on Stress .
Cambridge, MA : MIT Press .
Hayes , Bruce . Textsetting as constraint confict. In Toward a
Typology of Poetic Forms , ed. Jean-Louis Aroui and Andy Arleo .
Amsterdam: Elsevier. In press.
Jackendof , Ray , and Fred Lerdahl . 1983 . A Generative Teory of Tonal
Music . Cambridge, MA : MIT Press .
Kassler , Jamie . 2005 . Representing speech through musical notation .
Journal of Musicological Research 24 : 227 39.
Liberman , Mark . [1975] 1979. The Intonational System of English .
New York and London: Garland .
Mithen , Stephen . 2005 . Te Singing Neanderthals: Te Origins of Music,
Language, Mind and Body . Londo n: Weidenfeld and Nicolson .
Patel , A. D ., and J. R. Daniele . 2003 . An empirical comparison of rhythm
in language and music . Cognition 87 : B35 B45
Rousseau , Jean-Jacques . [1763] 1997. Essai sur lorigine des langues .
Paris : H. Champion . Reproduction of the Neuchtel manuscript.
Steele , Joshua . 1775 . An Essay Towards Establishing the Melody and
Measure of Speech to Be Expressed and Perpetuated by Peculiar Symbols .
London : J. Almon .
Swain , Joseph . 1997 . Musical Languages . New York : Norton .
Tarlinskaya , Marina . 2002 . Verse text: Its meter and its oral rendition.
In Meter, Rhythm, and Performance , ed. C. Kueper , 3955. Frankfurt
am Main : Peter Lang .
N
NARRATIVE, GRAMMAR AND
Analysts of stories have developed two broad strategies for
studying the relations between narrative and grammar. Te frst
emphasizes grammar of narrative, and the second grammar in
narrative.
Drawing on defnitions of grammar as a model of the catego-
ries and processes underlying intuitive knowledge of (or com-
petence in) a language, theorists working in felds such as
cognitive psychology, artifcial intelligence research, and nar-
ratology have proposed story grammars , that is, gram-
mars of narrative. (A key question is whether this work involves
Narrative, Grammar and
537
Diferent distributions of verb types in texts cue interpreters to
reconstruct storyworlds in which these aspectual values may play
a more or less prominent role. Contrast the emphasis on accom-
plishments in sports broadcasts and hard-boiled detective fctions
with the emphasis on mental states in the modernist novel of con-
sciousness or narratives of the self told in therapeutic settings .
What is more, in ways that M. A. K. Hallidays (1994) func-
tional grammar helps illuminate, patterns of verb selection
assign more or less static or dynamic roles to participants in sto-
ryworlds. From a functionalist perspective, and in parallel with
cognitive linguistic research on how grammar refects
underlying perceptual and conceptual processes used to make
sense of the world (see also cognitive grammar ), verbs
encode construals of experience in terms of processes of vari-
ous types; in turn, each such process type specifes preferences
for assigning roles to the participants involved. For example, the
material process type, encoded in verbs like put or get , assigns
the roles of actor and goal to participants: e.g., She [actor] kicked
the ball [goal] . By contrast, the mental process type, encoded
in verbs like saw , felt , and thought , assigns to participants the
roles of senser and phenomenon : e.g., He [senser] saw the sunrise
[phenomenon] . Although the functionalist approach originally
focused on process types and participant roles at the level of
the clause, aspects of the approach can be scaled up to account
for discourse-level patterning in narrative. Tus, Herman
(2002) suggests that storytelling genres can be characterized
as preference rule systems in which variable preference
rankings obtain for diferent kinds of process types, yielding,
in turn, preferred and dispreferred role assignments. Whereas
epics preferentially rely on material processes, with participants
slotted in the roles of actor and goal, psychological novels prefer
mental over material processes, and with them the participant
roles of senser and experiencer .
Deixis in Narrative
Deictic expressions such as I , here , and now that is, expressions
with interpretations that depend on who utters them to whom
in what communicative circumstances (see pragmatics )
are another part of the grammatical system exploited by nar-
ratives in distinctive ways. At the text level, deictic expressions
serve to locate narrators and characters in time and space vis-
-vis objects, events, and situations in the storyworld, whose
space-time coordinates often do not match those of the current
moment of narration.
Consider, for example, the frst two sentences of Ernest
Hemingways 1927 story Hills Like White Elephants: Te hills
across the valley of the Ebro were long and white. On this side
there was no shade and no trees and the station was between
two lines of rails in the sun. Te preposition across in the frst
sentence and the demonstrative pronoun this in the second sen-
tence must both be interpreted in light of the assumed position
of the vantage point from which events are being narrated a
vantage point that here overlaps with that of the characters .
David A. Zubin and Lynne E. Hewitt (1995) propose the notion
of deictic shift to account for such displaced or transposed
modes of deictic reference, which must be anchored in the sto-
ryworld evoked by the text, rather than in the world(s) that the
text producer or the text interpreter occupies when producing or
profle 1 via grammatical paradigms. Studied as both textual
structure and narrational process, narrative can be analyzed as a
discourse genre that draws in distinctive ways on the same stock
of grammatical resources used diferently in other forms of dis-
course, such as lists, scientifc descriptions, lyric poems, and so
on. Although a range of grammatical resources from lexical
relations and phrase structure to discourse anaphora ,
gapping (see ellipsis ), and topicalization are all poten-
tially relevant for the study of grammar in narrative, in this brief
discussion I focus on just two elements of morphosyntax verbs
and deictic expressions (see deixis ) and map out some of
their narrative functions at both the level of text and the level of
narration .
Functions of Verbs in Narrative
Verbs perform crucial functions in narratively organized dis-
course. At the level of narration, the selection of a particular
tense for the primary or matrix narrative can be used to indi-
cate the relation between event time and narration time, as
when past tense verbs are used to signal retrospective narra-
tion. Further, shifts in verb tense can be used to mark especially
salient episodes, as when (in English-language narratives) sto-
rytellers engage in shifts between the simple past tense and the
conversational historical present as a strategy for underlining the
signifcance of the events being recounted (Wolfson 1982). Te
selection of particular verbal moods can also be used to posi-
tion a tellers account on the continuum stretching between the
realis and irrealis modalities . At issue is a scale that ranges
from expressions indicating a speakers full commitment to the
truth of a proposition about the narrated world to more or less
hedged expressions, which indicate varying degrees of noncom-
mitment. In a foundational contribution to the study of stories
told in face-to-face communicative interaction, William Labov
(1972) argued that one of the identifying features of properly nar-
rative clauses (i.e., clauses that cannot be reordered in discourse
without changing the original semantic interpretation of the
narrative that they convey) is their reliance on past tense indica-
tive verbs. By contrast, subjunctive and other nonindicative ver-
bal moods are used by storytellers to evaluate (signal the point
of) the narrated events. In a series of studies, however, David
Herman found that tellers of supernatural tales in face-to-face
interaction regularly use nonindicative verbal moods (not went ,
but would go or used to go ) to accomplish fuzzy or strategically
inexact reference to the spatiotemporal positions and behavior
of ghosts (cf. Herman 2002, 335).
Meanwhile, at the level of text, verbs play a key role in the pro-
cess that can be characterized as storyworld (re)construction
that is, the use of textual cues to encode or interpret the special
class of mental models that support narrative understanding
(Herman 2002) . For one thing, both conversational narratives
and more complex literary narratives rely on verbs like come
and go (and cognate forms) to map characters paths of motion
through space and time in a process sometimes correlated with
patterns of alliance or confict, as well as with internal or psy-
chological growth, as in the classical Bildungsroman , or novel of
development. Verbs and verb phrases also express Aktionsarten ,
or aspectual values, including states ( I was a hiker ), activities
( I hiked ), and accomplishments ( I hiked up the mountain ).
Narrative, Grammar and
538
Mandler , J. M . 1984 . Stories, Scripts, and Scenes . Hillsdale, NJ : Lawrence
Erlbaum .
Pavel , Tomas G . 1985 . Te Poetics of Plot . Minneapolis : University of
Minnesota Press .
Prince , Gerald . 1973 . A Grammar of Stories . Te Hague : Mouton .
Propp , Vladimir . [1928] 1968. Morphology of the Folktale . Trans. Laurence
Scott, rev. Louis A. Wagner . Austin: University of Texas Press .
Wolfson , Nessa . 1982 . Te Conversational Historical Present in American
English Narrative . Dordrecht, the Netherlands : Foris .
Zubin , David A. , and Lynne E. Hewitt . 1995 . Te deictic center: A the-
ory of deixis in narrative. In Deixis in Narrative: A Cognitive Science
Perspective , ed. Judith F. Duchan , Gail A. Bruder, and Lynne E. Hewitt ,
12955. Hillsdale, NJ : Lawrence Erlbaum .
NARRATIVE, NEUROBIOLOGY OF
Despite cognitive neurosciences long-standing interest in the
processing of individual words and sentences , most neuro-
biological enquiry into the comprehension and production of
more holistic and contextual forms of narrative discourse has
come only recently. Chief among the obstacles to such studies
has been a tension between scientifc control and ecological
validity: Experimental control demands repeatable and rather
predictable stimuli and constrained responses, whereas free-
ranging narrative discourse eschews these properties. In addi-
tion, an analytical focus on mapping individual brain regions
to individual phonological, morphological , and local
contextual properties has upstaged examination of the ways in
which the construction of meaning may be subserved by network
interactions among many brain regions. With newly broadened
experimental paradigms, though, and with novel analytical meth-
ods addressing functional interactions, neuroscience promises
to add a biological dimension to cognitive psychological theories
of narrative discourse.
Central to cognitive neuropsychological explorations of nar-
rative has been the realization of shared structure between the
explicit narratives of literature and the implicit narratives of
everyday mental representations : Information processing
in both these spheres depends on the representational power
conferred by narrative abstraction. In the everyday world no
less than the literary, disconnected percepts gain meaning and
separability from time, place, and action insofar as they become
transformed into representative mental texts, stories whose dis-
tinct scenes contain recognizable characters that act in coherent
plots and evince meaningful themes . A neurobiological perspec-
tive only reinforces these observations from human thought and
behavior: Neurophysiological study of brain dynamics reveals
that human cognitive architecture may be engineered to
represent its processing in series of discrete frames somewhat
analogous to those of cinema (Freeman 2006), and cognitive
neuroimaging has begun to reinforce a neuropsychological
view of thought as an activity of constructing blended spaces
between narrative schemata . Te central role of narrative
scripting in cognition was explored in connection with early
eforts at constructing symbolic systems capable of artifcial
intelligence, and was latent in literary criticism even before the
cognitive revolution of the late twentieth century. What is
new about this connection between thought and narrative is its
decoding these textual signals. Tis model builds on a number
of prior theoretical frameworks, including Karl Bhlers account
of Deixis am Phantasma (= imaginary relocation to the alterna-
tive sets of space-time coordinates implied by utterances about
fctional or imaginary situations) and Kte Hamburgers argu-
ment that only fctional narrative can provide direct access to the
consciousness or I-originarity of another to felt, experien-
tial knowledge of the world as presented via someone elses van-
tage point on events.
In natural-language narratives told in contexts of face-to-
face communication, deictic expressions can serve other func-
tions as well. In particular, when stories are told on-site, that
is, where the events being recounted are purported to have
taken place, deictic references can cue recipients to map fea-
tures of the here-and-now circumstances of narration onto the
space-time environment evoked by the narrative text. Tus, in
the narrative discussed in Herman (2007), Monicas use of per-
son deixis her references to I and we create a referential link
between Monica as the teller in the here and now and Monica
as the coexperiencer (with her friend Renee) of the supernatu-
ral encounter she tells about. More than this, however, Monica
refers deictically to spatial features of the current communica-
tive context, as indicated by the items in bold in the following
partial transcript:
Because she is telling her story on-site, Mary can use deictic
expressions to recruit from features of the current environment
and thereby orient her interlocutors vis--vis the storyworld;
those features provide spatiotemporal coordinates for the situa-
tions and events of which Mary is giving an account.
More generally, on the basis of fundamental cognitive abili-
ties studied in research on conceptual blending , storytellers
like Mary exploit aspects of grammar to build complex mapping
relationships between two mentally projected worlds: the world
evoked by the narrative text and world in which the act of narra-
tion unfolds .
David Herman
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Genette , Grard . [1972] 1980. Narrative Discourse: An Essay in Method .
Trans. Jane E. Lewin, Ithaca, NY : Cornell University Press .
Halliday , M. A. K . 1994 . An Introduction to Functional Grammar . 2d ed.
London : Edward Arnold .
Herman , David . 2002 . Story Logic: Problems and Possibilities of Narrative .
Lincoln : University of Nebraska Press .
. 2007 . Storytelling and the sciences of the mind: Cognitive nar-
ratology, discursive psychology, and narratives in face-to-face interac-
tion . Narrative 15 : 30634.
Johnson-Laird , P.N . 1983 . Mental Models . Cambridge : Harvard University
Press .
Labov , William . 1972 . Te transformation of experience in narrative syn-
tax. In Language in the Inner City , 35496. Philadelphia : University of
Pennsylvania Press .
Monica: we walkin up the hill,
this way , coming up through here .
[]
And Im like on this side and Renees right here .
Narrative, Grammar and Narrative, Neurobiology of
539
to anteriorly, progressing toward the anterior medial temporal
lobe and its memory-related structures. Tese physiological
results in normal volunteers agree very well with the locations of
lesions that impair comprehension in aphasia patients .
An important question is whether the N400 refects the com-
parison of new information against the context maintained in
working memory , or the encoding and integration of this
information into the context, or both these processes in combi-
nation. In addition to these associations with working memory,
N400 amplitude seems correlated with the dif culty of retriev-
ing related information from long-term memory: Words that
are used rarely, for instance, evoke greater N400s than do com-
mon words, and semantic incongruities that introduce unre-
lated categories (buttersand) evoke greater N400s than
within-category violations of congruity (butteroil). N400
amplitude may thus refect the complexity of constructing
blended spaces between the semantic space signifed or evoked
by a term and the space established by its context. It remains an
open question as to what extent similar processes may under-
lie the construction of more complex and temporally extended
blends during the comprehension of complex discourses. It is
interesting to note that negative voltages with timing similar
to the N400 are evoked by all manner of nonlinguistic stimuli,
such as pictures and drawings, suggesting that all forms of
semantic evaluation may involve processes akin to those active
during narrative comprehension or, more abstractly, that the
computations involved in all forms of cognition may have nar-
rative character .
Te stronger N400 in the left hemisphere seems more driven
by category structure and afected by the retrieval of informa-
tion from long-term semantic memory, whereas the right
hemisphere may be more driven by broader contextual inte-
gration and afected by the retrieval and/or updating of working
memory. Te left hemisphere is, therefore, most afected by the
sense of a word considered individually or in relation to its local
context, and the right hemisphere by the broader context of the
narrative (Gernsbacher and Kaschak 2003). Tis computational
distinction of more local semantic evaluation by the left hemi-
sphere and more extended contextual evaluation by the right
hemisphere may map fairly directly onto an anatomical dis-
tinction of small and more isolated dendritic arbors in the left
hemisphere and larger and more overlapping dendritic arbors in
the right (Jung-Beeman 2005), although evoked potentials sug-
gest that this relation may be more a product of left hemisphere
specialization for local processing than of any complementary
right hemisphere specialization for broader context (Coulson
and Van Petten 2007). Cognitively, the pattern is refected in the
right hemispheres involvement in strongly context-dependent
constructions, for instance, those involving frame shifts, such as
metaphor or humor (see verbal humor, neurobiology
of ). Activation of the right hemisphere strengthens as one pro-
ceeds from the level of individual words to sentences to entire
discourses and as a narratives contextual complexity builds
from its beginning to its resolution (Xu et al. 2005).
In addition, in comparison to words and sentences, discourse
uniquely activates medial prefrontal cortex, the temporo-parietal
junction, and the precuneus (at the junction of posterior parietal
and anterior occipital lobes ), as well as subcortical regions
explicit elaboration in light of cognitive neuroscience, connect-
ing literature and philosophy with psychobiological information
and constraint .
Neuropsychologically, comprehension has been studied more
completely than production, and this entry focuses on compre-
hension. Narrative organization is implemented by interacting
and not entirely separable processes of perceptual organization,
attention, and memory . Perceptual organization is the process
that binds separate physical stimuli into coherent higher-order
objects within a scene, replacing, for example, a horizontal plane
and four perpendicular posts with the single entity of a table .
Attention, a group of many subprocesses, focuses processing
on those parts of a scene deemed relevant to the current script
or story schema . Memory encoding , maintenance, and
retrieval, by holding in mind the higher-order representations
of what one has seen before and what one expects to see next,
inform attention and perceptual organization with the context of
this story (Gerrig and McKoon 2001). Although much is known
about the neural substrates of these processes individually, their
signifcance in narrative processing lies in their interactions. An
understanding of these interactions subserving the comprehen-
sion or production of narrative discourse might best begin at the
beginning, with a discussion of contextual integration at the level
of individual words.
Te principal physiological index of a words integration into
its context is the N400 (Kutas and Federmeier 2000), a negative
voltage produced by the brain in response to a word, and maxi-
mal about 400 milliseconds after the word is presented. Te N400
refects a truly textual process largely independent of the partic-
ular sensory mode of representation: Although there are some
more subtle efects of sensory modality, the N400 arises no mat-
ter whether the word is read from a page or heard from a speaker.
Te N400 is thought to refect a process or processes of contex-
tual integration since its amplitude varies parametrically with a
words predictability; the canonical method of evoking a large
N400 is to present a word whose semantics violate contextual
expectation, for instance, At breakfast we ate toast with sand.
Te initial words in this sentence, and the syntax into which
theyre arranged, prime activations for appropriate breakfast
foods. Te ongoing construction of meaning is then challenged
by the non sequitur sand, eliciting a large N400 response. Te
N400 arises no matter whether the conficting context is estab-
lished by a surrounding sentence, as in this brief example, or
simply by a single adjacent word, or by an entire discourse. For
instance, the following discursive context reduces the N400 in
the example above: We camped on the beach. A stif wind blew
of the dunes into all our supplies. At breakfast we ate toast with
sand. Tus, words that are semantically related to their sur-
roundings or are otherwise contextually expected evoke small
N400s, whereas words that cannot be predicted from context
and which, by inference, supply new information with which the
context must be updated evoke large N400s.
Anatomically, electromagnetic source localization and func-
tional neuroimaging place the generators of the N400 primarily
in the superior temporal lobe and temporo- parietal junc-
tion (Van Petten and Luka 2006). Tese sources are mainly in the
left hemisphere but have some contribution from the right.
Across the period of the N400 response, they proceed posteriorly
Narrative, Neurobiology of
540
frontal activation refects a more general association with com-
plex social narratives perhaps related to contextual selection of
details that build coherence within a narrative and that engage a
works suggestion structure to make it relevant to personal
experience and self-representation (Ferstl and von Cramon
2002) . Such contextual selection may instantiate the prefrontal
cortexs more general involvement in the inhibition of responses
deemed inappropriate to the current behavioral and cognitive
context . In contrast, an experimentally based argument has been
made for a more selective association of temporo-parietal junc-
tion with the late-developing, belief-oriented variety of theory of
mind (Saxe and Powell 2006) .
It remains an open question in evolutionary psychol-
ogy as to what extent theory of mind may be a modular cogni-
tive capacity independent of other aspects of cognition, versus to
what extent it may depend on developmental specialization aris-
ing in the interaction of earlier-maturing, more general capaci-
ties for social perception and executive function; recent views
on genes and language suggest that human cognitive adap-
tation for narrative discourse may combine these modular and
generalist perspectives, perhaps by putting to novel uses a large
collection of small modules specialized for cognitive processes
that are applicable to language but not necessarily restricted to
language (Bookheimer 2002). Proponents of the modular view
have often pointed to autism and language , or more spe-
cifcally to autisms impairments in social communication, as
evidence for a modular dysfunction of theory of mind. However,
many people with autism pass theory of mind tests; the dys-
function of medial prefrontal cortex found in imaging studies of
autism seems consistent with a more general defcit in automati-
cally engaging contextual evaluation and self-representation,
and in any case, such abnormalities within specifc regions in the
autistic brain may be refections of a more fundamental disrup-
tion in the information transfer and integration between regions.
Behaviorally observed defcits in theory of mind may thus stem
from a more general perturbation of narrative processing, and
may appear especially prominent only because theory of mind is
so frequently applied in everyday social interaction .
In addition to those regions uniquely activated by discourse
processing, most other brain regions involved in language
or higher-order cognition become more heavily recruited by
(caudate nucleus and dorsomedial thalamus ) that communicate
with prefrontal cortex (Xu et al. 2005). (See Figure 1.) Te involve-
ment of these regions likely refects discourses demands to imag-
ine scenes perceptually and especially visually, to place scenes and
events in spatial relation and temporal sequence, to take up and to
shift between spatial perspectives and personal points of view ,
and to emote and empathize. In particular, the precuneus seems
associated with visual spatial perception and attention, medial
prefrontal cortex and its linked subcortical nuclei with perception
of events in sequence and context, the temporo-parietal junction
with theory of mind , and medial temporal lobe structures with
emotion and memory. Narrative representation can be viewed as
an emergent property of interactions among these and other struc-
tures subserving a broad array of cognitive processes .
One of the most discussed capacities involved in narrative
comprehension and production is theory of mind. Teory of
mind was frst characterized as the general ability to understand
or to model the thoughts and beliefs of other people. However,
more recent neuroscientifc results show that a great deal of
such social attribution can be accomplished using principally
perceptual mechanisms. Tese social perceptual capacities are
computationally and developmentally prior to theory of mind
and include specialized representations for qualities that typify
agency, such as autonomous movement and direction of gaze.
Such perceptual qualities underlie the attribution of volitional
mental states (she/he/it wants or she/he/it wants to) and
perceptual mental states (she/he/it sees) attributions that
form a ubiquitous shorthand in narrative descriptions even in
the case of plainly mechanical and nonsubjective entities (for
example, the computer doesnt see the network, or the printer
wants attention). Teory of mind in its most specifc sense is
essential only for the attribution of belief (she/he/it thinks/
believes/knows), is associated with activation of brain regions
distinct from those subserving more elementary forms of social
attribution, and is distinguished from these more elementary
forms by its appearance at a later stage in child development,
at or near four years of age. Tis developmental connection
is signifcant: Teory of mind seems to arise from simpler pro-
cesses that deal in sensory and especially visual data. Although
early studies associated theory of mind most strongly with the
medial frontal cortex, later work has suggested that this medial
Figure 1.
Narrative, Neurobiology of
541
that, has thus far been realized in multisubject functional neu-
roimaging. Contributing to this limited resolution is a potentially
high degree of variation in detailed functional anatomy across
individual subjects. A further challenge is an intersubject vari-
ability in information transfer between brain areas that actually
refects individual diferences in cognitive style: For instance, dif-
ferent individuals may make more or less use of working mem-
ory in comprehending and producing narrative, and there are
indications that these diferences may be refected in functional
connectivity with cooperating structures in the medial temporal
lobe related to long-term memory .
Tis connectivity frame holds out the potential for a rap-
prochement between connectionism and explicitly represen-
tational, modularist views of language and narrative processing,
since modular functions may reside not so much in any par-
ticular anatomical locus as in the incoming and outgoing links
among these loci: In this sense, the more closely the localization
problem is examined, the more ill-posed it may become. Te
aforementioned characterizations in terms of regional functional
mapping may, therefore, be understood as a frst approach to a
description in terms of regional functional interaction. It is in this
sense that the new neuroscientifc study of discourse processing
is approaching an understanding of narrative connectivity in
terms of neural connectivity .
Tis entry is current as of 15 August 2007.
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bookheimer , Susan . 2002 . Functional MRI of language: New approaches
to understanding the cortical organization of semantic processing.
Annual Review of Neuroscience 25 : 151 88.
Coulson , Seanna , and Cyma Van Petten . 2007 . A special role for the right
hemisphere in metaphor comprehension? ERP evidence from hemi-
feld presentation . Brain Research 1146 : 128 45.
Ferstl , Evelyn C. , and D. Yves von Cramon . 2002 . What does the fronto-
median cortex contribute to language processing: Coherence or theory
of mind? NeuroImage 17.3 : 1599 1612.
Freeman , Walter J . 2006 . A cinematographic hypothesis of cortical
dynamics in perception. International Journal of Psychophysiology
60.2 : 149 61.
Gernsbacher , Morton Ann , and Michael P. Kaschak . 2003 . Neuroimaging
studies of language production and comprehension. Annual Review
of Psychology 54 : 91 114.
Gerrig , Richard J. , and Gail McKoon . 2001 . Memory processes and
experiential continuity. Psychological Science 12.1 : 81 5.
Jung-Beeman , Mark . 2005 . Bilateral brain processes for comprehending
natural language . Trends in Cognitive Sciences 9 : 5121 8.
Karunanayaka , Prasanna R. , Scott K. Holland , Vincent J. Schmithorst ,
Ana Solodkin , E. Elinor Chen , Jerzy P. Szafarski , and Elena Plante .
2007 . Age-related connectivity changes in fMRI data from children
listening to stories . NeuroImage 34.1 : 349 60.
Kutas , Marta , and Kara D. Federmeier . 2000 . Electrophysiology reveals
semantic memory use in language comprehension . Trends in
Cognitive Sciences 4 : 463 70.
Mar , Raymond A. 2004 . Te neuropsychology of narrative: Story compre-
hension, story production, and their interrelation . Neuropsychologia
42 : 1414 32.
Saxe , Rebecca , and Lindsey J. Powell . 2006 . Its the thought that
counts: Specifc brain regions for one component of theory of mind .
Psychological Science 17 : 692 9.
discourse than by individual sentences or words. In particu-
lar, the middle frontal gyrus, on the dorsolateral surface of the
prefrontal cortex, seems involved in placing sentential or other
discursive elements in temporal, causal, or logical sequence
(Gernsbacher and Kaschak 2003; Mar 2004). Tis prefron-
tal sequencing and coordination of ideas seems analogous to
the more concrete executive sequencing and coordination of
body movements implemented in more posterior regions of
the frontal cortex. In a computational sense, therefore, narra-
tive comprehension can be viewed as an elaboration of motor
control or, conversely, motor control itself can be said to
have narrative character, in the sense that it fundamentally
involves sequencing and relating actions in contexts. Tis rela-
tion between language and action gives crucial context to evo-
lutionary biologys eforts to explain the phylogenesis of such
an abstract cognitive capacity, in that the roots of this capacity
may rest in the very concrete domain of motor control. Along
the same lines, the anterior inferior frontal gyrus seems distin-
guished from neighboring cortex by an involvement in selecting
semantic relations in communication with semantic retrieval
processes in the temporal lobe, whereas posterior regions of the
inferior frontal gyrus are more immediately bound up with the
more concrete, sequencing-related details of syntax and pho-
nology (Bookheimer 2002; Jung-Beeman 2005).
In general, the neural implementation of narrative compre-
hension seems to take advantage of individual capacities for
movement, sensory perception, and emotion, activating these
systems in an internal simulation of the events evoked by the
narrative. Te neural implementation of narrative understanding
thus depends crucially on embodiment . At the most concrete
level, that of simulating movements, this process of comprehen-
sion engages the mirror neuron system (see mirror systems,
imitation, and language ) in the ventrolateral frontal lobe
and the supplementary motor area in the dorsomedial frontal
lobe (Wilson, Molnar-Szakacs, and Iacoboni 2008).
Cognitively, narrative understanding seems to emerge from
the interaction of many specialized subsystems. Neurally, there-
fore, a full description of narrative processing must encompass not
only the individual brain regions engaged, but also the dynamics
with which these regions connect and interact over a wide vari-
ety of narrative processes and subprocesses. Tese analyses of
functional connectivity are just beginning (Karunanayaka et al.
2007) via techniques including structural equation modeling,
dynamic causal modeling, and model-free multivariate meth-
ods such as partial least squares and independent components
analysis and initial work has demonstrated information fow
from regions of the temporal lobe adjoining the auditory cortex
(often described as wernickes area , though the anatomical
defnition of this term has never been precise) to higher-order
processing in the inferior frontal lobe ( brocas area ) and near
the temporo-parietal junction. From these areas, the network of
interactions becomes more complex, with the inferior frontal lobe
projecting to the dorsolateral prefrontal cortex, temporo-parietal
junction, and medial frontal cortex, implementing wide-rang-
ing efects of core language processing on complex semantics,
sequencing and coordination of ideas, and social attribution.
Te data now in hand suggest functional specialization and
subdivision of brain regions beyond the anatomical resolution
Narrative, Neurobiology of
542
their formal devices are limited in number while also being
universal in nature. In a way, the Russian Formalists took up
where Aristotle had left off in his Poetics and his Rhetoric on
the study of the architecture of narrative. But their work, as in
the case of Aristotle, did not go beyond the boundaries of clas-
sification and typology.
Te research program developed by the French Structuralists
in the 1960s and 1970s in the guise of a narratology draws on the
work of the Russian Formalists and that of Vladimir Propp con-
cerning the structural analysis of fairy tales (a group included
in the Aarne-Tompson Index of folktale types). Te program
also consists mainly in a taxonomy, but its clarifcation of con-
cepts and criteria of classifcation is much more advanced and
sophisticated .
Te defnition of narrative proposed by H. Porter Abbott is
simple, capacious, and suf ciently precise for our purpose here.
Abbott says: Narrative is the representation of an event or a series
of events (2002, 13). And he explains: Te diference between
events and their representation is the diference between story
(the event or sequence of events) and narrative discourse (how
the story is conveyed) (2002, 15). From this point of view, narra-
tive can be studied without regard to the medium through which
the event or events are represented. Because of this property,
narrative in texts, in flms, in comic books, and so on can be stud-
ied by the means furnished by narratology, with a small amount
of adjustments in each case. As the most fundamental taxonomy,
narratology deals with formal universals, and in that capacity it
is indispensable for a research program seeking to establish the
scientifc study of narrative.
Since the mid-1990s there has been a renewed interest in the-
matics and its empirical study with the purpose of determining
how narratives relate to human universals. Te multidisciplinary
research by Max Louwerse , Willie van Peer , and Donald Kuiken
should be mentioned here, as well as the imaginative and solid
research program being developed by Patrick Colm Hogan on the
basis of his identifcation of three narrative universal structures
or genres (romantic, heroic, and sacrifcial tragicomedic) gener-
ated by emotion prototypes. Another group of approaches con-
cerns the study of the universal features of the relation between
reader or audience and narrative. Here, empirical work, such as
that being realized by David Miall and Kuiken, as well as Deirdre
Wison and Dan Sperber , to name but a few once again, is also
contributing to a scientifc approach to very complex relations.
Evolutionary psychology and cognitive science are tools used by
these researchers in the wake of Joseph Carroll and Robert Storey .
And with respect to the universal features of rhythm, metered
speech, onomatopaeia, and prosody, the long and patient work
of Reuven Tsur has led to many fruitful results. Te same is true
of the much more recent research relating music, neurobiology,
and the emotions.
Indeed, a common experiential link among these groups
of approaches is the emotions. Recent neuroscience has given
solid evidence to the hypothesis that emotions, too, are univer-
sal. Tis is a very important fnding, since emotions are a central
concern in the study of narrative . Te discovery of a mirror neu-
ron system in humans (after its discovery in macaque monkeys
in 1996) has opened the way to a multitude of new explorations
concerning the brain and our social behavior, including the
Van Petten , Cyma , and Barbara J. Luka . 2006 . Neural localization of
semantic context efects in electromagnetic and hemodynamic stud-
ies . Brain and Language 97.3 : 279 93.
Wilson , Stephen M. , Istvan Molnar-Szakacs , and Marco Iacoboni . 2008 .
Beyond superior temporal cortex: Intersubject correlations in narra-
tive speech comprehension . Cerebral Cortex 18.1: 23042.
Xu , Jiang , Stefan Kemeny , Grace Park , Carol Frattali , and Allen Braun .
2005 . Language in context: Emergent features of word, sentence, and
narrative comprehension. NeuroImage 25.3 : 1002 15.
NARRATIVE, SCIENTIFIC APPROACHES TO
Knowledge is acquired in many ways and may exhibit many
degrees of generality, certitude, and power one can use to pre-
dict the future. It can be the product of direct observation and
a small number of general assumptions, or the result of a very
elaborate and long chain of hypotheses and deductions. It can
possess a rich factual content or be almost devoid of it, but it
must always lead back to factual observations. Knowledge, in its
most developed form, called science, rests on the basic assump-
tion that the whole universe is structured and functions accord-
ing to laws that hold without exceptions, in a precise way, and
throughout all time. Sciences main tool is mathematics, which
is as universal as science but is not itself a science; it is sciences
language (or logic). In its turn, science (and sometimes its lan-
guage, mathematics) is used as a tool by many intellectual activi-
ties and felds of study that do not have the status of a science.
Such is the case of narrative, a cognitive endeavor wherein the
methods used and the classifcations arrived at are still far from
yielding scientifc (factual) observations and scientifc (factual)
predictions. Classifcations are an important step in science; sci-
entists generally proceed by identifying and isolating a group of
phenomena that seem related, formulating hypotheses about
their main characteristics, and trying to connect them by means
of a theory. When they succeed, a branch of science becomes
established.
Te study of narrative as such has not followed this path far
enough to become a scientifc discipline. But during the last 15
years, more and more scholars have specialized in the study of
narrative have been using fndings from neurobiology, cogni-
tive science, and evolutionary psychology to enlarge our under-
standing of narrative and to ground it in the architecture of the
human brain .
The first attempts to establish narrative as an autonomous
scientific discipline took place in Russia just before the 1917
Revolution and continued until about 1930, when they were
stopped by Stalins regime in the Soviet Union. The Russian
Formalists, as the researchers were called, made salient con-
tributions to their field such as the search for literariness ,
or the formal properties defining the literary text; the dis-
tinction between plot ( syuzhet ) and story ( fabula ); the con-
cept of ostraniene , translated into English by the neologism
estrangement; the notion that a literary text is a system, as
is literature itself; and the setting of boundaries between the
study of a text itself and the scientifically irrelevant study of
its production or its reception. These contributions were con-
ceived within a research program that acknowledges the fact
that narratives are infinite in subject and presentation, but
Narrative, Scientifc Approaches to
543
literary prosody is available online at: http://www.arsversifcandi.net/
backissues/vol1/essays/tsur.html.
Wilson , Deirdre , and Dan Sperber . 2006 . Relevance theory. In
Handbook of Pragmatics , ed. Gregory Ward and Laurence R. Horn ,
60732 . Oxford : Blackwell .
NARRATIVE UNIVERSALS
In one Arawak story, a tiger sees a great hunter in the forest,
changes herself into a woman, and marries him. Te two have a
happy married life until the good wife suggests that they visit the
hunters family. She warns him that he must not reveal her ori-
gin to anyone. Te hunter, however, tells his mother the secret.
Feeling ashamed in front of the community, the woman changes
back into a tiger and returns to the forest. Te poor husband
would often go into the bush and call his wife, but there never,
never came a reply (Roth 1915, 2034).
At frst glance, the story might seem strange and clearly not
universal. A man marries a tiger and must keep it a secret from
his mother? Claude Lvi-Strauss analyzes the story in relation
to a complex of South American myths (see homologies and
transformation sets ) that recount the decline from a golden
age. He also connects the tale with cannibalism (1973, 259). Lvi-
Strausss analysis does implicitly link the tale with other tradi-
tions, simply by referring to a decline from a golden age, for this
topic is found in diferent cultures. In the Judeo-Christian tradi-
tion, the story of the Fall is a case in point. However, in some ways,
this link only makes the story more alien. Pairing Adam and Eve
with a hunter and a tiger seems to highlight cultural diference
and pairing the eating of an apple with cannibalism seems to put
the Arawak at quite a distance from anything Western.
On the other hand, there is something deeply familiar about
the story. It tells of a couple joined by attachment (witness the
husbands pathetic search at the end of the story), sufering con-
fict due to the husbands divided loyalties to his mother and his
wife; it treats shared secrets, feelings of betrayal and shame, and a
concern about social origins. Moreover, there is nothing in the tale
itself that suggests cannibalism, despite Lvi-Strausss analysis.
In short, it is a story that seems both strange and familiar, both
culturally particular and imbued with cross-cultural concerns.
Cultural Construction, Universality, and Narrative
Tere is a common view in cultural studies the interdisciplinary
area of the humanities and social sciences devoted to analyzing
culture that practices, dispositions, artifacts, communicative
actions are socially constructed. Tis is to say that they are
not innate or biologically determined but result from cultural
developments. In this view, anything from individual emotions
to political structures might not be analyzed in terms of rela-
tively constant human predispositions, but rather in terms of the
historically contingent organizations and imperatives of social
practice or performance. Additionally, writers in cultural studies
commonly understand social construction as widely variable. In
the most extreme versions, this variation may be seen as limited
by little beyond the laws of physics.
An alternative view, often associated with evolutionary
approaches to culture, takes a wide range of social practices to
be very narrowly constrained by genetic propensities. Tese
production and reception of narrative. Te mirror neuron sys-
tem in humans has been consistently reported as being related
to imitation, action observation, intention understanding, and
understanding of the emotional states of others, to mention a
few of the human faculties that are essential for the right percep-
tion of fctional narratives and, of course, for the survival and
evolution of the human race. And there seems to be even more
to it. In 2004, Vittorio Gallese , Christian Keysers , and Giacomo
Rizzolatti published a paper in which they explore the possibil-
ity that the mirror neuron system, by providing us with an expe-
riential (precognitive) insight into other minds, could provide
the frst unifying perspective of the neural basis of social cogni-
tion (2004, 401). Te mirror neuron system is apparently the
basic mechanism that allows us to grasp the intentions of others
and to experience similar emotions (empathy). As these authors
say in their article , A crucial element of social cognition is the
brains capacity to directly link the frst- and third-person expe-
riences of these phenomena (i.e., link < I do and I feel > with < he
does and he feels >) (ibid., 396).
Te question arises: Would all art then be nothing but a spe-
cifc and specialized activity aimed at fring the mirror neurons
in a certain direction? Would the will to style displayed by
authors, composers, flm directors, and so on be nothing but the
deliberate use of syntax, semantics, pragmatics, and phonemics
to trigger the mirror neuron system in specifc and predeter-
mined ways in order to elicit specifc and predetermined insights
and emotions in the reader and the audience?
Frederick Aldama
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Abbott , H. Porter . 2002 . Te Cambridge Introduction to Narrative .
Cambridge : Cambridge University Press .
Aristotle . 1984 . Te Rhetoric and Poetics of Aristotle . Trans. Rhys Roberts
and Ingram Bywater. New York : McGraw-Hill .
Carroll , Joseph . 2004 . Literary Darwinism: Evolution, Human Nature,
and Literature . New York : Routledge .
Gallese , Vittorio, Christian Keysers , and Giacomo Rizzolatti . 2004 . A uni-
fying view of the basis of social cognition. Trends in Cognitive Science
8.9 : 396403.
Hogan , Patrick Colm . 2003 . Te Mind and Its Stories: Narrative Universals
and Human Emotion . Cambridge : Cambridge University Press .
Louwerse , Max, and Don Kuiken , eds. 2004. Te Efects of Personal
Involvement in Narrative Discourse Processes . Philadelphia: Lawrence
Erlbaum .
Louwerse , Max, and Willie van Peer , eds. 2002. Tematics: Interdisciplinary
Studies . Amsterdam: John Benjamins .
Miall , David . 2006 . Literary Reading: Empirical and Teoretical Studies.
New York: Peter Lang .
Miall , David , and Donald Kuiken . 1994 . Foregrounding, defamiliariza-
tion, and afect response to literary stories. Poetics 22 : 389 407.
Propp , Vladimir . 1968 . Morphology of the Folk Tale . Austin : University of
Texas Press .
Rizzolati , Giacomo, Vittorio Gallese, Leonardo Fogassi , and Luciano Fadiga .
1996 . Action recognition in the premotor cortex. Brain 119 : 593 609.
Storey , Robert . 1996 . Mimesis and the Human Animal: On the Biogenetic
Foundations of Literary Representation . Evanston, IL : Northwestern
University Press .
Tsur , Reuven . 1997 . Poetic rhythm: Performance patterns and their
acoustic correlates . Versifcation 1 : 1997. Tis electronic journal of
Narrative, Scientifc Approaches to Narrative Universals
544
Tese processes have adaptive value because they approximate
the function of giving us access to other peoples states of mind.
However, since they are mere approximative mechanisms, they
are fallible a point that is highly consequential both in life and
in stories. Consider narratives that focus particular attention on
the relative opacity of others intentions (e.g., between lovers in
cases of misguided jealousy). If these recur cross-culturally (as
they do), they do so not because adaptive mechanisms succeed
but because they fail in certain systematic ways.
nongenetic universals may also arise due to patterns in
childhood development that are not genetically programmed.
For example, we seem to have innate predispositions to emo-
tional attachment. However, our attachment responses are not
wholly hardwired. Tey are shaped in some crucial ways by
childhood experiences, as a number of writers have stressed fol-
lowing John Bowlby. Tere appear to be some crucial parameters
in our early childhood experiences that have lasting efects on
the quality and durability of our attachments in later life. While
cultures may vary in the degree to which one or another parent-
ing/attachment style predominates, it is inevitably the case that
every society has variation in parenting/attachment styles. Tus,
insofar as narratives cross-culturally focus attention on social
emotion, we would expect to fnd cross-cultural expression of
the same basic attachment styles .
Te preceding example indicates that there are two prob-
lems with the usual framing of the division between those who
claim that there are cross-cultural universals and those who
deny that claim. First, universality does not entail innateness. A
pattern may be universal without being genetically determined.
Second, social construction does not entail cultural diference.
Tough they include a genetic component, attachment styles are
socially constructed in that they result from the childs experi-
ence of parenting. Yet it seems likely that similar divisions of, for
example, secure and insecure attachment will recur everywhere,
even if they do so in diferent proportions. Tese considerations
suggest that common dichotomies regarding universality and
cultural construction are false. Te point has consequences for
our understanding of universality in a range of areas, not only
narrative.
Indeed, the point goes further. Research in group dynamics
and elsewhere (see network theory; self-organizing
systems; pragmatics, universals in ) suggests that many
patterns may arise through convergent development (independ-
ent processes in diferent societies that give rise to parallel prac-
tices) for example, patterns in the ways social networks operate
to defne in-groups and out-groups, intragroup inequality, inter-
acting subgroups, and so on. Given the importance of group
antagonism, social hierarchy, and subgroup divisions for any
society and, thus, their importance for the lives of individual
agents we might expect narratives to emplot these relations fre-
quently. Insofar as these relations derive from group dynamics,
cross-cultural patterns of such emplotment would not result
from genetic predispositions per se but from convergent social
developments .
With these points in mind, we might return to the story of
the tiger woman or jaguar woman, in Lvi-Strausss version.
It is, as it turns out, framed by attachment (along with sexuality)
and group opposition. Moreover, there are hints that the group
propensities are thought to have resulted from adaptations that
are specifcally social. While writers in cultural studies tend to see
social practices as quite variable, writers adopting this approach
tend to see societies as manifesting a wide range of univer-
sals . Language study has been one area in which universalism
has been prominent, though there has been some disagreement
as to the precise evolutionary origins of language (see
biolinguistics ).
narratology has incorporated both tendencies. A com-
mon view in cultural studies is that narrative is socially con-
structed and can vary widely from culture to culture. In contrast,
some researchers drawing on models from linguistics and psy-
chology have argued that there are remarkably consistent nar-
rative patterns across cultures. Tere are, however, diferences
between the study of narrative and the sorts of study that have
occupied linguists for example, the study of syntax . While
writers in cognitive linguistics have viewed syntactic prin-
ciples as resulting from general cognitive structures and pro-
cesses, a common view within the feld is that there are some
aspects of cognitive architecture that are specially devoted to
syntactic processing (see autonomy of syntax ). Te case for
an autonomy of narrative is much weaker. It seems much more
likely that narrative results from the interaction of various cog-
nitive structures and processes that are not specially devoted to
narrative. As such, narrative is a less likely candidate for simple
evolutionary analysis. Put diferently, if narrative results from
our cognitive abilities to draw causal inferences, to attribute
intentions, to imagine counterfactual or hypothetical situations,
to adopt varying physical points of view, to simulate experiences,
and so on, then it is less likely that there is any single adaptive
function for cross-cultural narrative patterns (comparable to the
commonly posited communicative function for language) .
Te point is consequential for a number of reasons, relating
to narrative and to other areas of study in the language sciences.
Specifcally, if narrative patterns are unlikely to be genetically
coded in any detail, one of two conclusions may be drawn: One
may simply see this as further evidence for the culturalist posi-
tion, further reason to believe that narrative may vary across
cultures with few limitations. However, the existence or nonex-
istence of universals is an empirical issue. It cannot be decided
a priori. If one believes that the evidence supports conclusions
of universality, then one is likely to draw a diferent conclusion
from the indirectness of the relation between narrative and
adaptation. Certainly, some universal patterns will derive more
or less directly from aspects of cognitive architecture (e.g., causal
attribution) that are defned by genetic programs resulting from
selective pressure. Others will derive from commonalities in the
physical environment. But that is not all.
Sticking close to biology, we may note that some cross-cul-
tural patterns are likely to derive from the fact that adaptations
are mechanisms, not functions which means that there are
cases where the mechanism fails. More exactly, the genetic pre-
dispositions that serve us so well in daily life do so because they
set out relatively simple procedures that approximate advanta-
geous functions. For example, although evolutionary psy-
chologists often refer to our ability to read minds, we do not
directly know other peoples intentions. Rather, we engage in
complex processes of simulating and inferring those intentions.
Narrative Universals
545
force of arms. However, Layla is married to another man. When
Layla and Majnun die, they are reunited in paradise .
Romance and Prototypes
Although this is only a tiny selection of narratives, it is signifcant
in part due to their prominence in distinct narrative traditions.
(On complications of establishing this distinctness in literary
study, see areal distinctness and literature .) We may
already begin to see the ways in which these narratives may
share certain prototypical characteristics. I say prototypi-
cal characteristics because these examples do not suggest a set
of necessary and sufficient conditions but a gradient of
more or less standard cases. Specifcally, one common sort of
narrative that recurs cross-culturally tends toward a prototype
involving the following elements.
We begin with two lovers. Teir mutual interest combines
sexuality and attachment. However, they face inhibition.
Tat inhibition is frequently a matter of confict with author-
ity usually parental or religious or group division, or both.
(Te priming of religious fgures due to the prototype may
explain their surprising presence as helpers in some stories.)
Te group division is itself regularly one of in-group hierarchy
or in-group/out-group antagonism. Works that do not involve
such a confict commonly suggest it, as in akuntal . After a
brief union, the lovers are separated often, one is confned
to home while the other is sent away. In tragic versions, one or
both die. In comic versions, the separation may be associated
with death. During this separation, one lover proves himself
(or, less often, herself) worthy of the beloved, sometimes by
defeating the rival (who may ultimately die). Tis demonstra-
tion may overturn the disapproval of the parents or society. In
the end, the lovers are reunited and the conficting families are
reconciled.
Of course, individual stories must vary this pattern. However,
as a standard case, it appears to be remarkably consistent across
cultures and across time. Te Arawak story varies the pattern
more than the others we have considered. But it remains rec-
ognizable. Te diferences are largely a matter of order. Te
hunter proves himself worthy of the beloved through his suc-
cessful hunting right at the outset (the point is related to Arawak
cultural practices in which potential bridegrooms must prove
themselves; see Roth 1915, 31516). Te confict occurs after
marriage, rather than before, and it is in part the fault of the
man for violating the trust of his wife and preferring his mothers
interests over hers, no matter how briefy. Is there a reason for
these diferences? Te comic form of the romantic plot involves
suggestions of death or unending separation of the lovers
prior to their ultimate reunion in part because this intensifes
the fnal joy of their union. Te point is a simple matter of the
psychology of emotion the joy of an outcome is intensifed by
the dif culty of achieving the outcome (see Ortony, Clore, and
Collins 1988, 73) and by the gradient of change from a previous
emotional state. Te same point holds for the Arawak story, but
in reverse. Here, the atypically early union, the apparently hap-
pily ever after condition of the couples married life, and the
hunters subsequent tragic error serve to intensify the pathos of
the conclusion .
opposition may point either toward in-group hierarchy or toward
in-group/out-group antagonism. Alexandra Aikhenvald reports
that at least among some Arawaks, members of one low-prestige
subgroup are referred to as people of jaguar (2006, 12). Walter
Roth cites an Arawak proverb that identifes tigers with enemies
(1915, 367). Moreover, the story relies on, indeed elaborates on,
the failure of mind reading, which is precisely what allows the
secret and the issue of spousal loyalty to arise in the frst place.
So there is certainly commonality here, commonality that makes
cognitive sense. But is that all there is to it? After all, we knew that
there was some commonality already. Is there any greater univer-
sality to this story or to narrative patterns more broadly?
Attachment combined with sexual desire (as in pair bonding
or marriage) points toward a set of stories that recur across cul-
tures. Perhaps we will get a better understanding of the issues if
we consider some other stories of this sort, particularly paradig-
matic works from other traditions. (In a brief entry, we cannot
consider many stories, or other evidence for narrative universals.
For a range of cases, and for references to other accounts of nar-
rative universality, see Hogan 2004.)
Four Romances
To begin, lets consider what is almost certainly the paradigm of
romantic narrative in the English-speaking world Romeo and
Juliet . Romeo and Juliet fall in love. However, they are prevented
from uniting by the group antagonism of their parents. With the
help of a friar, they are briefy united, but then Romeo is exiled
and Juliet is confned to the home. Juliet is to be married to a
rival by her father. She fakes her death to escape this fate. Romeo
returns, kills his rival, then commits suicide just at the moment
when he might have been united with Juliet. Juliet, too, kills her-
self, but after their deaths, the families are reunited .
Now consider the Romance of the Western Chamber , Chinas
most popular love comedy, both on stage and in print, begin-
ning in the twelfth century (Idema 2001, 800). Chang and Ying-
Ying fall in love. Chang goes of to take the imperial exams.
Meanwhile, a rival comes to marry Ying-Ying with her mothers
approval. Chang succeeds in the examination and returns to
elope with Ying-Ying. He is successful due to the help of a monk.
Te rival commits suicide (see Idema 2001, 798800) .
Te Recognition of akuntal , the most revered work of
Sanskrit drama, begins when Duyanta and akuntal fall in
love. Duyanta worries that they cannot marry due to caste
(thus, an internal group hierarchy). akuntal worries that they
cannot marry due to her fathers disapproval. Both turn out to
be mistaken. Tey are united. However, akuntal violates her
obligations to a holy man, who curses her with separation from
Duyanta. In consequence, akuntal is exiled, while Duyanta
remains at home, sufering confict with the demands of an earl-
ier wife (thus, a rival). Duyanta defeats an army of demons
in battle and is subsequently reunited with akuntal and
their son .
In the Arab and Muslim world, few stories have been as popu-
lar and infuential as that of Layla and Majnun. Layla and Majnun
fall in love, but Laylas father refuses the marriage. Majnuns
father tries to cure Majnun of his love madness through religion,
but Majnun only calls on God to make him worship Layla more.
Majnun wanders the desert, eventually trying to win Layla by
Narrative Universals
546
narrative universals would seem to have consequences for our
understanding of universals elsewhere and for our understand-
ing of the place of both biology and social construction in an
account of universals .
Patrick Colm Hogan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aikhenvald , Alexandra . 2006 . A Grammar of Tariana . Cambridge :
Cambridge University Press .
Bowlby , John . 1982 . Attachment and Loss . New York : Basic Books
Hogan , Patrick Colm . 2004 . Te Mind and Its Stories: Narrative Universals
and Human Emotion . Cambridge : Cambridge University Press .
. Afective Narratology: Te Emotional Structure of Stories .
Lincoln: University of Nebraska Press. In press.
Idema , Wilt . 2001 . Traditional dramatic literature. In Te Columbia
History of Chinese Literature , ed. Victor Mair , 785847. New
York : Columbia University Press .
Lvi-Strauss , Claude . 1973 . From Honey to Ashes (Introduction to a
Science of Mythology: 2). Trans. John and Doreen Weightman. New
York : Harper & Row .
Ortony , Andrew, Gerald Clore , and Allan Collins . 1988 . Te Cognitive
Structure of Emotions . Cambridge : Cambridge University Press .
Roth , Walter . 1915 . An Inquiry into the Animism and Folk-Lore of the
Guiana Indians: Tirtieth Annual Report of the Bureau of American
Ethnology to the Secretary of the Smithsonian Institution : 19081909 .
Washington, DC : Government Printing Of ce , 103453.
NARRATIVES OF PERSONAL EXPERIENCE
Te study of narrative extends over a broad range of human
activities: novels, short stories, poetic and prose epic, flm, folk-
tale, interviews, oral memoirs, chronicles, histories, comic strips,
graphic novels, and other visual media. Tese forms of commu-
nication may draw upon the fundamental human capacity to
transfer experience from one person to another through oral
narratives of personal experience.
A focus on spontaneous recounting of experience was greatly
stimulated by the development of sociolinguistic research in
the 1960s, designed to capture the closest approximation to
the vernacular of unmonitored speech. Narratives of personal
experience were found to reduce the efects of observation to a
minimum (Labov 2001). Since then it has appeared that such
narratives are delivered with a similar organization in a wide vari-
ety of societies and cultures as, for example, in the Portuguese of
fshermen in northeastern Brazil (Maranho 1984). Te follow-
ing discussion of oral narratives is based on the initial analysis of
William Labov and Joshua Waletzky (1967), as developed further
in the suggested reading.
Te discussion frst treats the structural organization of nar-
rative (temporal organization, orientation, coda), then turns to
the evaluative component and fnally to the construction of nar-
rative as a folk theory of causality instrumental to the assignment
of praise and blame.
Structural Organization
A narrative is defned here as one way of recounting past events,
in which the order of narrative clauses matches the order of
events as they occurred. Example (1) is a minimal narrative orga-
nized in this way:
Understanding Narrative Universals
Perhaps surprisingly, the explanation of the Arawak storys dif-
ference from the cross-cultural prototype begins to suggest why
there is a cross-cultural prototype to begin with. Cross-culturally,
there are two common purposes of narrative verbal art the
communication of emotionally satisfying experiences (roughly,
a psychological purpose) and the treatment of thematically sig-
nifcant issues, often ethical or political (roughly, a social pur-
pose). Te explanation of narrative universals bears importantly
on these two elements.
Narratives involve sequences of action engaged in by inten-
tional agents pursuing goals that we share and that engage
us emotionally. One thing that cross-cultural patterns sug-
gest is that these narrative goals are much more limited, and
much more cross-culturally widespread, than one might have
imagined. For example, they include union with a partner in
an enduring relationship that is both sexual and founded in
attachment thus, romantic love (on other happiness goals, the
related emotions, and the associated narrative structures, see
Hogan 2004). Te precise development of narratives results in
part from the means necessary to intensify emotional experi-
ences such as creating a relatively sharp change from separa-
tion anxiety to reunion, enhancing confict by involving people
who themselves have attachment bonds (e.g., parents and chil-
dren), and so on .
Again, the development of romantic narratives also crucially
includes real social concerns. Most obviously, these involve in-
group/out-group divisions and group hierarchies, which pre-
sumably result from group dynamics. But group organization
does not delimit the entire social world. Individual biological
endowments, developmental idiosyncrasies, and experiential
accidents in later life guide personal af liations. Tere is, in con-
sequence, no way of guaranteeing that personal af liations will
conform to the principles of group hierarchization or in-group/
out-group antagonism. Societies are, then, condemned to face
conficts between interpersonal attachments and the segrega-
tions imposed by social organization. Romantic plots tell the
story of that confict .
In sum, there seem to be signifcant narrative universals
(many, of course, statistical; others absolute ). Tese uni-
versals arise from a complex interaction of factors, including
biological endowment (e.g., in basic emotional responses), pat-
terns in childhood development, and convergent developments
arising through group dynamics. In this way, narrative univer-
sals are in part derived from biological adaptations. However,
they are no less derived from social constructions, which are
themselves universal. An understanding of narrative universals
is important for at least three reasons: 1) Narratives are a cen-
tral part of human life everywhere. Understanding narratives is
therefore crucial to understanding the human mind and human
experience. 2) Te precise narrative universals we discover tell
us some surprising things about human society. For example,
it is striking that most romantic plots develop our sympathy for
the lovers, not for the society. Tis suggests not only that certain
sorts of confict are inevitable in society but also that we share
a deep sympathy with individuals or couples working against
social hierarchization and group antagonism a surprising and
in many ways hopeful fact. Finally, 3) the complex nature of
Narrative Universals Narratives of Personal Experience
547
fact realized. Frequently, such evaluative clauses are concen-
trated in an evaluation section, suspending the action before
a critical event and establishing that event as the point of the
narrative.
Evaluative clauses vary along a dimension of objectivity. At
one extreme, narrators may interrupt the narrative subjectively
by describing how they felt at the time:
(8) a. I couldnt handle any of it
b. I was hysterical for about an hour and a half
In a more objective direction, narrators may quote themselves
(I said to myself Tis is it), or with more credibility, cite a third
party witness, as in (5). At the other extreme, objective events
speak for themselves, as in the account of a plane developing
motor trouble over Mexico City:
(9) And you could hear the prayer beads going in the back of the
plane.
Evaluation provides justifcation for the narratives claim on
a greater portion of conversational time than most turns of talk,
requiring an extended return of speakership to the narrator until
it is fnished (Sacks 1992). Evaluation thus provides a response
to the potential question So what? (Spanish Y que? ; French Et
alors? ).
Narratives of personal experience normally show great vari-
ation in the length of time covered by the clauses in the orien-
tation, complicating the action and evaluation sections, ranging
from decades to minutes to seconds. Sequences of clauses of
equal duration may be termed chronicles ; these are not designed
to report and evaluate personal experience .
Reportability and Credibility
A reportable event is one that itself justifes the delivery of the
narrative and the claim on social attention needed to deliver it.
Some events are more reportable than others. Te concept of
reportability or tellability (Norrick 2005) is relative to the situation
and the relations of the narrator with the audience. At one end of
the scale, death and the danger of death are highly reportable in
almost every situation. At the other end, the fact that a person ate
a banana for lunch might be reportable only in the most relaxed
family setting. Most narratives are focused on a most reportable
event . Yet reporting this event alone does not make a narrative; it
only forms the abstract of a narrative.
For a narrative to be successful, it cannot report only the most
reportable event. It must also be credible if the narrative is not
to be rejected as a whole by the listener. Tere is an inverse rela-
tionship between reportability and credibility: Te more report-
able, the less credible. Narrators have available many resources
to enhance credibility. In general, the more objective the evalua-
tion, the more credible the event .
Narrative Preconstruction
When a narrator has made the decision to tell a narrative, he or
she must solve the fundamental and universal problem: Where
should I begin? Te most reportable event, which will be desig-
nated henceforth as e
0
, is most salient, but one cannot begin with
it. Given the marked reportability of e
0
and the need to estab-
lish its credibility, the narrator must answer the question How
(1) a. Well, this man had a little too much to drink
b. and he attacked me
c. and a friend came in
d. and she stopped it.
Te same events could have been reported in the non-narrative
order c,d,a,b as in (2), which employs a variety of grammatical
devices within a single clause.
(2) A friend of mine came in just in time to stop this person who
had had a little too much to drink from attacking me.
Narrative structure is established by the existence of temporal
juncture between two independent clauses. Temporal juncture
is said to exist between two such clauses when a change in the
order of the clauses produces a change in the interpretation of
the order of the referenced events in past time. Tese are nar-
rative clauses. Narrative clauses respond to a potential question
what happened then? and form the complicating action of the
narrative.
A narrative normally begins with an orientation , introduc-
ing and identifying the participants in the action: the time, the
place, and the initial behavior. The orientation section pro-
vides answers to the potential questions who? when? where?
what were they doing? In the minimal narrative (1), the first
clause (a) is the orientation. More information is usually
provided:
(3) a. my son has awell, it was a fairly new one then.
b. Its a 60 cc Yamaha.
c. and it could move pretty good.
d. Tis fella and I were going down the road together
Te end of a narrative is frequently signaled by a coda , a state-
ment that returns the temporal setting to the present, precluding
the question and what happened then?
(4) a. And you know the man who picked me out of the water?
b. Hes a detective in Union City,
c. and I see him every now and again .
Evaluation
Most adult narratives are more than a simple reporting of events.
A variety of evaluative devices are used to establish the evalua-
tive point of the story (Polanyi 1989). Tus, we fnd that narra-
tives, which are basically an account of events that happened,
frequently contain irrealis clauses negatives, conditionals,
futures which refer to events that did not happen or might have
happened or had not yet happened:
(5) And the doctor just says Just that much more, he says, and
youd a been dead.
(6) Ill tell you if I had ever walloped that dog Id have felt some
bad.
(7) a. And he didnt come back.
b. And he didnt come back.
Irrealis clauses serve to evaluate the events that actually did
occur in the narrative by comparing them with an alternate
stream of reality: potential events or outcomes that were not in
Narratives of Personal Experience
548
insertions of pseudoevents and removing them, detecting dele-
tions and replacing them, and exchanging excuses for the action
excused. It is then possible to approximate the original chain of
events on which the narrative is based. A useful exercise is to
develop a complementary sub rosa case in the interests of the
antagonist. Te comparison of these two constructions deepens
our understanding of how narrative skills are enlisted to trans-
form the social meaning of events without violating our commit-
ment to a faithful rendering of the past .
William Labov
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Gofman , Erving . 1981 . Forms of Talk. Oxford : Blackwell .
Labov , William . 2001 . Te Social Stratifcation of English in New York City .
2d ed. Cambridge : Cambridge University Press .
. 2003 . Uncovering the event structure of narrative. Georgetown
University Round Table 2001 . Ed. Deborah Tannen and James Alatis,
6383. Washington, DC: Georgetown University Press. Tis article
develops the search for the events underlying the narrative.
. 2004 . Ordinary events. In Sociolinguistic Variation: Critical
Refections . ed. C. Fought , 3143. Oxford : Oxford University Press . An
exploration of the evaluative efect of inserting ordinary events into
narrative.
. 2006 . Narrative preconstruction . Narrative Inquiry 16 : 3745. A
fuller development of this topic.
Labov , William , and Joshua Waletzky . 1967 . Narrative analysi s. In Essays
on the Verbal and Visual Arts , ed. J. Helm, 1244. Seattle: University
of Washington Press. Repr. Journal of Narrative and Life History 7
(1997) : 3 38.
Maranho , Tulio . 1984 . Te force of reportive narratives. Papers in
Linguistics 17 (3): 235 65.
Norrick , Neal R . 2005 . Te dark side of tellability . Narrative Inquiry
15 : 323 44.
Polanyi , Livia . 1989 . Telling the American Story . Cambridge, MA : MIT
Press .
Sacks , Harvey . 1992 . Lectures on Conversation . Vols. 1 and 2. Ed. Gail
Jeferson . Oxford : Blackwell .
Trabasso , T. , and P. van den Broek . 1985 . Causal thinking and the
representation of narrative events . Journal of Memory and Language
27 : 1 22.
NARRATOLOGY
Te French term narratologie (formed in parallel with biology ,
sociology , etc. to denote the study of narrative) was coined by
Tzvetan Todorov in his 1969 book Grammaire du Dcamron.
Te early narratologists participated in a broader structuralist
revolution that sought to use Saussurean linguistics as a pilot
science for studying diverse forms of cultural expression, which
structuralist theorists characterized as rule-governed signifying
practices or languages in their own right (see structuralism ;
Culler 1975). Likewise, narratologists such as Todorov, Roland
Barthes , Claude Bremond , Grard Genette , and Algirdas Julien
Greimas , adapted Ferdinand de Saussures distinction between
la parole and la langue to construe particular stories as individ-
ual narrative messages supported by an underlying semiotic
code (see semiotics ). And just as Saussurean linguistics privi-
leged code over message, focusing on the structural constituents
and combinatory principles of the semiotic system of language,
did this (remarkable) event come about? Te answer requires a
shift of focus backwards in time to a precursor event e
-1
, which is
linked to e
0
in the causal network in which events are represented
in memory (Trabasso and van den Broek 1985). In traversing this
network in reverse, the causal links found may be event-to-goal,
goal-to-attempt, or attempt-to-outcome. Te process will con-
tinue recursively to e
-2
, e
-3
, and so on. until an ordinary, mun-
dane event e
-n
is reached, for which the question Why did you
do that? is absurd, since e
n
is exactly what we would expect the
person to do in the situation described. Te event e
n
is, of course,
the orientation. Tus, a narrator telling of a time he was on shore
leave in Buenos Aires begins,
(10) a. Oh, I was settin at a table drinkin.
TRIGGERING EVENTS. Given the mundane and nonreportable
character of the orientation, it follows that the frst link in the
causal chain is a triggering event , which drives the narrative
along the chain toward the most reportable event. Tus, (10) is
followed by (11):
(11) b. an this Norwegian sailor come over
c. an kep givin me a bunch o junk about how I was sittin
with his woman.
How ordinary situations like (10) can give rise to the report-
able and violent events that followed is a mystery that narrative
analysis can only contemplate, since they are part and parcel of
the contingent character of history .
The Transformation of Experience
Te participants in many narratives include protagonist,
antagonist, and third-party, witnesses, of which the frst is the
most complex. Elaborating on Gofman (1981, 1445), one can
identify many egos present: the self as original author of the
narrative and its immediate animator ; the self as actor ; the self
as generalized other (normally as you); the anti-self as seen
by others; and the principal , the self in whose interest the story
is told. Tat interest is normally advanced through a variety
of techniques that do not require any alteration in the truth-
fulness of the events reported. Te re-creation of the causal
network involves the assignment of praise and blame for the
critical events and their outcomes. Most narratives of confict
involve linguistic devices that contribute to the polarization
of protagonist and antagonist, though within the family, other
linguistic forms lead to the integration of participants. Te
devices used to adjust praise and blame include most promi-
nently the deletion of events, an operation that can often be
detected by close reading. Key elements in further manipula-
tion are the grammatical features of voice: active versus pas-
sive, but also zero causatives that assign agency (He drove
through town with a chaufeur) or verbs that imply the exer-
tion of authority and resistance to it (My dad let me go with
him). Other narrative devices function to increase the impres-
sion of agency: pseudoevents that may not correspond to any
physical event (I turned to him and, I took this girl and, I
started to hit him but).
Narrative analysis can show how the prima facie case is built
to further the interests of the principal. Tis involves detecting
Narratives of Personal Experience Narratology
549
between nuclei and catalyzers in his 1966 Introduction to the
Structural Analysis of Narratives (Barthes [1966] 1977). Renamed
kernels and satellites by Chatman (1978), these terms refer to core
and peripheral elements of story content, respectively. Delete or
add to the kernel events of a story and you no longer have the
same story; delete or add to the satellites and you have the same
story told in a diferent way. Related to Tomashevskiis work on
free versus bound motifs, Viktor Shklovskiis early work on plot
as a structuring device established one of the grounding assump-
tions of structuralist narratology: namely, the fabula - sjuzhet or
story-discourse distinction (see story and discourse ), that
is, the distinction between the what and the how, or what is being
told versus the manner in which it is told .
Another important precedent was furnished by Vladimir
Propps Morphology of the Folktale ([1928] 1968), whose frst
English translation appeared in 1958. Propp distinguished
between variable and invariant components of higher-order nar-
rative structures more specifcally, between changing dramatis
personae and the unvarying plot functions performed by them
(e.g., act of villainy, punishment of the villain, etc.). In all, Propp
abstracted 31 functions, or character actions defned in terms of
their signifcance for the plot, from the corpus of Russian folktales
that he used as his data set; he also specifed rules for their distri-
bution in a given tale. His approach constituted the basis for later
accounts of narrative structure. For instance, extrapolating from
what Propp had termed spheres of action, Greimas ([1966]
1983) sought to create a typology of general behavioral roles to
which particularized actors in narratives could be reduced. He
initially identifed a total of six roles (which he termed actants )
underlying individual narrative actors: subject, object, sender,
receiver, helper, and opponent .
Establishing the Field: Structuralist Narratology
I have already begun to discuss how the structuralist narratolo-
gists built on Russian Formalist ideas to help consolidate what I
am referring to as the classical tradition of research on narrative.
As originally conceived (cf. Barthes [1966] 1977), the new science
of narratology aimed to be not a school or method of literary crit-
icism that is, not a way of interpreting novels or other specif-
cally literary narratives but, rather, a transmedial investigation
of stories of all kinds, naturally occurring as well as artistically
elaborated, verbal (spoken or written) as well as image based,
painted as well as flmed. It also aimed to be transcultural and
transgeneric, investigating everything from legends and fables
to epics and tragedies. Ethnographic and sociological impulses,
refecting the linguistic, anthropological, and folkloristic bases
for structuralist analysis of narrative, reveal themselves, when
Barthes writes: All classes, all human groups, have their narra-
tives, enjoyment of which is very often shared by men with difer-
ent, even opposing, cultural backgrounds ([1966] 1977, 79).
Narratologys grounding assumption is that a common, more
or less implicit model of narrative explains peoples ability to
recognize and interpret many diverse productions and types
of artifacts as stories; the same model allows them to compare
an anecdote with a novel or an opera with an epic. In turn, the
raison dtre of narratological analysis is to develop an explicit
characterization of the model underlying peoples intuitive
knowledge about stories, in efect providing an account of what
rather than on situated uses of that system, structuralist narratol-
ogists privileged narrative in general over individual narratives,
emphasizing the general semiotic principles according to which
basic structural units (characters, states, events, actions, etc.) are
combined and transformed to yield specifc narrative texts.
In this brief overview, I trace in further detail some of the
developments from which structuralist narratology took rise
and outline key contributions by early theorists. I also review
limitations of the structuralist approach to narrative inquiry
limitations that manifested themselves as story analysts began
to engage more fully with recent research in the language sci-
ences, among other areas of study. To map the evolution of the
feld, I draw a distinction between classical and postclassical
approaches to narratological analysis (cf. Herman 1999) . Classical
narratology encompasses the tradition of research, rooted in
Russian Formalist literary theory as well as earlier precedents,
that was extended by structuralist narratologists starting in the
mid-1960s and refned and systematized up through the early
1980s by scholars such as Mieke Bal , Seymour Chatman , Wallace
Martin , Gerald Prince , and others. Te Anglo-American tradition
of scholarship on fctional narrative can also be included under
the rubric of classical approaches, though for reasons of space,
this discussion focuses mainly on the Formalist-structuralist tra-
dition . Postclassical narratology , meanwhile, designates frame-
works for narrative research that build on this classical tradition
but supplement it with concepts and methods that were unavail-
able to story analysts during the heyday of structuralism. In
developing postclassical approaches, which not only expose the
limits but also exploit the possibilities of older models, theorists
of narrative have drawn on a range of felds, from gender theory,
philosophical ethics, and comparative media studies to socio-
linguistics , the philosophy of language, and cognitive science.
Given the focus of the present encyclopedia, I concentrate here
on productive synergies between postclassical narratology and
research in the language sciences .
The (Recent) Prehistory of Narratology
Te Russian Formalists authored a number of pathbreaking
studies that served as foundations for narratological research.
Crucially, the Formalists sought to create a stylistics suit-
able for larger verbal structures found in prose narratives of all
sorts, from Leo Tolstois historically panoramic novels to tightly
plotted detective novels to (Russian) fairy tales. Tis widened
investigative focus would prove to be a decisive development
in the history of modern-day narratology. Te new focus helped
uncouple theories of narrative from theories of the novel , shifting
scholarly attention from a particular genre of literary writing to
all discourse or, in a broader interpretation, all semiotic activities
that can be construed as narratively organized. Te Formalists
thus set a precedent for the transgeneric and indeed transmedial
aspirations of French structuralist theorists such as Bremond
and Barthes, who came later.
Not only was the general orientation of Formalist research
narratologically productive; more than this, specifc Formalist
concepts were taken over more or less directly by structuralist
story analysts. For example, in distinguishing between bound
(or plot-relevant) and free (or non-plot-relevant) motifs,
Boris Tomashevskii provided the basis for Barthess distinction
Narratology
550
the storyworld. In singulative narration, there is a one-to-one
correspondence between these frequency rates; in repetitive
narration, events are recounted more often than they occur;
and in iterative narration, events that happen more than once
are recounted fewer times than the frequency with which they
actually occur .
For all the gains it achieved by drawing on linguistics as a pilot
science (or, rather, as a metaphor for disciplinary practice), how-
ever, structuralist narratology was also limited by the linguistic
theories it treated as exemplary. Barthes unintentionally reveals
the limits of structuralist narratology when he remarks that a
narrative is a long sentence, just as every constative sentence is in
a way the rough outline of a short narrative, suggesting that one
fnds in narrative, expanded and transformed proportionately,
the principal verbal categories: tenses, aspects, moods, persons
([1966] 1977, 84). By contrast, post-Saussurean language theory
has underscored that certain features of the linguistic system
conversational implicatures , discourse anaphora , pro-
tocols for turn-taking in conversation (see adjacency pair ),
and so on emerge only at the level beyond the sentence. In
other words, attempting to bring to bear on narrative texts a
code-centered linguistics that ignores distinctive features of lan-
guage in use, the early narratologists lacked crucial resources for
the analysis of stories. Te problem, then, is not with the origi-
nal intuition of the narratologists namely, that linguistics can
serve as a pilot science for narratological research. Te problem,
rather, is with the particular linguistic concepts they used to fesh
out that intuition .
Beyond Structuralism: Postclassical Narratology and the
Sciences of Language
Ironically, the narratologists embraced Saussures structural-
ist linguistics as their point of reference just when its defcien-
cies were becoming apparent in the domain of linguistic inquiry
itself. Te limitations of the Saussurean paradigm were thrown
into relief, on the one hand, by emergent formal (e.g., genera-
tive-grammatical) models for analyzing language structure
(see generative grammar ). On the other hand, powerful
tools were being developed in the wake of Ludwig Wittgenstein ,
J. L. Austin , H. P. Grice , John Searle , and other post-Saussurean
language theorists interested in how contexts of language use
bear on the production and interpretation of socially situated
utterances. Teorists working in this tradition began to question
what they viewed as counterproductive modes of abstraction and
idealization in both structuralist linguistics and the Chomskyan
paradigm that displaced it. Indeed, the attempt by later narrative
scholars to incorporate ideas about language and communica-
tion that postdate structuralist research has been a major factor
in the advent of postclassical models for research on stories and
storytelling. To put the same point another way, one reason for
the shift from classical to postclassical narratology has been an
ongoing efort to move from using linguistics as a metaphor for
narrative research to using linguistic models in the actual prac-
tice of narratological inquiry.
Te following are just some of the domains of narratological
research in which theorists have begun to import concepts and
methods from the modern-day sciences of language, in an efort
to build models with greater descriptive and explanatory power
constitutes humans narrative COMPETENCE . Hence, having
conferred on linguistics the status of a founding model ([1966]
1977, 82), Barthes identifes for the narratologist the same object
of inquiry that ( mutatis mutandis ) Saussure had specifed for the
linguist: the code or system from which the infnity of narrative
messages derives and on the basis of which they can be under-
stood as stories in the frst place.
Narratologists like the early Barthes used structuralist linguis-
tics not just to identify their object of analysis but also to elabo-
rate their method of inquiry. In this connection, the adaptation of
structuralist-linguistic concepts and methods was to prove both
enabling and constraining. On the positive side, the example of
linguistics did provide narratology with a productive vantage
point on stories, afording terms and categories that generated
signifcant new research questions. For example, the linguistic
paradigm furnished Barthes with what he characterized as the
decisive concept of the level of description (Barthes [1966]
1977: 8588). Imported from grammatical theory, this idea sug-
gests that a narrative is not merely a simple sum of propositions
but, rather, a complex structure that can be analyzed into hierar-
chical levels in the same way that a natural-language utterance
can be analyzed at the level of its syntactic , its morphological ,
or its phonological representation. Barthes himself distin-
guishes three levels of description. At the lowest or most gran-
ular level are basic meaning-bearing elements that he termed
functions , which can be mapped out both distributionally and
in terms of paradigmatic classes; then come characters actions
that collocate to form narrative sequences; and fnally there is
the level of narration , or the profle that narrative assumes when
viewed as a communicative process.
Likewise, Genette ([1972] 1980) drew on a broadly gram-
matical paradigm in using the categories of tense , mood , and
voice to characterize the relations among the story (= the basic
sequence of states, actions, and events recounted), the text on
the basis of which interpreters reconstruct that story, and the
act of narration that produces the text. Indeed, Genettes work
in the area of narrative temporality constitutes one of the truly
outstanding achievements in the feld. Developing distinctions
that bear an interesting resemblance to Hans Reichenbachs
(1947) discriminations among event time, reference time, and
speech time, Genette focuses on two kinds of temporal rela-
tionships: 1) that between between narration and story and
2) that between text and story. In connection with the frst,
Genette distinguishes between simultaneous, retrospective,
prospective, and intercalated modes of narration; in connec-
tion with the second, he develops the categories of duration,
order, and frequency. Duration can be computed as a ratio
between the length of time that events take to unfold in the
world of the story and the amount of text devoted to their nar-
ration, with speeds ranging from descriptive pause to scene to
summary to ellipsis. Order can be analyzed by matching the
sequence in which events are narrated against the sequence
in which they can be assumed to have occurred, yielding
chronological narration, analepses or fashbacks, and pro-
lepses or fashforwards, together with various subcategories
of these nonchronological modes. Finally, frequency can be
calculated by measuring how many times an event is narrated
against how many times it can be assumed to have occurred in
Narratology
551
a greater awareness that narratology is itself one of the sciences
of language specifcally, the domain of inquiry whose focal con-
cern is narratively organized sign systems across all media and
communicative settings. Once a subdomain of literary study,
narratology is now coming into its own as the comprehensive
science of narrative-pertinent phenomena originally envisioned
by the structuralist narratologists.
Evidence of this reconfguration of the feld can be found in
narratologists reengagement with natural-language narratives
told in contexts of face-to-face interaction. Although William
Labov and Joshua Waletzky (1967) developed their model for the
analysis of narratives told in contexts of face-to-face communica-
tion just as structuralist narratologists were proposing their key
ideas, and although the Labovian model has been extraordinarily
infuential in social-scientifc research for some four decades, ini-
tially there was little interaction between sociolinguistic research
on storytelling and other traditions of narrative scholarship. But
now there is increasing interest in building an integrative theory
that can accommodate both the study of written, literary narra-
tives and the analysis of everyday storytelling (Fludernik 1996;
Herman 2002). At the same time, among researchers concerned
with face-to-face narrative communication, there has been a shift
analogous to the one I have characterized as a transition from
classical to postclassical approaches. Precipitating this shift is
the recognition that the Labovian model captures one important
subtype of natural-language narratives namely, stories elicited
during interviews but does not necessarily apply equally well
to other storytelling situations, such as informal conversations
between peers, he-said-she-said gossip, conversations among
family members at the dinner table, or, for that matter, written,
literary texts (see conversation analysis ).
Tis convergence of sociolinguistic, discourse-analytic, and
narratological research suggests that narratology is now coming
into its own as a bona fde member of the language sciences. As
such, its chief aim is to enhance our understanding of stories not
only as a means of artistic expression or a resource for communi-
cation but also as a fundamental human endowment .
David Herman
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Barthes , Roland . [1966] 1977. Introduction to the structural analysis
of narratives. In Image Music Text , trans. Stephen Heath, 79124.
New York : Hill and Wang .
Chatman , Seymour . 1978 . Story and Discourse: Narrative Structure in
Fiction and Film . Ithaca, NY : Cornell University Press .
Culler , Jonathan . 1975 . Structuralist Poetics: Structuralism, Linguistics,
and the Study of Literature . Ithaca, NY : Cornell University Press.
Emmott , Catherine . 1997 . Narrative Comprehension: A Discourse
Perspective . Oxford : Oxford University Press .
Fludernik , Monika . 1996 . Towards a Natural Narratology . London :
Routledge .
Genette , Grard . [1972] 1980. Narrative Discourse: An Essay in Method.
Trans. Jane E. Lewin. Ithaca, NY : Cornell University Press .
Greimas , Algirdas-Julien . [1966] 1983. Structural Semantics: An Attempt
at a Method . Trans. Danielle McDowell, Ronald Schleifer, and Alan
Velie. Lincoln : University of Nebraska Press .
Herman , David . 1999 . Introduction. In Narratologies: New Perspectives
on Narrative Analysis , ed. David Herman , 130. Columbus : Ohio State
University Press .
than those developed by the classical narratologists. In each of
these domains, story analysts are working to adopt more narra-
tive-appropriate tools from the language sciences that is, tools
that can throw light on how narrative, as a distinctive kind of lan-
guage use, constitutes a cognitive and communicative resource
by means of which human beings make sense of themselves, one
another, and the world.
Narrative Comprehension : To explore aspects of narrative
processing, story analysts have drawn on a range of theoreti-
cal frameworks that were unavailable to the structuralist nar-
ratologists, including artifcial intelligence research (work on
knowledge representations), accounts of mental models ,
cognitive linguistics , and research on text processing.
For example, Catherine Emmott (1997) presents a powerfully
integrative theory of narrative comprehension as a process of
using textual cues to build and update complex mental rep-
resentations that she terms contextual frames , which contain
information about narrative agents, their situation in time
and space, and their relationships with one another. Mark
Turner (2003), meanwhile, relates story comprehension to
more general cognitive processes that involve conceptual
blending .
Speech and Tought Representation : Narratologists have
drawn on felds ranging from dialectology, pragmatics,
discourse analysis (linguistic) , and historical
linguistics to study aspects of speech representation in
narrative, including dialect representations and fctional
portrayals of scenes of conversational interaction. Likewise,
to study representations of characters mental function-
ing, analysts have begun to work toward a rapprochement
between narratological theory and ideas from cognitive and
social psychology, research on emotion, cognitive linguistics,
and other frameworks for inquiry .
Focalization Teory : Initially given impetus by Genettes
([1972] 1980) attempt to reformulate theories of narrative
perspective or point of view in more rigorous terms, focal-
ization theory has in recent years taken on an increasingly
interdisciplinary profle. Manfred Jahn (1996) has drawn on
the cognitive science of vision to propose a powerful account
of the perspective-marking features of narrative. Meanwhile,
David Herman (2009) uses ideas from cognitive grammar
to propose refnements to Genettes theory .
Quantitative, Corpus-Based Research : Story analysts have
begun to work with large text corpora (see corpus lin-
guistics ) to study whether the distributional facts support
accounts proposed by earlier narratologists on the basis of
their own readerly intuitions. On the one hand, hypothesis-
driven approaches to corpus study use a top-down method,
attempting to map assumed categories of structure onto spe-
cifc texts or corpora to test the validity of prior theories. On
the other hand, bottom-up approaches, seeking to reduce
theoretical presuppositions to a minimum, work to induce
categories and models from surface features that can be
identifed through automated analysis of narrative corpora .
It is not just that narratologists have begun to engage more fully
with concepts and methods from the language sciences, how-
ever; more than this, the feld is currently being revolutionized by
Narratology
552
whole set of ethnolinguistic nationalisms in Europe immedi-
ately afterwards. Tough the historical diferences between the
various social movements cannot be elided, they were inspired
by a number of German post-Kantian idealist thinkers. J. G.
Herders assertion in 1768 that each national language forms
itself in accordance with the ethics and manner of thought of
its people (2002, 50) was an important articulation of the link
between language and nation; by the time that William von
Humboldt gave his defnition of a nation in 1836 (a body of men
who form language in a particular way [1988, 153]), the connec-
tion appeared almost axiomatic. In 1808, J. G. Fichte spelled out
the political signifcance of linguistic nationalism by arguing that
wherever a separate language is found, there a separate nation
exists, which has the right to take charge of its independent
afairs and to govern itself (1968, 49). Te implications of the
doctrine were realized in the role that it played in national inde-
pendence campaigns conducted by Greeks, Czechs, Hungarians,
Bulgarians, Ukrainians, Finns, Norwegians, Afrikaners, and
the Irish. Some postcolonial activists today, the Kenyan writer
Ngg wa Tiongo , for example, use the same model of linguistic
nationalism in their contemporary struggle, not so much against
colonialism but in order to counter the legacy of colonial rule .
Andersons account of the nation as an imagined commu-
nity drew attention to the constructedness of the concept by
pointing to its precise historical origins. Yet the role of language
in the imagining of the community of the nation is also one that
arises at particular moments in history and serves specifc func-
tions; it is neither transhistorical nor general. It is also worth
noting that the conception of language underpinning this act of
imagination is one that has been criticized . Tus, M. M. Bakhtin,
in the important essay Discourse in the Novel, points to the fact
that national languages are produced by various types of insti-
tutional forces intellectual (linguistic theorizing), educational
(grammars and dictionaries), political (legislation) which act
centripetally in order to create a determinate, fxed, and know-
able form. As part of this process, the realities of heteroglossia
(see dialogism and heteroglossia ) social diference
inscribed in language by means of variation past and present
have to be banished. Historians, such as E. J. Hobsbawm (1990),
have noted the historical signifcance of such linguistic selection
and ranking, while linguistic anthropologists have drawn atten-
tion to the fact that the homogeneous language of nationalism is
as imaginary as the community that accompanies it (Irvine and
Gal 2000).
Te extent to which such insights will have an impact in
political and linguistic thought remains to be seen. It is certainly
the case, however, that the postulated relationship between lan-
guage and nation is now treated much more skeptically. At the
reactionary edge of forms of linguistic nationalism, there are still
those who argue for the purity of language as a way of guar-
anteeing the integrity of the nation. But the very fact that the
vast majority of nations past and present have been multilingual
communities including a number of those whose very entrance
into history depended on an emphasis on their supposed mono-
lingualism radically undermines the ideological case for lin-
guistic nationalism .
Tony Crowley
. 2002 . Story Logic: Problems and Possibilities of Narrative .
Lincoln : University of Nebraska Press .
. 2009 . Cognitive approaches to narrative analysis. In Cognitive
Poetics: Goals, Gains, and Gaps , ed. Geert Brne and Jeroen Vandaele ,
79118 . Berlin : Mouton de Gruyter .
Jahn , Manfred . 1996 . Windows of focalization: Deconstructing and
reconstructing a narratological concept . Style 30.3 : 241 67.
Labov , William , and Joshua Waletzky . 1967 . Narrative analysis: Oral ver-
sions of personal experience. In Essays on the Verbal and Visual Arts ,
ed. June Helm , 1244. Seattle : University of Washington Press .
Propp , Vladimir . [1928] 1968. Morphology of the Folktale . Trans. Laurence
Scott, rev. Louis A. Wagner. Austin : University of Texas Press .
Reichenbach , Hans . 1947 . Elements of Symbolic Logic . New
York : Macmillan .
Turner , Mark . 2003 . Double-scope stories. In Narrative Teory and the
Cognitive Sciences , ed. David Herman , 11742. Stanford, CA : CSLI .
NATIONALISM AND LANGUAGE
While the link between language and nationality is often pre-
sented as though it developed at some primordial point in the
past, its appearance is, in fact, quite recent. Tis is hardly surpris-
ing inasmuch as the conception of the nation itself is relatively
modern. Tus, the idea that language is the medium by which
nationality is established, that language is the key to the nation,
has to be traced historically. Two distinct but related contexts
may serve as examples concerning how and why the connection
was made.
In the sixteenth century, the Tudor monarchy sought to exer-
cise its dominion over Ireland, a colony which had been nomi-
nally under English rule since 1169 but which had never quite
been successfully subjugated. Part of its centralizing project was
the imposition of English upon the whole of the island of Ireland
on the ground that the use of the native language, Gaelic, along
with other cultural factors such as behavior and dress, led the
Irish to think of themselves as being of sundry sorts, or rather of
sundry countries ( Statutes 1786, 28H8.cxv) rather than as mem-
bers of one polity united under the English crown. Tis stress on
the signifcance of linguistic diference, embodied in the Act for
the English Order, Habit and Language (1537), formed the basis
of the English policy of linguistic colonialism in Ireland, but, of
equal importance, it heralded the connection between language
and national identity . In his 1617 Itinerary , Fynes Moryson ,
an English adventurer in Ireland, articulated the lesson that the
colonialists learned from their struggle to impose English lan-
guage and order: [C]ommunion or diference of language hath
always been observed a special motive to unite or alienate the
minds of all nations. And in general all nations have thought
nothing more powerful to unite minds than the community of
language (Moryson [1617] 1903, 213). Under specifc historical
conditions the clash between an early modern nation-state and
one of its colonies linguistic diference came to signify national
diference through the operation of military and discursive
power. Te link established in this context served as a portent
of a more general connection that appeared later in Europe and
beyond.
Although his seminal account of nationalism identifes its ori-
gins in the New World in the late eighteenth and early nineteenth
centuries, Benedict Anderson also discusses the appearance of a
Nationalism and Language
553
Various terms and phrases can be cited as examples of
non -natural kind terms. Student with a long nose who visited
Malaysia denotes a kind whose defning properties are not
related together in any lawlike regularity, and is, therefore, of
no use for the understanding of nature. A term like nonhuman
designates a group that is too heterogeneous. Another example
often cited is that of artifcial kind terms, such as pencil or apart-
ment . But this is perhaps problematic: It seems to presuppose
that humans, with their artifacts, constitute a kingdom within
a kingdom. But if Homo sapiens sapiens is a natural kind, and
as such part of nature, then terms useful for describing its life
and behavior for example, apartment should perhaps count
as NKTs.
Recent philosophical discussion has concentrated on the
meaning of NKTs. Until the 1960s, philosophers spoke of these
terms as if they were synonymous with a group of identifying
descriptions of the kinds. Te statement that some liquid is water,
say, would then be synonymous with the statement that it has (at
least most of) the properties that would be used, for instance, in a
good, scientifcally informed dictionary to characterize water.
Tis description theory of NKTs is problematic. According to
it, if a scientist asks a child for a glass of water, what the scien-
tist means by water is very diferent from what the child means
by it, and the latter cannot even understand the former. But
this is unacceptable, for fuent communication is a criterion for
understanding .
Te most infuential theory of the meaning of NKTs nowa-
days, essentialism , was developed during the 1970s by Saul
Kripke (1980) and Hilary Putnam (1975). Both claimed that the
meaning of an NKT is determined not by descriptions but by
ostensive reference to samples. Natural kinds are assumed to
have essential properties, and the NKT means something having
the same essential properties as (most of) these samples , although,
as a rule, when introducing an NKT, people would be ignorant of
these essential properties .
Kripke also claimed that NKTs are rigid , but this seems con-
fused. First, an NKT say, tiger is not rigid in the sense of des-
ignating the same particulars in every possible world , since
in diferent possible worlds there exist diferent tigers. Secondly,
it is not rigid in the sense that if it designates a particular in one
possible world, it designates it in every possible world in which it
exists : Te queen bee is presumably a natural kind, but whether
larvae develop into queen bees depends on how they are fed. So
an insect that is a queen bee might not have been one, and queen
bee designates it only in some of the possible worlds in which it
exists. Lastly, and perhaps more importantly, if what was meant
in calling NKTs rigid is that they preserve their meaning across
possible worlds, then this is true of non-NKTs as well, such as
student with a long nose who visited Malaysia, and it would
trivialize the meaning of rigidity (cf. Schwartz 2002) .
A hypothetical example supporting essentialism that many
found convincing was developed by Putnam. He asks us to imag-
ine a remote planet identical to ours (Twin Earth), apart from
the fact that instead of water, that is, H
2
O, it contains a super-
fcially identical liquid of an entirely diferent composition, say
XYZ. (Let us ignore the fact that such a liquid would not quench
our thirst, and so wouldnt even be superfcially like H
2
O.) Since
Twin Earths liquid is superfcially indistinguishable from water,
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Benedict . 1991 . Imagined Communities: Refections on the
Origin and Spread of Nationalism . London : Verso .
Bakhtin , M. M . 1981 . Te Dialogic Imagination: Four Essays . Ed. Michael
Holquist , trans. Caryl Emerson and Michael Holquist . Austin : University
of Texas Press .
Barbour , Stephen, and Cathie Carmichael , eds. 2002. Language and
Nationalism in Europe . Oxford : Oxford University Press .
Fichte , J. G . [1808] 1968. Addresses to the German People . Ed. G. Armstrong
Kelly . New York : Harper.
Herder , J. G . [1768] 2002. Philosophical Writings . Ed. and trans. Michael
N. Forster . Cambridge : Cambridge University Press .
Hobsbawm , E. J . 1990 . Nations and Nationalisms since 1780 .
Cambridge : Cambridge University Press .
Humboldt , William von . [1836] 1988. On Language: On the Diversity
of Human Language Construction and Its Infuence on the Mental
Development of the Human Species . Ed. Michael Losonsky , trans. Peter
Heath . Cambridge : Cambridge University Press .
Irvine , Judith T. , and Susan Gal . 2000 . Language ideology and linguis-
tic diferentiation. In Regimes of Language: Ideologies, Polities and
Identitie s , ed. Paul Kroskrity , 3583. Oxford, UK : James Currey .
Joseph , John . 2004 . Language and Identity: National, Ethnic, Religious .
London : Palgrave .
Moryson , Fynes . [1617] 1903. Shakespeares Europe: Unpublished
Chapters of Fynes Morysons Itinerary . Ed. C. Hughes . London : Sherratt
and Hughes .
Ngg Wa Tiongo . 1986 . Decolonising the Mind: Te Politics of Language
in African Literature . London : James Currey .
Te Statutes at Large Passed in the Parliaments Held in Ireland . 1786
1801. 20 vols. Dublin.
NATURAL KIND TERMS
Natural kind terms (NKTs) are, to use Platos ancient metaphor,
those terms that carve Nature at her joints; they are the terms
that correspond to unities and diversities in nature ( Phaedrus ,
265e266b). Tey therefore enable lawlike generalizations,
descriptions of natural patterns, and explanations of natural
phenomena .
From this characterization of NKTs it is clear that science
strives to use such terms in its classifcation and explanation
of nature. It is also clear that, as a rule, NKTs are developed
together with the growth of our knowledge of nature, and they
both result from a better understanding of phenomena and
advance that understanding. For instance, the biblical classifca-
tion of plants into grass, the herb yielding seed and fruit tree
yielding fruit whose seed is in itself (Genesis 1:11) is no longer
used in botany, which classifes some trees together with some
grass as angiosperms, the fowering plants, in contrast to some
other trees, which are gymnosperms. Te same point is illus-
trated by the recent scientifc controversy over the defnition of
planet : Scientists aimed at forming a concept that would refect
and allow a better understanding of the diferent characteristics
and origins of bodies orbiting the sun.
Te most common examples of NKTs are names of sub-
stances. Gold , water , alcohol , and metal are names of natural
kinds of matter; Homo sapiens sapiens , primates , mammals , ani-
mals , and eukaryotes are names of natural kinds of organisms.
But often enough one fnds names of natural phenomena, such
as heat or pain , counted among these terms as well .
Natural Kind Terms
554
Tis debate can be understood as being about the necessary
and suf cient conditions for the correct predication of Taitu-
as-afraid. While some claim that a particular confguration of
changes in the autonomic nervous system are both necessary
and suf cient for predicating of Taitu that she is afraid, others
respond that while such changes might be necessary conditions
they do not amount to suf cient conditions, for such changes can
be brought about by the administration of drugs. Tus, the truth
conditions for the predication of Taitu-as-afraid must involve
more than the existence of these bodily changes, for example,
Taitu also holding the belief such that she is subject to a threat and
the desire to act in a way that diminishes or absents that threat.
Furthermore, some might (and do) hold that conditions such as
patterned changes in the autonomic nervous system might be
suf cient for a particular token, such as Taitu being afraid on
this occasion, but not for all applications the term afraid . Tis
latter claim might then draw on Ludwig Wittgensteins discus-
sion of family resemblance and hold that all correct appli-
cations of a term are not necessarily conditional on something
being common to all .
Phil Hutchinson
NEGATION AND NEGATIVE POLARITY
Negation is a linguistic, cognitive, and intellectual phenomenon.
Ubiquitous and richly diverse in its manifestations, negation is
fundamentally important to all human thought. As Laurence R.
Horn and Yasuhiko Kato put it:
Negative utterances are a core feature of every system of human
communication and of no system of animal communication.
Negation and its correlates truth-values, false messages, contra-
diction, and irony can thus be seen as defning characteristics of
the human species. (2000, 1)
Cognitively, negation is elementary of-line thinking; it involves
some comparison between a real situation lacking some partic-
ular element and an imaginal situation that does not lack it. Te
particular element in focus anchors and contextualizes the nega-
tive element (which, being constrained by grammar, frequently
doesnt provide enough information for a listener to determine
what its focus is intended to be). Tere are many diferent con-
versational and written strategies for indicating and interpreting
focus elements, and even more for modulating them.
Formally (see logic and language ), a functor called by
logicians negation is the only signifcant monadic functor; its
behavior is described by the most basic axiom of logic, the Law
of Contradiction ( (pp ), NKpNp , also known as Te Law of
Non-Contradiction), which asserts that no proposition is both
true and not true. Pragmatically (see pragmatics ), negation
provides, among many other concepts, the basic cancelation test
for presupposition, as well as the fundamental observations that
underlie theories of politeness.
In natural language, negation functions as an operator ,
along with quantifers (see quantification ) and modals
(see modality ); operators are more basic and have more
properties than ordinary predicates or functors. In particu-
lar, operators have a scope ; that is, there is always some other
we would mistake it for water; but then, the argument continues,
because of its diferent essential properties (composition, in this
case), it is not water. And thus we are supposed to conclude that
the essential properties, even if unknown to us, determine the
meaning of our NKTs.
Nonetheless, this example is problematic. Ever since the
composition of water was discovered, it is among waters known
and defning properties. Accordingly, after this discovery, we
wouldnt consider XYZ water because we would know it isnt, and
so this case does not support the claim that essential unknown
properties are sometimes involved in NKTs meanings. On the
other hand, no case has been made for the claim that we would
have been mistaken if, before that discovery, we had considered
XYZ water. We would have been mistaken had we then claimed
that Twin Earths liquid has the same unknown composition as
Earths water; but the moot point is whether the claim that it is
water would then have implicitly involved such an additional
claim. Examination of actual similar cases does not support
Putnams contention; moreover, essentialism has been shown to
be problematic in additional respects as well (cf. Ben-Yami 2001,
and additional references there). So despite its current popular-
ity, essentialism remains far from established .
Hanoch Ben-Yami
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ben-Yami , H . 2001 . Te semantics of kind terms . Philosophical Studies
102 : 155 84.
Kripke , S. 1980. Naming and Necessity . Oxford : Blackwell .
Putnam , H . 1975 . Te meaning of meaning. In Mind, Language and
Reality: Philosophical Papers , II: 21571. Cambridge : Cambridge
University Press.
Schwartz , S. P . 2002 . Kinds, general terms, and rigidity . Philosophical
Studies 109 : 265 77.
NECESSARY AND SUFFICIENT CONDITIONS
If we take two conditions, A and B, and we take A to be a nec-
essary and suf cient condition (or set, thereof) for B, then con-
dition B cannot hold in the absence of condition A. However, if
A is merely a necessary condition (or set, thereof) for condition
B, then the presence of condition A does not entail condition B.
Further, if A is merely a suf cient condition (or set, thereof) for B,
then it is possible that condition B holds in the absence of condi-
tion A.
In the philosophy of language, necessary and suf cient con-
ditions have been employed by some philosophers in response
to the question: What are the conditions for the correct applica-
tion of a word?
Take our employment of the words fear and afraid . To say of
a person that he or she is afraid, we want to know how we might
judge whether that application is correct. We are here asking for
the truth conditions : Tat is, what needs to hold such that our
predication of Taitu-as-afraid is true? One answer might be that
Taitu must hold a belief such that she is subject to a threat and a
consequent desire to act in a way that will diminish that threat.
Alternatively, one might argue that predicating of Taitu that she
is afraid is conditional upon a particular sensation, or patterned
change in the autonomic nervous system, being elicited in Taitu.
Necessary and Suffcient Conditions Negation and Negative Polarity
555
Besides NPIs, English also has positive-polarity items
( would rather, sorta ) that dont occur in negative polarity con-
texts; possible polarity items ( tell time ) that can occur only
within the scope of a possible -type modal; and combinations,
like the impossible polarity item fathom that require both nega-
tive scope and a modal .
John M. Lawler
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Atlas , Jay D . 1996 . Only noun phrases, pseudo-negative generalized
quantifers, negative polarity items, and monotonicity. Journal of
Semantics 13 .4: 265 328. Negative polarity infltrates logic.
Baker , C. L . 1970 . Double negatives . Linguistic Inquiry 1 : 169 86.
Horn , Laurence R . 1969 . A presuppositional analysis of only and even .
Chicago Linguistics Society: CLS 5 : 97 108.
. 1989 . A Natural History of Negation . Chicago: University of
Chicago Press . Horns revision and extension of his 1972 dissertation.
Te classical neo-Gricean analysis.
Horn , Laurence R. , and Yasuhiko Kato . 2000 . Introduction: Negation and
polarity at the millennium. In Studies in Negation and Polarity , ed.
Laurence R. Horn and Yasuhiko Kato, 119. Oxford : Oxford University
Press . An excellent survey.
Israel , Michael . 2004 . Te pragmatics of polarity . In Te Handbook of
Pragmatics , ed. L. Horn and G. Ward , 70123. Oxford : Blackwell.
Klima , Edward S . 1964 . Negation in English. In Te Structure of
Language , ed. J. Fodor and J. Katz , 246323. Englewood Clifs,
NJ: Prentice Hall . Te frst modern syntactic/semantic study.
Ladusaw , William A . 1980 . Polarity Sensitivity as Inherent Scope Relations .
New York : Garland . Te origins of the downward-entailment theory,
using visual metaphors like scope and focus.
Lakof , Robin . 1970 . Some reasons why there cant be any some-any
rule . Language 45 : 608 15.
Lawler , John M . 1974 . Ample negatives. Chicago Linguistics Society: CLS
10 : 35777. After Klima, negative polarity was developed extensively in
the generative semantics tradition (Horn, Lakof, Ross, Lawler,
and many others), largely using a negative polarity feld metaphor,
along with negative triggers and secondary triggering.
Linebarger , Marcia . 1981 . Te grammar of negative polarity . Ph.D. diss.,
Massachusetts Institute of Technology. A then-orthodox generative
treatment, using rule-based metaphors like NPIs being licensed.
. 1991 . Negative polarity as linguistic evidence . Papers from the
Parasession on Negation . Chicago Linguistic Society: CLS 27 : 165 88.
McCawley , James D . 1993 . Everything Tat Linguists Have Always Wanted
to Know About Logic (But Were Ashamed to Ask) . Chicago: University
of Chicago Press. Oxford : Blackwell . McCawleys modern generative
semantic analysis; for example, In natural language, negation is not
truth-functional.
Ross , John R . 1973 . Negginess. Paper delivered to the Winter Meeting of
the Linguistic Society of America, San Diego.
van der Wouden , Ton . 1996 . Negative Contexts: Collocation, Polarity
and Multiple Negation . London and New York : Routledge . A useful
University of Groningen dissertation.
Zeijlstra , Hedde, and Jan-Philipp Soehn , eds. 2007. Proceedings of the
Workshop on Negation and Polarity . Tbingen : University Collaborative
Research Center . Recent evidence of polarity research expansion into
other languages.
NETWORK THEORY
Network theory concerns itself with the study of elements, called
vertices (e.g., words), and their connections, called edges or links
element either assumed or verbally present in the discourse
to which a negative, modal, or quantifer refers. Tat linked ele-
ment is said to be the focus or to be in the scope of the negative
(or modal; quantifers are said to bind rather than focus on
another element) .
Negation produces signifcant complexities and occasional
ambiguities when it interacts with other scope operators, because
the scopes can get twisted about. Every boy didnt leave is ambig-
uous, depending on the relative scope of the negative didnt and
the quantifer every (rather like Every boy read some book , where
two diferent quantifers produce ambiguity). Negation com-
bines in idiosyncratic ways with modals; for example, in You
may not go, and thats fnal! the deontic may not means not
possible, but in Tis may not be the place , the epistemic may
not means possibly not.
Every language develops its own idiomatic sets of negative
elements, and its own rules for using them. English negative phe-
nomena are by far the best studied; examples include syntactic
constructions ( Tis is it, isnt it? Not any big ones, he didnt ), vari-
ation ( so didnt I; aint got none ), morphology ( nt, -free, un- ),
(morpho)phonology ( do/dont ), intonations ( Riight ), and lex-
emes sporting negation overt ( never ), incorporated ( doubt, lack ),
calculated ( few ), entailed ( prohibit ), or presupposed ( only ) .
Included also is a large, complex, and diverse system of nega-
tive polarity items (NPIs like ever in He didnt ever see it ), which
felicitously occur only in the scope of some negative element
( *He ever saw it ). Te details of what scope actually is, and of how
and which and why NPIs can occur within it, vary among specifc
negative and NPI elements.
Negative polarity is a variety of negative concord (e.g., French
Je ne regrette rien ; literally I do nt regret nothing ; Yiddish Ix
hob nit kin gelt ; literally I do nt have no money), but instead
of negative concord, which uses negative elements in the focus
of another negative, negative polarity uses other, non-negative
elements, which can sometimes pick up negativity by association
and occur without overt negative ( could care less < could nt care
less ). An interesting typological question is whether languages
like English that lack signifcant negative concord develop more
negative polarity phenomena to compensate .
NPI is a term applied to lexical items, fxed phrases, or syn-
tactic construction types that demonstrate unusual behavior
around negation. NPIs might be words or phrases that occur
only in negative-polarity contexts ( fathom , in weeks ) or have an
idiomatic sense in such contexts ( not too bright, drink a drop );
or they might have a lexical afordance that only functions in
such contexts ( need/dare (not ) reply ); or a specifc syntactic rule
might be sensitive to negation, like subject - verb inversion with
adverb fronting in Never/*Ever/*Frequently have I seen such a
thing.
Te grammatical occurrence of NPIs in an utterance is prima
facie evidence that it contains some sort of negation, and this
allows NPIs to function as indicators for various types of seman-
tic opposition and syntactic structure. Tis has turned out to be a
sensitive tool in other research areas of linguistics, and linguists
using NPIs have discovered many covert negative phenomena;
for instance, NPIs can also occur in questions ( Have you ever
been there? ), hypothetical clauses ( Tell me if he ever arrives ), and
comparatives ( Hes better than we ever expected ) .
Negation and Negative Polarity Network Theory
556
the small world phenomenon, high clustering, and heteroge-
neous degree distribution are common properties of linguistic
networks (Mehler 2008). Most models are based on the prefer-
ential attachment principle proposed by Albert Lszl Barabsi
and Rka Albert (1999): Vertices with many connections are
more likely to become more connected in the future than those
with few connections (Steyvers and Tenenbaum 2001, 2005;
Dorogovtsev and Mendes 2001; Motter et al. 2002) .
Te challenges of the application of network theory are to
explain the properties of these networks (most studies are merely
descriptive); to incorporate deeper statistical techniques, for
example, degree correlation analysis (Serrano et al. 2007); and
to extend the studies to more languages (most studies are in
English). For these reasons, it is too early to argue that the het-
erogeneous degree distributions and other statistical patterns
constitute laws of language in the sense of absolute and
statistical universals . When applied to syntactic networks,
network theory has helped to explain the origins of the properties
of the syntactic dependency structure of sentences, for example,
the exceptionality of syntactic dependency crossings (Ferrer
i Cancho 2006) and has provided new tracks for understand-
ing syntax at the large scale of syntactic organization (Ferrer i
Cancho, Sol, and Khler 2004), above the traditional sentence
level (see syntax, universals of) .
In their pioneering application of network theory, Mark
Steyvers and Joshua B. Tenenbaum (2001, 2005) studied the
large-scale organization of various kinds of semantic networks
(e.g., word association networks, as in Figure 1) and proposed
a simple model for explaining the small-worldness, high clus-
tering and a heterogeneous degree distribution of semantic
networks. Over time, new vertices (e.g. words) are added and
attached to existing vertices according to two principles: a)
Barabsi and Alberts preferential attachment and b) diferentia-
tion. Diferentiation means that a new vertex tends to mimic the
connectivity pattern of an existing vertex .
Network theory has shed new light on the evolution of lan-
guage by defning the necessary conditions for the existence of
language (e.g., word ambiguity ) and also by suggesting the
possibility that language could have appeared for free as a side
(e.g., two words are connected if one word has been elicited by
the other in a word association experiment; see Figure 1). Tis
theory has many applications in language sciences and is the
outcome of intersecting work of mathematicians and physicists,
who usually call it graph theory (Bollobs 1998) or complex
network theory (Newman 2003), respectively.
One of the major contributions of physicists has been to
unravel the statistical properties of real networks (Newman
2003), for example, the World Wide Web or protein interaction
networks . Firstly, physicists discovered that practically all real
networks exhibited the small world phenomenon. Te term
small world comes from the observation that everyone in the
world can be reached through a short chain of social acquain-
tances, although the number of people in the whole social net-
work is huge. In the word association network, partially shown
in Figure 1, volcano is reached from ache through a chain of at
least four links, while only one link separates fre from volcano .
Secondly, physicists found that many real networks had a het-
erogeneous degree distribution. Loosely speaking, this property
means that there are vertices (words) with a disproportionately
large number of connections (the so-called hubs). For instance,
in the network partially shown in Figure 1, the fve words with the
highest degrees are food , money , water , car , and good (Steyvers
and Tenenbaum 2005). Finally, another fundamental property of
real networks is clustering; that is, roughly speaking, if two verti-
ces are connected to the same vertex they are likely to be directly
connected as well .
Network theory has contributed to the study of language in
three ways: a) by characterizing the statistical properties of lin-
guistic networks, such as networks of word association (Steyvers
and Tenenbaum 2005), thesauri (Sigman and Cecchi 2002),
and syntactic dependencies (Ferrer i Cancho, Sol, and Khler
2004); b) by modeling the properties of these networks (Steyvers
and Tenenbaum 2005; Motter et al. 2002); and c) by proposing
abstract models that provide a further understanding of the fac-
ulty of language (Ferrer i Cancho, Riordan, and Bollobs 2005).
Although the systematic application of network theory to lan-
guage is a young feld (starting in the early twenty-frst century)
within quantitative linguistics , it can be concluded that
Figure 1. A subset of a word association network appearing in
Steyvers and Tenebaum (2005, 50). Links go from the stimu-
lus to the response word. Reproduced by permission of the
Cognitive Society, Inc., copyright 2005.
Network Theory
557
Design complexity is one of the hallmarks of adaptive function
and, thus, inquiry into the brain systems that support the lan-
guage faculty is mandatory if one wishes to understand the evo-
lutionary history and potential adaptive functions of language.
Neurochemical, neurophysiological, and neuroanatomical
studies defne a widely distributed neural network that supports
speech and language functions. Tis network includes the motor
and supplementary motor area (SMA) of the prefrontal lobes;
brocas area in the dorsal prefrontal region; wernickes
area in the medial temporal lobe ; the anterior cingulate
gyrus and the subcortical basal ganglia; and the periaqeduc-
tal gray matter (PAG). Te anterior cingulate gyrus sends efer-
ents directly onto the PAG central gray and appears to infuence
the initiation and voluntary control of vocalization. Destruction
of the central gray substance at the subcortical level or the SMA at
the cortical level can cause mutism. Patients with bilateral lesions
within the cingulate area often undergo a period of mutism, fol-
lowed by slow recovery during which speech is aprosodic and
initiation of speech is rare. Te anterior cingulate gyrus receives
eferents from the dopamine-rich supplementary motor area in
the cortex and sends aferents, along with other dopaminergic
fbers coming from the basal ganglia, up and into the prefrontal
regions. Tus, all of these language-related areas are intercon-
nected via dopaminergic fbers and the prefrontal cortex .
Te prefrontal cortex (PFC) constitutes approximately one-
third of the human cortex and is the last part of the human
brain to become fully myelineated in ontogeny, with maturation
occurring in late childhood/early adolescence (Huttenlocher
and Dabholkar 1997). Te PFC receives projections from the
mediodorsal nucleus and encompasses primary motor cortex,
as well as premotor, supplementary motor, and the dorsal and
orbital sectors of the prefrontal (proper) lobes. All of these PFC
areas are addressed by mesocortical dopaminergic projections
and play a role in language functions .
Dopamine (DA) is manufactured in the pigmented neurons of
the substantia nigra (SN) and the ventral tegmental area (VTA).
Tere are three major ascending dopaminergic systems: the stri-
ato-nigral tract, which ascends from the SN to the corpus stria-
tum; the mesolimbic system, which ascends from the SN and
medial VTA to limbic sites, including the cingulate gyrus; and the
mesocortical system, which ascends from the anteromedial teg-
mentum and VTA to neocortical sites, including supplementary
motor area and prefrontal cortex (Nieoullon 2002; Girault and
Greengard 2004). Important language regions are linked directly
to these dopaminergic frontal lobe structures. Brocas area, for
example, is in the frontal lobes , as is the SMA. Posterior
language sites, such as Wernickes area, the angular gyrus, and
the inferior and superior parietal lobules, are densely intercon-
nected via the superior and inferior longitudinal fasiculi, with
meso-prefrontal dopaminergic systems .
Language-related semantic and working memory net-
works in humans can be modulated by dopaminergic stimulation
(Williams and Goldman-Rakic 1995; Kischka et al. 1996; Luciana
and Collins 1997; Jay 2003; Angwin et al. 2004). Dopaminergic
activation may even support key components of the sentence
comprehension system in patients with Parkinsons disease
(PD) (Grossman et al. 2001). Dopaminergic agents may be efect-
ive treatment for nonfuent aphasia . M. Albert and colleagues
efect of communication principles (Ferrer i Cancho, Riordan,
and Bollobs 2005) .
Ramon Ferrer i Cancho
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bibliography on linguistic and cognitive networks. Available
online at: http://www.lsi.upc.edu/~rferrericancho/linguistic_and_
cognitive_networks.html.
Bollobs , Bla . 1998 . Modern Graph Teory . New York : Springer . A help-
ful introduction to graph theory.
Barabsi , Albert Lszl , and R ka Albert . 1999 . Emergence of scaling in
random networks Science 286 : 509 12.
Dorogovtsev , Sergey , and Jos Fernando Mendes . 2001 . Language as an
evolving word web . Proceedings of the Royal Society of London Series
B , Biological Sciences 268 : 2603 6.
Ferrer i Cancho , Ramon . 2006 . Why do syntactic links not cross?
Europhysics Letters 76 : 1 228 34.
Ferrer i Cancho , Ramon, Oliver Riordan , and B la Bollobs . 2005 .
Te consequences of Zipfs law for syntax and symbolic reference .
Proceedings of the Royal Society of London Series B 272 : 561 5.
Ferrer i Cancho , Ramon, Ricard V. Sol , and Reinhard K hler . 2004 .
Patterns in syntactic dependency networks . Physical Review E
69 : 051915.
Mehler , Alexander . 2008 . Large text networks as an object of corpus lin-
guistic studies. In Corpus Linguistics: An International Handbook of
the Science of Language and Society , ed. Anke Ldeling and Merja Kyt ,
32882 . Berlin and New York : de Gruyter .
Motter , Adilson E., Alessandro P. S. de Moura, Ying-Cheng Lai , and
Partha Dasgupta . 2002 . Topology of the conceptual network of lan-
guage . Physical Review E 65 : 065102.
Newman , Mark . 2003 . Te structure and function of complex networks .
SIAM Review 45.2 : 167 256. A helpful introduction to complex network
theory.
Serrano , Mari ngeles, Marian Bogu , Romualdo Pastor-Satorras ,
Alessandro Vespignani . 2007 . Correlations in complex networks.
In Large Scale Structure and Dynamics of Complex Networks: From
Information Technology to Finance and Natural Science , ed. Guido
Caldarelli and Alessandro Vespignani , 3565. Singapore : World
Scientifc .
Sigman , Mariano , and Guillermo A. Cecchi . 2002 . Global organization of
the Wordnet lexicon . Proceedings of the National Academy of Sciences
USA 99 : 1742 7.
Steyvers , Mark , and Joshua B. Tenenbaum. 2001. Te large-scale struc-
ture of semantic networks: Statistical analyses and a model of seman-
tic growth . Available online at: http://arxiv.org/ftp/cond-mat/
papers/0110/0110012.pdf. Tis is the frst version of their 2005 journal
article.
. 2005 . Te large-scale structure of semantic networks: Statistical
analyses and a model for semantic growth . Cognitive Science
29.1 : 41 78.
NEUROCHEMISTRY AND LANGUAGE
Why study a potential neurochemistry of language? Tere are
two reasons, one practical, the other theoretical. First, pharma-
cologic treatments of various speech and language disorders
depend on, and will be enhanced by, an understanding of how
selective neurochemical networks facilitate, inhibit, or mediate
language functions. Second, understanding the adaptive func-
tion of a given trait requires detailed knowledge of the design
specifcations that mediate or implement the trait in question.
Neurochemistry and Language
558
(i.e., dopamine transporter reuptake, as in the striatum) is largely
nonexistent. In humans, the COMT gene contains a highly func-
tional and common variation in its coding sequence that appears
to be a unique human mutation because it has not been found
in great apes. Tis uniquely human change in dopaminergic
functional capacity in the prefrontal cortex suggests that it may
have been a factor in the evolution of the human prefrontal cor-
tex and thereby of human speech and language functions more
generally .
An inherited defcit in spoken grammatical language among
several members of a family (family KE) in England has been
associated with a mutation in the forkhead box P2 ( FOXP2 ) gene
on chromosome 7 (see genes and language ). Persons with
the FOXP2 mutation evidence underactivity in dopaminergic
neural networks linking subcortical striatal networks with pre-
frontal cortical sites, including Brocas area during word-genera-
tion tasks. FOXP2 is subject to the efects of genomic imprinting,
with relatively high expression from the paternal chromosome
(Feuk et al. 2006). Such a pattern of gene expression evolves
in the context of evolutionary confict due to paternity uncer-
tainty in polygynous mating systems, as is the case with most
mammals including humans. Genetic confict occurs between
asymmetrically related kin (i.e., between mothers and ofspring,
and between siblings in the context of paternity uncertainty),
with genes that are paternally expressed in ofspring promoting
behaviors in ofspring that are designed to monopolize resources
from the mother and exclude resources going to siblings. Te
FOXP2 -related defect implies that some aspects of spoken lan-
guage may have evolved under pressures of genetic confict.
Tis confict view of the evolution of language is consistent
with recent fndings linking handedness and cognitive/language
defcits of schizophrenics (another disorder involving dopamin-
ergic dysfunction) to the parent of origin efects (Francks et al.
2003), as well as other evidence identifying potential imprinting
efects on genes that regulate dopaminergic systems of the lan-
guage-related areas of the prefrontal cortex. In short, investiga-
tion of dopaminergic infuences on language functions leads us
into two seemingly disparate realms of inquiry: 1) the develop-
ment of rational pharmacotherapeutic strategies for treatment
of language disorders (e.g., dopaminergic drugs for fuency
disorders), and 2) reconstruction of the evolutionary conficts
that led to the emergence of speech and language functions
themselves .
Patrick McNamara
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Albert , M . 2000 . Towards a neurochemistry of naming and anomia. In
Language and the Brai n , ed. Y. Grodzinsky , L. Shapiro, and D. Swinney ,
15765. San Diego, CA : Academic Press .
Albert , M., D. L. Bachman, A. Morgan , and N. Helm-Estabrooks . 1988 .
Pharmacotherapy for aphasia . Neurology 38 : 877 9.
Angwin , A. J., H. J. Chenery, D. A. Copland, W. L. Arnott, B. E. Murdoch ,
and P. A. Silburn . 2004 . Dopamine and semantic activation: An
investigation of masked direct and indirect priming . Journal of the
International Neuropsychological Society 10.1 : 15 25.
Bannon , M. J., E. B. Bunney , and R. H. Roth . 1981 . Mesocortical dopa-
mine neurons: Rapid transmitter turnover compared to other brain
catecholamine systems . Brain Research 218.1 : 376 82.
(1988), using an on/of design, reported improved fuency
and naming scores in a patient with nonfuent aphasia treated
with bromocriptine (a drug that stimulates selected dopamine
receptors in the brain). Fluency and naming scores returned to
baseline after the drug was discontinued. S. R. Gupta and A. G.
Mlcoch (1992) replicated the efect of improved fuency scores
after bromocriptine in two aphasic patients, but L. Sabe and
colleagues (1995) and D. L. MacLennan and colleagues (1991)
could not document any improvement in speech and language
scores in nonfuent aphasics who were treated with bromocrip-
tine late in the recovery process . Y. Tanaka and D. L. Bachman
(2000) conducted a double-blind, crossover study with bromo-
criptine. Tey administered the drug (57.5 mg/day for four
weeks) to 10 patients with a Broca-type aphasia. Statistically sig-
nifcant improvement (pre- to posttreatment) on naming and
fuency scores was documented in the mild aphasics, but not
in the severely impaired aphasics. M. Bragoni and colleagues
(2000) used a double-blind, placebo-controlled design focused
only on chronic nonfuent aphasics at a dosage of 30 mg/day,
with participants maintained at that dosage for three months.
While signifcant gains in verbal fuency were evidenced with
bromocriptine, these fndings are based on the performance of
only fve participants. Consistent, however, with the claim of
positive dopaminergic efects on fuency is the fact that the dopa-
minergic drug levodopa (LD) has also demonstrated benefcial
efects on speech fuency in midstage patients with Parkinsons
disease (McNamara and Durso 2000). S. Knecht and colleagues
(2004) showed that healthy volunteers given 100 mg of LD per
day exhibited more rapid and more accurate learning of verbal-
visual associations than a group of controls given a placebo.
Learning efects in this carefully controlled study could not be
attributed to changes in arousal, autonomic function, motor
response times, afect, or response biases .
Acetylcholine, one of the neurotransmitters that interacts
with dopaminergic systems at the level of the cortex, has also
been implicated in language functions. Tanaka, M. Miyazaki,
and Albert (1997) documented naming and comprehension
improvement in fuent aphasics using the cholinergic agent bife-
melane. Tey built on the work of L. Moscowitch, P. McNamara,
and Albert (1991) who reported that an anticholinesterase agent
(which boosts cholinergic activity) improved language perform-
ance in eight fuent semantic aphasics. Albert (2000) provides
an in-depth discussion of both dopaminergic infuences on
nonfuent aphasia and cholinergic infuences on fuent aphasia.
His conclusion is that there appears to be a strong and consist-
ent efect of dopaminergic agents on verbal fuency and mild
efects of cholinergic agents on naming and semantic memory.
Dopaminergic efects, however, are better documented than
cholinergic efects .
One of the major regulatory genes that control dopamines
metabolic pathways in the prefrontal cortex is the gene that
codes for the enzyme catechol-O-methyltransferase (COMT).
Signifcant associations between COMT variations with varia-
tions in prefrontal cognitive function have been identifed (Egan
et al. 2001; Joober et al. 2002). Studies in rats, knockout mice,
and monkeys suggest that COMT is of particular importance
with respect to intrasynaptic dopamine regulation in the pre-
frontal cortex, where an alternative route of dopamine removal
Neurochemistry and Language
559
Nieoullon , A . 2002 . Dopamine and the regulation of cognition and atten-
tion . Progress in Neurobiology 67 : 52 83.
Sabe , L., F. Salvarezza, A. Garcia Cuerva, R. Leiguarda , and S. Starkstein.
1995 . A randomized, double-blind, placebo-controlled study of bro-
mocriptine in nonfuent aphasia . Neurology 45 : 2272 4.
Tanaka , Y. , and D. L. Bachman . 2000 . Pharmacotherapy of Aphasia. In
Neurobehavior of Language and Cognition: Studies of Normal Aging
and Brain Damage , ed. and M. Albert , L. Connor, and L. Obler , 15978.
Boston : Kluwer Academic .
Tanaka , Y., M. Miyazaki , and M. Albert . 1997 . Efects of cholinergic activ-
ity on naming in aphasia . Lancet 350 : 1161 7.
Tierry , A. M., J. P. Tassin, G. Blanc, L. Stinus, B. Scatton , J. Glowinski .
1977 . Discovery of the mesocortical dopaminergic system: Some
pharmacological and functional characteristics . Advances in
Biochemistry and Psychopharmacology 16 : 5 12.
Williams , G. V. , and P. S. Goldman-Rakic . 1995 . Modulation of mem-
ory felds by dopamine D1 receptors in prefrontal cortex . Nature
376 : 572 5.
NEUROIMAGING
Neuroimaging technologies provide a major source of new data
about how the language system is organized in the brain. In par-
ticular, activation imaging approaches, in which brain activity is
monitored while subjects perform some language task, allow us
to visualize various aspects of language processing in the normal
brain and test hypotheses about component language functions
or systems. Among the imaging techniques most commonly in
use for understanding language in the brain are structural mag-
netic resonance imaging (MRI), functional MRI (fMRI), positron
emission tomography (PET), electroencephalography (EEG),
and magnetoencephalography (MEG).
Structural Magnetic Resonance Imaging
Te lesion defcit model, where one deduces the function of
a brain region by observing what it cannot do when damaged,
marks the basis of our understanding of language organization
in the brain, originating with Paul Brocas work . Whereas Broca
had to wait until his patients death to determine where the
lesion was located, structural or conventional MRI scanning
allows lesion-behavior correlation in vivo.
Te MRI scanner is essentially composed of a large, high feld
magnet that delivers magnetic pulses and records small changes
in the magnetized atoms in your brain or body. Tese signals are
picked up by an antenna and, through several transformations, are
translated into pictures of the brain. By altering the direction, fre-
quency, and readout times of these magnetic perturbations, difer-
ent MRI pulse sequences produce variations in the signals generated
by each tissue type. Te diference between these signals is referred
to as contrast, and using these variations, the radiologist can deter-
mine what is normal brain tissue, what looks like a clot of blood or a
fatty tumor, what tissue has had a disruption in the normal difusion
of water molecules, and so on. MRI has excellent spatial resolution ,
which refers to the precision with which one can see details. Typical
MRI scans resolve 1 mm; thus, it is easy to locate even small brain
lesions that might explain a particular abnormality.
In the past few years, the sophistication of structural MRI
techniques has increased markedly, ofering new approaches
for identifying structure-function correlations in the language
Bragoni , M., M. Altieri, V. Di Piero, A. Padovani, C. Mostardini , and
G. L. Lenzi . 2000 . Bromocriptine and speech therapy in non-fuent
chronic aphasia after stroke. Neuroscience 21.1 : 19 22.
Egan , M. F., T. E. Goldberg, B. S. Kolachana, J. H. Callicott,
C. M. Mazzanti, R. E. Straub, D. Goldman ,and D. R. Weinberger . 2001 .
Efect of COMT Val108/158 Met genotype on frontal lobe function
and risk for schizophrenia . Proceedings of the National Academy of
Sciences USA 98 : 6917 22.
Feuk , L., A. Kalervo, M. Lipsanen-Nyman, J. Skaug, K. Nakabayashi,
B. Finucane, D. Hartung, M. Innes, B. Kerem, M. J. Nowaczyk,
J. Rivlin, W. Roberts, L. Senman, A. Summers, P. Szatmari, V. Wong,
J. B. Vincent, S. Zeesman, L. R. Osborne, J. O. Cardy, J. Kere, S. W.
Scherer , and K. Hannula-Jouppi . 2006 . Absence of a paternally
inherited FOXP2 gene in developmental verbal dyspraxia . American
Journal of Human Genetics 79 : 965 72.
Francks , C., L. E. DeLisi, S. H. Shaw, S. E. Fisher, A. J. Richardson,
J. F. Stein , and A. P. Monaco . 2003 . Parent-of-origin efects on hand-
edness and schizophrenia susceptibility on chromosome 2p12-q11 .
Human Molecular Genetics 12 : 3225 30.
Gardner , E. L. , and C. R. Ashby , Jr . 2000 . Hetereogeneity of the meso-
telencephalic dopamine fbers: Physiology and pharmacology .
Neuroscience and Biobehavioral Reviews 24 : 115 28.
Girault , J. A. , and P. Greengard . 2004 . Te neurobiology of dopamine
signaling . Archives of Neurology 61 : 641 4.
Greener , J., P. Enderby , and R. Whurr . 2002 . Pharmacological treatment
for aphasia following stroke . Cochrane Database Systematic Reviews
4 : CD000424.
Gupta , S. R. , and A. G. Mlcoch . 1992 . Bromocriptine treatment of non-
fuent aphasia . Archives of Physical Medicine and Rehabilitation
73.4 : 373 6.
Grossman , M., G. Glosser, J. Kalmanson, J. M. Morris, M. B. Stern ,
H. I. Hurtig . 2001 . Dopamine supports sentence comprehen-
sion in Parkinsons Disease . Journal of the Neurological Sciences ,
184.2 : 123 30.
Huttenlocher , P. R. , and A. S. Dabholkar . 1997 . Regional diferences in
synaptogenesis in human cerebral cortex . Journal of Comparative
Neurology 387.2 : 167 78.
Jay , T. M . 2003 . Dopamine: A potential substrate for synaptic plasticity
and memory mechanisms . Progress in Neurobiology 69 : 375 90.
Joober , R., J. Zarate, G. Rouleau, E. Skamene , and P. Boksa . 2002 .
Provisional mapping of quantitative trait loci modulating the acous-
tic startle response and prepulse inhibition of acoustic startle .
Neuropsychopharmacology 27 : 765 81.
Kischka , U., T. Kammer, S. Maier, M. Weisbrod, M. Timm , and
M. Spitzer . 1996 . Dopaminergic modulation of semantic network acti-
vation . Neuropsychologia 34 : 1107 13.
Knecht , S., C. Breitenstein, S. Bushuven, S. Wailke, S. Kamping, A. Floel,
P. Zwitserlood , and B. Ringelstein . 2004 . Levodopa: Faster and better
word learning in normal humans . Annals of Neurology 56.1 : 20 6.
Luciana , M. , and P. Collins . 1997 . Dopaminergic modulation of working
memory for spatial but not object cues in normal humans . Journal of
Cognitive Neuroscience 9 : 330 47.
MacLennan , D. L., L. E. Nicholas, G. K. Morley , and R. H. Brookshire .
1991 . Te efects of bromocriptine on speech and language function
in a man with transcortical motor aphasia. In Clinical Aphasiology ,
ed. T. E. Prescott , 14555. Boston : College Hill .
McNamara , P. , and R. Durso . 2000 . Language functions in Parkinsons
disease: Evidence for neurochemistry of language. In Neurobehavior
of Language and Cognition: Studies of Normal Aging and Brain
Damage , ed. L. Obler and L. T. Conner , 20112. New York : Kluwer
Academic .
Moscowitch , L., P. McNamara , and M. L. Albert . 1991 . Neurochemical
correlates of aphasia . Neurology 41 (Supplement 1): 410.
Neurochemistry and Language Neuroimaging
560
of tractography, where the WM tracts in and out of a language
area have been mapped. Tis approach can identify inputs and
outputs to language regions, indicating possible mechanisms for
distal efects of local lesions through disruption on the connect-
ing pathways .
Activation Imaging and Neurovascular Coupling
Several brain imaging technologies take advantage of neuro-
vascular coupling in identifying brain regions associated with
language performance. Neurovascular coupling refers to the
fact that when neurons increase their fring rate (because that
brain area is working harder), blood fow increases to that
region. Typically, the correlation between blood fow and neu-
ronal activity is extremely high, although the blood fow increase
is delayed in onset by several seconds and falls of gradually in
comparison to neuronal activity (Buxton et al. 2004). In several
clinical conditions it may be decoupled, such as in acute stroke.
Both PET and fMRI take advantage of neurovascular coupling to
identify brain activity.
POSITRON EMISSION TOMOGRAPHY. PET scanning is an imag-
ing tool in which a radioactively labeled compound is injected
into the body and taken up in the brain. Compounds such as
glucose or water are labeled with positrons that rapidly decay;
during the decay process the positrons hit other nuclei and are
annihilated, which causes the emission of two photons that
shoot of in opposite directions simultaneously. Te PET scan-
ner is composed of a ring of detectors that detect these simul-
taneous photons, and software reconstructs their originating
positions, revealing where the compound traveled. Te resulting
PET image is a blurry picture showing the amount of radioactive
substance reaching every pixel in the brain. Diferent radioactive
compounds measure various brain processes, each with a char-
acteristic half-life. For instance, 18-fuorodeoxyglucose (
18
FDG)
has a half-life of about 45 minutes and measures glucose metab-
olism. More useful for language research is the compound H
2
15
O,
radioactive water. Because of its short half-life (about 2 minutes),
H
2
15
O scans can be repeated after a delay of 1012 minutes, up to
between 6 and 10 scans. For this reason, H
2
15
O has been used for
system. One approach involves warping scans from diferent
patients together, showing areas, for example, where lesions
overlap among patients with the same language impairment. An
illustrative study comes from Bates et al. (2003). Here, the authors
used a technique called voxel based morphometry . For each voxel
(a three-dimensional pixel) in the brain, a t-test compares the
extent to which patients with a lesion encompassing that voxel
difer signifcantly from subjects whose lesions do not encom-
pass that area on a language task. E. Bates and colleagues (2003)
correlated lesion location with verbal fuency (Color Plate 5) .
Te areas of the brain showing the most signifcant diferences
between groups are depicted in red, indicate the brain area that
is most likely responsible for the defcit observed. Contrary to
Brocas report, reduced verbal fuency was associated not pri-
marily with brocas area lesions but, rather, with lesions of
the insula and underlying white matter. Indeed, a recent MRI
study of Brocas patient confrmed involvement of these struc-
tures (see Figure 1).
Another approach in MRI analysis compares aspects of brain
structure, such as gray matter thickness or sulcal position, to per-
formance on language tasks using voxel-based correlations. For
instance, L. Lu and colleagues (2007) examined the relationship
between the thickness of the cortex in the left inferior frontal
region and ability on a phonological processing task in children.
Te development of gray matter changes in this area correlated
with improving scores on phonology tasks, indicating a dynamic
relationship between emerging brain growth and language
development (Color Plate 6) .
A third structural MRI approach examines the integrity of the
white matter underlying the cortical ribbon. Difusion tensor
imaging is an MRI approach that measures the difusion of water
molecules in the brain. White matter (WM) fbers tend to be bun-
dled together, lined up in parallel sheaths. Because water is more
likely to difuse in parallel to these white matter tracts as opposed
to crossing them on the perpendicular, imaging techniques that
track difusion will tend to emphasize the direction of these fber
tracts. Image-processing techniques can identify the unifor-
mity of these fber directions in each voxel in the brain, indicat-
ing whether the WM is intact. Color Plate 7 shows an example
Figure 1. Top left: Original drawing by Broca.
Bottom left: preserved whole brain of Brocas
patient. Right: axial MRI slice through Brocas area
showing damage to insula, striatum, and underlying
white matter (Dronkers et al. 2007).
Neuroimaging
561
processes of nearly any complexity, resulting in many important
fndings. For instance, in the reading system, years of controversy
about whether reading involves a single system versus parallel
systems have been largely resolved. Te physical presence of two
anatomically distinct pathways diferentially engaged by difer-
ent reading demands and subject groups lends strong credence
to the dual-route hypothesis (Pugh et al. 1996). Another interest-
ing set of fndings explores the role of the right hemisphere
in many aspects of language processing, including afective
prosody , metaphor analysis, and contextual processing (see
Bookheimer 2002 for a review) . Within the frontal lobe there
appear to be separate regions for processing aspects of expres-
sive language, including phonological processing, syntax, and
semantic integration. Tus, fMRI appears to have resolved an
ongoing debate over whether part of Brocas area is specialized
for syntax, as opposed to secondary to increased working mem-
ory demands of complex syntactic structures, with a recent study
indicating syntax specifcity in this region (Santi and Grodzinsky,
2007). In general, language research from fMRI indicates a level
of organization that is far more complex, detailed, and specifc
than envisioned on the basis of lesion-defcit studies. Color Plate
9 shows an fMRI exam during a series of language tasks. Even
within a single subject, clear evidence for at least nine diferent
brain regions contributing to language can be observed .
EEG and MEG. While fMRI ofers a tremendous advantage in
both spatial and temporal resolution over PET, the fMRI response
is extremely sluggish in comparison to neural activity measured
directly. Two other technologies ofer vastly improved temporal
resolution. Electroencephalography measures the combined
electrical activity of a wide area of the brain. Because electrical
activity directly measures neural fring, EEG runs very close to
real-time activity of neurons. It also has several major disadvan-
tages. Te spatial resolution is poor: Electrical signals represent
an average over many centimeters of activity, and results are gen-
erally confned to entire lobes. Further, the signals come mostly
from surface brain structures. Finally, data must be averaged over
many trials to yield an averaged electrical response to a class of
stimuli. Nonetheless, EEG is the method of choice for high tem-
poral resolution work, and for young children or infants.
Years of EEG research reveal expected patterns of electri-
cal responses to certain classes of stimuli. Averaged electrical
responses to a class of stimuli are referred to as event- related
potentials or ERPs. An example is the N400 response, meaning
a negatively directed signal occurring 400 milliseconds after
the stimulus. Te N400 is found when the subject experiences
an anomalous or unexpected event. For instance, A. Hahne,
K. Eckstein, and A. Friederici (2004) examined ERPs in response
to semantic and syntactic violations embedded within sen-
tences. When there was a syntactic violation, the authors found a
component in anterior brain regions, termed the ELAN (early left
anterior negativity). Tis was followed by a late positive response
(P600) . Semantic violations produced the N400 response. In a
second experiment, instructions to ignore the syntactic viola-
tions and focus on the semantic task could not override the brain
response to syntactic violations. Tese data indicate that syn-
tactic processing is mandatory and not under efortful control.
While EEG cannot locate the brain structures that generate these
language activation studies, where subjects might receive sev-
eral injections while performing one or more language tasks and
during control tasks (Color Plate 8).
One of the frst PET studies to visualize language areas in vivo
was that of S. E. Petersen and colleagues (1988). In this study,
normal volunteers performed a series of language tasks ordered
hierarchically: viewing a crosshair on a screen; seeing printed
words on a screen or hearing words over headphones, reading or
repeating heard words, or generating an action verb correspond-
ing to a visually or auditorally presented noun. By subtracting
lower-level tasks from higher-order language tasks, the authors
isolated areas of the brain involved in word generation, while
removing unwanted efects of sensory stimulation. Tis sub-
tractive logic forms the basis of activation imaging experiments.
While there are theoretical dif culties with assumptions of hier-
archical organization, cognitive subtraction models remain the
mainstay of activation imaging research.
An important disadvantage of PET is the need to expose sub-
jects to radioactivity; a second disadvantage is that subjects must
perform the same task for several minutes continuously to obtain
a single brain image. Also, PET is a relatively noisy methodol-
ogy, and the signal-to-noise ratio is low enough that scans must
be averaged over a group of subjects. Finally, the spatial resolu-
tion of PET is low, usually about 6 mm, so that individual brain
structures cannot be resolved and the areas of signifcant acti-
vation are not easily localized to a specifc brain structure. Most
investigators solve this problem by performing an MRI scan for
each subject, mathematically moving or registering the brain
images so they are in the same space, then overlaying the PET
activation regions onto the corresponding MRI scans to localize
the regions of activity .
FUNCTIONAL MRI. In the early 1990s, two groups independently
discovered that blood fow increases during neural activity
could be measured directly with MRI (Kwong et al. 1992; Ogawa
et al. 1992). Tis is due to the accident that oxygenated blood and
deoxygenated blood have slightly diferent magnetic properties.
During increased brain activity, an increase in blood fow is not
matched by an increase in oxygen consumption; consequently,
more oxygenated blood spills over to the venous side of the cap-
illary bed. Scans of the brain taken during this state of increased
oxyhemoglobin concentration have slightly higher MRI signals
than those taken in the resting state. Tus, fMRI measures this
blood-oxygen-level dependent, or BOLD, signal when compar-
ing scans taken in diferent cognitive states. Tanks to the discov-
ery of ultrafast MRI scanning, typically using an approach called
echo-planar imaging or EPI, fMRI takes a complete picture of the
brain as quickly as once per second. By taking complete brain
volumes every few seconds over a period of several minutes or
more, fMRI can track those magnetic changes that correlate with
blood fow during experimental and control tasks that can be
varied with tremendous experimental complexity. Because fMRI
has signifcantly greater spatial and temporal resolution than
PET, it is often possible to see signifcant language-related brain
activity within a single individual in a matter of minutes.
Te study of language organization in the brain has been
revolutionized by fMRI. Because these studies are relatively easy
and inexpensive to conduct, it is possible to examine language
Neuroimaging
562
imaging of human brain activity during primary sensory stimulation .
Proceedings of the National Academy of Science 89 .12: 5675 9.
Lee , A., V. Kannan , and A. E. Hillis. 2006 . Te contribution of neuroim-
aging to the study of language and aphasia . Neuropsychology Reviews
16 .4: 171 83.
Lu , L., C. Leonard, P. Tompson, E. Kan, J. Jolley, S. Welcome, A. Toga ,
and E. Sowell . 2007 . Normal developmental changes in inferior fron-
tal gray matter are associated with improvement in phonological pro-
cessing: A longitudinal MRI analysis . Cerebral Cortex 17 .5: 1092 9
Ogawa , S., D. W. Tank, R. Menon, J. M. Ellermann, S. G. Kim, H. Merkle , and
K. Ugurbil . 1992 . Intrinsic signal changes accompanying sensory stimu-
lation: Functional brain mapping with magnetic resonance imaging .
Proceedings of the National Academy of Science 89.13 : 5951 5.
Petersen , S. E., P. T. Fox, M. Mintun , and M. E. Raichle . 1988 . Positron
emission tomographic studies of the cortical anatomy of single-word
processing . Nature 331 .6157: 585 9.
Pugh , K. R., B. A. Shaywitz, S. E. Shaywitz, R. T. Constable, P. Skudlarski,
R. K. Fulbright, R. A. Bronen, D. P. Shankweiler, L. Katz, J. M. Fletcher ,
J. C. Gore . 1996 . Cerebral organization of component processes in
reading . Brain . 119.4 : 1221 38.
Salmelin , R., P. Helenius , and E. Service . 2000 . Neurophysiology of fu-
ent and impaired reading: A magnetoencephalographic approach .
Journal of Clinical Neurophysiology 17 : 163 74.
Santi , A. , and Y. Grodzinsky . 2007 . Working memory and syntax interact
in Brocas area . Neuroimage 37 .1: 8 17.
Tarkiainen , A., P. Helenius, P. L. Hansen, P. L. Cornelissen , and
R. Salmelin . 1999 . Dynamics of letter string perception in the human
occipitotemporal cortex . Brain 122: 2119 32.
Wise , R. J. 2003 . Language systems in normal and aphasic human sub-
jects: Functional imaging studies and inferences from animal studies .
British Medical Bulletin 65 : 95 119.
NUMBER
Number is a grammatical feature that quantifes the denotation
of a linguistic element. It can refer to entities or events, and in
language we fnd both nominal number (very common, dis-
cussed in the following) and verbal number (less common, real-
ized on the verb to indicate the number of events or the number
of participants; also called pluractionality ).
Languages vary with regard to the part of their nominal inven-
tory that is involved in the number system. In diferent languages,
the split into nominals that do and do not express number may
occur at diferent points of the animacy hierarchy : speaker (frst
person pronouns) > addressee (second person pronouns) > third
person pronouns > kin > rational > human > animate > inanimate.
Furthermore, not all nouns are number diferentiable. Two types of
noun are traditionally distinguished: count nouns and mass nouns,
the latter regarded as lacking the number distinction . At the level of
semantics , the countmass distinction can be captured with two
semantic features boundedness and internal structure (Jackendof
1991), which corresponds to the distinction between temporally
bounded and unbounded events in verbal semantics. But count-
ability is really a characteristic of nominal phrases (Allan 1980),
since many nouns can appear in both count and mass syntactic
contexts, for example, Would you like a cake/some cake? We need a
bigger table/ Tere is not enough table for everyone to sit at .
When nominal number is found expressed on the noun or the
noun phrase as such, it is considered inherent. When found on other
elements of the noun phrase or on the verb, it is contextual. Te
signals, the high temporal resolution of EEG makes it possible
to analyze on-line processing of language, addressing questions
that difer fundamentally from those tapped by fMRI .
Magnetoencephalography bears many similarities to EEG
while ofering improved spatial resolution. MEG measures mag-
netic moments created by electrical activity in the brain that are
picked up by a large magneto-detector sensitive to very small
changes in magnetic felds. Tese felds are closely related to neu-
ral activity and can be tracked over temporal intervals efectively in
real time. Localizing the source of MEG signals remains a signif-
cant dif culty with this technique, and sources generated close to
the brain surface are easier to detect than those from deep struc-
tures. Nonetheless, source localization appears to be far more
precise than with EEG, while temporal resolution is equivalent.
MEG is more technically challenging than EEG and fMRI and is
also less widely available. Nonetheless, signifcant contributions
to the feld of language continue to emerge from MEG studies.
For example, MEG studies have indicated that letter-string rec-
ognition occurs 150 msec after presentation of a word, whereas
children with dyslexia show a weaker response (Tarkiainen et al.
1999). A corresponding delayed response in superior temporal
regions may refect an earlier-level dysfunction in letter-string
recognition areas (Salmelin, Helenius, and Service 2000).
Both EEG and MEG are limited to event-related designs that
measure the temporal dynamics of evoked responses to a brief
duration stimulus. In some cases, however, temporal resolution
isnt necessary. For instance, studying mood or drug states or
observing a cognitive strategy that evolves over time may require
longer time intervals to measure, favoring fMRI or PET .
Susan Bookheimer
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bates E. , S. M. Wilson, A. P. Saygin, F. Dick, M. I. Sereno, R. T. Knight ,
and N. F. Dronkers . 2003 . Voxel-based symptom mapping . Nature
Neuroscience 6 : 448 50.
Bookheimer , S. Y . 2002 . Functional MRI of language: New approaches
to understanding the cortical organization of semantic processing .
Annual Review of Neuroscience 25 : 151 88.
Bookheimer , S. Y., T. A. Zef ro, T. Blaxton, W. D. Gaillard, B. Malow , and
W. H. Teodor e. 1998. Regional cerebral blood fow during auditory
responsive naming: Evidence for cross- modality neural activation .
Neuroreport , 9 .10: 240913.
Bookheimer , S. Y., T. A. Zef ro, T. Blaxton, W. D. Gaillard , and W. H.
Teodore . 1995 . Regional cerebral blood fow during object naming
and word reading . Human Brain Mapping 3.2 : 93 106.
Buxton , R. B., K. Uludag, D. J. Dubowitz , and T. T. Liu . 2004 . Modeling
the hemodynamic response to brain activation . Neuroimage . 23
(Supplement 1): S220 3.
Cohen , M. S. , and S. Y. Bookheimer . 1994 . Functional magnetic reso-
nance imaging . Trends in Neurosciences 17 .7: 268 77.
Dronkers , N. F., O. Plaisant, M. T. Iba-Zizen , and E. A. Cabanis . 2007 .
Paul Brocas historic cases: High resolution MR imaging of the brains
of Leborgne and Lelong . Brain 130 .5: 1432 41.
Hahne , A., K. Eckstein , and A. Friederici . 2004 . Brain signatures of syn-
tactic and semantic processes during childrens language develop-
ment . J Cogn Neurosci 16.7 : 1302 18.
Kwong , K. K., J. W. Belliveau, D. A. Chesler, I. E. Goldberg, R. M. Weisskof,
B. P. Poncelet, D. N. Kennedy, B. E. Hoppel, M. S. Cohen, R. Turner, H.
Cheng, T. Brady , and B. Rosen . 1992 . Dynamic magnetic resonance
Neuroimaging Number
563
was only indirect in the sense that its role was limited to provid-
ing visual input processed later on by language centers in other
brain regions that contribute to certain aspects of language (such
as reading).
New data are now challenging this view of the occipital lobe
as solely processing visual information. First, several key studies
on the blind have shown the involvement of the occipital lobe
in processing other sensory modalities. In particular, there is
a clear link between occipital lobe processing in the blind and
language and verbal memory functions, where this pattern of
activation is attributed to massive reorganization of the occipi-
tal lobe in cases of blindness. Further, the involvement of the
occipital lobe in nonvisual processing has clearly been demon-
strated in the sighted (i.e., under normal development of the
occipital lobe), and much of todays research is exploring the
extent to which the occipital lobe is involved in language pro-
cessing under normal development. Tese topics are covered in
the following sections .
The Occipital Lobe and Vision
We live in a culture that relies heavily on vision and, accord-
ingly, vision research has motivated and dominated neurosci-
ence research. Te discovery and analysis of cortical visual areas
using electrophysiological and anatomical techniques was one
of the major milestones in visual neuroscience (e.g., Hubel and
Wiesel 1963 , 1965; Zeki 1978 ). On the basis of the vast amount
of anatomical studies in the primate, we now have a picture of
a highly diverse and hierarchically structured system (Felleman
and Van Essen 1991 ). Tis hierarchical organization originates
in the geniculostriate pathway from visual area 1 (V1, primary
visual cortex) and beyond to an array of visual areas. Along this
hierarchical organization there is an increase both in the recep-
tive feld size and in the complexity of the optimal stimulus neu-
rons in each area. Converging evidence suggests that the visual
cortex is structured according to several principles of organi-
zation and functional neuroanatomical schemes. Following
are descriptions of some of the most important organizing
principles:
1. Topographical Organization. Topographic mapping can
chart an orderly and gradual change in some functional prop-
erty of cortical neurons laid along the cortical surface. Te most
fundamental transformation in vision is retinotopy. Tis involves
the transformation from a Euclidean coordinate system in the
retina to polar coordinates in the visual cortex. In this transfor-
mation, each of the early visual areas maps the visual feld along
two orthogonal axes: the polar angle (points that lie on a spe-
cifc radius whose origin is at the fovea have an identical polar
angle) and eccentricity (the eccentricity distance from the fovea,
the center of the visual feld). Areas in the left central vision area
are projected to the back of the right occipital lobe, whereas the
more peripheral areas are projected more anteriorly. Te retinal
points on such radii are mapped onto parallel bands across cor-
tical areas. Te sequential layout of these bands reverses when
crossing from one visual area to another, providing a way for an
accurate delineation of the borders of these retinotopic areas
(Sereno et al. 1995 ). Tere is evidence that higher-order object-
related areas are also topographically organized, but the basis for
expressions of nominal number can involve special number words
(of diferent syntactic status); syntactic means (i.e., agreement ,
found most commonly on demonstratives and verbs but also on
articles, adjectives, pronouns, nouns in possessive constructions,
adverbs, adpositions, and complementizers); a variety of mor-
phological means (infections, stem changes, zero expressions,
clitics ); and lexical means (e.g., suppletion). Number is often
marked in more than one way within one language.
All nominal number systems are built on the primary opposi-
tion between singular (expressing the quantity one ) and plural
( more than one ). Other attested number values are dual ( two ),
trial ( three ), and paucal ( a few ). Tere may be further divisions
into paucal and greater paucal, plural and greater plural (the last
value may imply an excessive number, or all possible instances of
the referent). No genuine quadrals ( four ) have been found. Te
largest number systems involve fve values. In many languages,
the absence of plural marking does not necessarily imply the sin-
gular, but the form may be outside the number opposition and
express general number , that is, the meaning of the noun without
reference to number .
Associatives, distributives, and collectives all sometimes
listed as additional values of number are better analyzed as
independent features. Associativity expresses the meaning X
and the group associated with X; distributives indicate that
entities (whether count or mass ones), events, qualities, or loca-
tions are to be construed as distinct in space, sort, or time; and
collectives indicate that the members of a group are to be con-
strued together as a unit. Many languages have markers for these
categories in addition to various number markers .
Anna Kibort
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Allan , Keith . 1980 . Nouns and countability . Language 56 : 541 67.
Corbett , Greville G . 2000 . Number . Cambridge : Cambridge University
Press .
Jackendof , Ray . 1991 . Parts and boundaries . Cognition 41 : 9 45.
O
OCCIPITAL LOBE
Alcmaeon of Croton , probably the frst person to suggest that the
mind is located in the brain and not the heart, also suggested that
the optic nerves are light-bearing paths to the brain. His revo-
lutionary ideas, formulated about 2,500 years ago, were ignored
by Egyptian and Greek scholars alike (most notably by Aristotle).
We now know that visual information is delivered to the occipi-
tal lobe mainly via the thalamus. Te occipital lobe is the most
posterior of the four lobes of the brain (named after the four skull
bones beneath which they lie).
Even until very recently, it was believed that the occipital lobe
was involved only in the processing of visual information per
se. Language was considered to take place in dedicated areas,
mainly in the frontal and temporal cortex. Tus the relevance of
the occipital lobe to language processing and language research
Occipital Lobe
564
information processing in the primary and secondary sensory
areas is strictly modality-specifc. According to this view, the
occipital cortex processes vision, and integration of the difer-
ent senses occurs only in higher-level areas. Recent evidence
suggests that the occipital cortex does in fact process nonvi-
sual functions. We focus later on object recognition and object
naming, although similar results were obtained in the dorsal
stream (e.g., for visuo-tactile orientation, see Zangaladze et al.
1999 ).
Recognition of an object can involve a wide range of cues,
for example, a characteristic color, a unique texture, or a typical
sound. However, shape is a particularly fundamental feature for
recognizing and naming objects. Surprisingly, recent neuroim-
aging studies have found that visual and tactile object-related
information (both contribute to shape information) converges
in a lateral occipito-temporal ventral visual stream area (LOtv;
Amedi et al. 2001 ). A later study found that shape and not sen-
sory modality is indeed the crucial factor in activating these
regions. Te study used visual-to-auditory sensory substitution
devices (Bach-y-Rita and Kercel 2003 ) in which visual images
are captured by a camera and then transformed by a predeter-
mined algorithm into soundscapes that preserve shape infor-
mation. Te study found that recognizing objects by their typical
sounds or learning to associate specifc soundscapes with spe-
cifc objects do not activate this region. Critically, soundscapes
synthesized to preserve shape information did activate LOtv
robustly. Tis suggests that LOtv is driven by the presence of
shape information, rather than by the sensory modality that
provides this information (Amedi et al. 2007 ). It is interesting to
note that a similar phenomenon of a modal representation in
occipito-temporal cortex was found for word recognition. Te
study showed that the left basal occipito-temporal area shows
specifcity to word processing, regardless of the sensory modal-
ity used. Tis area showed selective activation to words versus
non-word letter strings when subjects read using vision or when
blind individuals read Braille using touch (Buchel, Price, and
Friston 1998 ) .
Another example of cross-modal interactions is the case of
integration of heard and seen speech (letter-sound associa-
tion), which is another crucial function for normal development
of language abilities. Tis function, however, was shown to
be mediated primarily by the temporal lobe (especially in the
superior temporal sulcus and gyrus, STS and STG, respectively).
For instance, a recent fMRI study (van Atteveldt et al. 2004 )
showed that bilateral STS/STG responded more strongly to
bimodal matching of letter-sound pairs than to their respective
unimodal components. Note that correspondences between
speech sounds and mouth movements are learned implicitly
and early in development by exposure to heard speech together
with the sight of the speaker. In contrast, the visual representa-
tion of spoken language by written language is a cultural arti-
fact. Terefore, associations between letters and speech sounds
are not learned automatically but require explicit instruction.
It is interesting that the learning of new letter-sound mappings
involves the occipital cortex, whereas the auditory association
cortex is active during the processing of previously acquired
matching letter-sound combinations (Hashimoto and Sakai
2004 ) .
this organization is still subject to debate (for a review, see Grill-
Spector and Malach 2004 and the following) .
2. Visual Pathways. Lesion studies in primates and human
fMRI studies both suggest that there are two processing streams
between early, retinotopic visual areas in the occipital cortex
and higher-order processing centers in the occipito-temporal
and occipito-parietal lobes. Tese two streams are referred to
as the dorsal and ventral streams, respectively (Ungerleider
and Haxby 1994 ). Because the dorsal stream is also involved
in visuo-motor transformations, a diferentiation is made
between vision for action (dorsal stream) and vision for per-
ception (ventral stream) (Goodale and Milner 1992 ). Te ven-
tral stream contains structures devolved to the fne analysis of
a visual scene such as form and color. Tus, it is also known
as the what pathway. It consists of areas V1V4 in the occipital
lobe and several regions that belong to the lateral and ventral
temporal lobe .
3. Functional Specialization. Tis principle of division of
labor, which leads to a specialization of function in the vari-
ous cortical areas, was originally suggested for the visual sys-
tem (Zeki 1978 ). Electrophysiological studies in nonhuman
primates have identifed organizing principles in addition
to retinotopy, such as selectivity for simple features like spa-
tial orientation in V1 and selectivity for categories of complex
stimuli like faces for spatial layouts in inferior temporal (IT)
cortex. In the last decade or so, the use of noninvasive func-
tional imaging, particularly fMRI, has dramatically increased
our knowledge of the functional organization of the human
visual cortex and its relation to vision, due to its ability to pro-
vide a large-scale neuro-anatomical perspective (e.g. Martin
and Chao 2001 ). An active debate is ongoing about the actual
organization of the ventral stream (Grill-Spector and Malach
2004 ). One area in the lateral occipital cortex, the lateral occipi-
tal complex (LOC; Malach et al. 1995 ) responds strongly to
pictures of intact objects by contrast to scrambled objects or
nonobject textures. In the ventral occipito-temporal cortex,
specialized areas for faces (fusiform face area, FFA; Kanwisher,
McDermott, and Chun 1997 ), scenes (Epstein and Kanwisher
1998 ), and human body parts (Downing et al. 2001 ) have been
described, as well as for visual word forms (visual word form
area, VWFA; McCandliss, Cohen, and Dehaene 2003 ) which
has direct implications for the discussion here. Developing a
theoretical framework that captures these specialized regions
continues to be problematic, although the notion of widely
distributed and overlapping cortical object representations
remains a likely principle of organization (Haxby et al. 2001 ).
Te efects of perceptual expertise for certain object categories
(Gauthier et al. 2000 ) and, more recently, diferent category-
related resolution needs (Grill-Spector and Malach 2004 ) have
also been put forward as candidate organizational principles of
the human ventral stream .
The Occipital Lobe, Multisensory Integration,
and Nonvisual Processing
Te perception of objects and the perception of space are cogni-
tive functions of prime importance. In everyday life, these func-
tions beneft from the coordinated interplay of vision, audition,
and touch. A central theme in sensory neurophysiology is that
Occipital Lobe
565
tasks and during speech processing (See Burton , Diamond, and
McDermott 2003 ; Amedi et al. 2003 ; Pascual-Leone et al. 2005 ;
Roder et al. 2002 ).
For instance, robust plasticity in the left occipital areas of
the blind is evident during verbalmemory tasks requiring the
retrieval of abstract words from long-term memory in early
blind individuals (Amedi et al. 2003 ). In this case, the observed
occipital activation occurred without introducing any tactile or
auditory sensory input. Notably, blind subjects showed superior
verbal memory capabilities, compared not only to age-matched,
sighted controls but also with reported population averages.
Furthermore, in the blind group only, a strong positive correla-
tion was found between the magnitude of V1 activation and the
verbal memory capabilities of individual subjects.
More directly related to language processing, several stud-
ies have used a verb-generation task, in which both blind and
sighted subjects were instructed to generate a verb in response
to a noun cue. Te sighted group showed activation in typical
language- related areas (e.g., brocas area in the prefrontal
cortex, which was activated as well in the blind), but no occipi-
tal activation. Te blind group, however, showed additional
robust activation in the occipital cortex (Burton, Diamond, and
McDermott 2003 ; Amedi et al. 2003 ). Furthermore, as research
in bilingual subjects has demonstrated, the convergence of
two languages in prefrontal cortex during semantic tasks (e.g.
Crinion et al. 2006 ), performed by bilingual blind subjects,
shows additional convergence of the two languages in the poste-
rior occipital cortex, including in the primary visual cortex (Ofan
and Zohary 2006 ). In addition, fMRI studies have shown that
efective connectivity between the prefrontal and occipital cor-
tex is increased in blind individuals during semantic processing.
Both early blind and sighted subjects activate a left-lateralized
fronto-temporal core semantic retrieval system. However, blind
subjects activate additional extra-striate regions, which are cou-
pled with frontal and temporal semantic regions (Noppeney,
Friston, and Price 2003 ; Liu et al. 2007 ).
Finally, it should be pointed out that neuroimaging, at best,
establishes an association between brain activity and task per-
formance. A causal link between occipital areas and semantic
processing was reported in a recent transcranial magnetic stimu-
lation study. TMS targeted over the left V1 or left occipito-tempo-
ral cortex led to a disruption and an increase in the error rate in a
similar verb-generation task in blind but not in sighted subjects
(Amedi et al. 2004 ). An analysis of error types revealed that the
most common error produced by the TMS was semantic (e.g.,
apple would lead to the verb jump ), whereas phonological errors
and interference with motor execution or articulation were rare .
Tese results suggest that processing language and verbal
memory in the blind incorporates a widespread network that
encompasses occipital visual brain areas, and that this type of
reorganization of language and memory is relevant to behavior.
It is clear, for example, that the functional and structural iden-
tity of the occipital cortex may switch from processing visual
information to processing information related to another sen-
sory modality or even diferent language functions. However,
is this a unique consequence of early blindness? As shown here
in some examples, the occipital cortex may inherently possess
the computational machinery needed for nonvisual information
Language and Verbal Processing in the Occipital Lobe
READING AND DYSLEXIA; NAMING AND ALEXIA. Reading words
and naming visual objects involves the association of visual stim-
uli with phonological and semantic knowledge. Damage to
the left occipital lobe can result in pure alexia: the inability to read
without losing the ability to write or the loss of any other major
language-related function (Damasio and Damasio 1983 ). More
recent neuroimaging studies support this view by showing cor-
relations between the left occipito-temporal cortex with linguis-
tic aspects of reading and object naming based on visual input
in normal subjects. No less informative is the study of abnormal
patterns of activation in this part of the brain in subjects with
developmental dyslexia (Schlaggar and McCandliss 2007 ) .
VISUAL WORD FORM IN THE SIGHTED AND TACTILE BRAILLE IN THE
BLIND. An interesting example of such language-related pro-
cessing in the occipital lobe is the case of visual word form in the
sighted. Related to this is the case of occipital activation during
Braille reading in the blind. In both cases, activation was found
for both words and letter strings. Te specifcs and signifcance of
these two examples are discussed in this section.
One of the most hotly debated topics in the context of modu-
lar versus general architecture organization in the occipital lobe
is the existence of the human visual word form area, which is
dedicated to the construction of visual words and thus is a key
player in our ability to read. Like other language-related areas
in the prefrontal and parietal cortex, this area also has a strong
hemispheric dominance located in the left occipito-temporal
sulcus bordering the fusiform gyrus (McCandliss, Cohen, and
Dehaene 2003 ). Other methodologies, such as recording feld
potentials in awake humans, showed selectivity to words and
letters strings in similar parts of the left occipital cortex (Nobre,
Allison, and McCarthy 1994 ). Recently, a causal link between
lesions in VWFA and acquired alexia without agraphia was dem-
onstrated in a patient who had a patch of his cortex removed in
surgery, causing activation in VWFA to disappear (Gaillard et al.
2006; Martin 2006).
As in the case of the fusiform face area, some investigators
have suggested that there is no reason to label VWFA as a sepa-
rate modular brain area purely because of the possible lack of
specifcity of a ventral occipito-temporal lesion to cause pure
alexia, and since the reading disorder might not be limited to
words or could be a manifestation of a more general visual pro-
cessing defcit (Price and Devlin 2003 ).
What happens when early blind subjects read using a difer-
ent sensory modality? Recent neuroimaging studies in the blind
have demonstrated robust occipital cortex activation during
Braille reading (see blindness and language ). In this case,
activation is not limited to occipito-temporal areas but stretches
all the way to the primary visual cortex (Sadato et al. 1996 ), and
interference in processing in the occipital cortex using tran-
scranial magnetic stimulation (TMS) increases the error rate in
Braille reading (Cohen et al. 1997 ) .
OCCIPITAL LOBE AND PLASTICITY IN LANGUAGE AND VERBAL
MEMORY FUNCTIONS. Recent neuroimaging studies in the blind
have demonstrated robust occipital cortex activation during
a wide variety of linguistic and specifcally semantic judgment
Occipital Lobe
566
Hubel , D. H. , and T. N. Wiesel. 1963 . R Shape and arrangement of col-
umns in cats striate cortex. Journal of Neurophysiology 165 : 55968.
. 1965 . Receptive felds and functional architecture in two non-
striate visual areas (18 and 19) of the cat . Journal of Neurophysiology
28 : 22989.
Kanwisher , N., J. McDermott , and M. M. Chun. 1997 . Te fusiform face
area: A module in human extrastriate cortex specialized for face per-
ception . Journal Neuroscience 17 : 430211.
Liu , Y., et al. 2007 . Whole brain functional connectivity in the early
blind . Brain 130 : 208596.
Malach , R., et al. 1995 . Object-related activity revealed by functional
magnetic resonance imaging in human occipital cortex . Proceedings
of the National Academy of Science 92: 81359.
Martin , A. 2006 . Shades of Djerine forging a causal link between the
visual word form area and reading. Neuron 50 : 1735.
Martin , A., and L. L. Chao. 2001 . Semantic memory and the
brain: Structure and processes. Current Opinion in Neurobiology
11 : 194201.
McCandliss , B. D., L. Cohen, and S. Dehaene. 2003 . Te visual word form
area: Expertise for reading in the fusiform gyrus. Trends in Cognitive
Science 7 : 2939.
Nobre , A. C., T. Allison, and G. McCarthy. 1994 . Word recognition in the
human inferior temporal lobe . Nature 372 : 2603.
Noppeney , U., K. J. Friston , and C. J. Price. 2003 . Efects of visual depriva-
tion on the organization of the semantic system. Brain 126 : 16207.
Ofan , R. H., and E. Zohary. 2006 . Visual cortex activation in bilingual
blind individuals during use of native and second language. Cerebral
Cortex 17 : 124959.
Pascual-Leone , A., A. Amedi, F. Fregni , and L. B. Merabet. 2005 . Te plas-
tic human brain cortex. Annual Review of Neuroscience 28 : 377401 .
Price , C. J., and J. T. Devlin. 2003 . Te myth of the visual word form area .
Neuroimage 19 : 47381.
Rder , B., O. Stock, S. Bien, H. Neville , and F. Rosler. 2002 . Speech pro-
cessing activates visual cortex in congenitally blind humans. European
Journal of Neuroscience 16 : 9306.
Sadato N. , A. Pascual-Leone, J. Grafman, V. Ibanez, M. P. Deiber, et al.
1996 . Activation of the primary visual cortex by Braille reading in
blind subjects. Nature 380 : 5268.
Schlaggar , B. L. , and B. D. McCandliss. 2007 . Development of neural sys-
tems for reading. Annual Reviews in Neuroscience 30 : 475503.
Serano , S.C., et al. 1995 . Borders of multiple visual areas in humans
revealed by functional magnetic resonance imaging. Science 268:
88993 .
Ungerleider , L. G. , and J. V. Haxby. 1994 . What and where in the
human brain. Current Opinion in Neurobiology 4 : 15765.
van Atteveldt , N., E. Formisano, R. Goebel , and L. Blomert. 2004 .
Integration of letters and speech sounds in the human brain. Neuron
43 : 27182 .
Zangaladze , A., C. M. Epstein, S. T. Grafton, and K. Sathian. 1999 .
Involvement of visual cortex in tactile discrimination of orientation.
Nature 401 : 58790.
Zeki , S. M. 1978 . Functional specialization in the visual cortex of the rhe-
sus monkey. Nature 274 : 4238.
OPTIMALITY THEORY
Optimality Teory (OT; Prince and Smolensky [ 1993 ] 2004) is a
formal theory of constraint interaction in grammar that seeks
to explain how and to what extent natural languages may vary.
In addition to this central question of generative grammar ,
research in OT addresses questions of the grammars use in per-
formance, its acquisition, and its neural realization.
processing. Under specifc conditions, this potential may be
materialized. If so, visual deprivation may simply allow for the
emergence of the true potential of certain brain regions. Tis
hypothesis also suggests that careful task choice and experimen-
tal design (e.g., blindfolding sighted subjects for several days)
may reveal additional nonvisual, linguistic roles in the occipital
cortex in the sighted (Pascual-Leone et al. 2005 ) .
Amir Amedi
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Amedi , A., et al. 2007 . Shape conveyed by visual-to-auditory sensory sub-
stitution activates the lateral occipital complex . Nature Neuroscience
10 : 6879.
Amedi , A. , R. Malach , T. Hendler , S. Peled , and E. Zohary . 2001 . Visuo-
haptic object-related activation in the ventral visual pathway. Nature
Neuroscience 4 : 32430.
Amedi , A , N. Raz , P. Pianka , R. Malach , and E. Zohary. 2003 . Early
Visual Cortex Activation Correlates with Superior Verbal Memory
Performance in the Blind . Nature Neuroscience 6 : 75866.
Amedi , A, A. Floel, S. Knecht, E. Zohary , and L. G. Cohen. 2004 .
Transcranial Magnetic Stimulation of the Occipital Pole Interferes
with Verbal Processing in Blind Subjects. Nature Neuroscience
7 : 126670 .
Bach-y-Rita , P. , and S. W. Kercel. 2003 . Sensory substitution and the
human-machine interface. Trends in Cognitive Neuroscience 7 : 5416.
Buchel , C., C. Price , and K. Friston . 1998 . A multimodal language region
in the ventral visual pathway . Nature 394 : 2747.
Burton , H., J. B. Diamond , and K. B. McDermott. 2003 . Dissociating cor-
tical regions activated by semantic and phonological tasks to heard
words: A fMRI study in blind and sighted individuals. Journal of
Neurophysiology 90 : 196582.
Crinion , J., et al. 2006 . Language control in the bilingual brain. Science
312 : 153740.
Cohen , L. G., P. Celnik, A. Pascual-Leone, B. Corwell, L. Falz, et al. 1997 .
Functional relevance of cross-modal plasticity in blind humans .
Nature 389 : 1803.
Damasio , A. R., and H. Damasio. 1983 . Te anatomic basis of pure
alexia . Neurology 33 : 157383.
Downing , P. E., Y. Jiang, M. Shuman , and N. Kanwisher. 2001 . A cor-
tical area selective for visual processing of the human body. Science
293 : 24703.
Epstein , R., and N. Kanwisher. 1998 . A cortical representation of the
local visual environment. Nature 392 : 598601.
Felleman , D. J. , and D. C. Van Essen. 1991 . Distributed hierarchical pro-
cessing in the primate cerebral cortex. Cerebral Cortex 1 : 147.
Gaillard , R., L. Naccache, P. Pinel, S. Clemenceau, E. Volle, et al. 2006 .
Direct intracranial, fMRI, and lesion evidence for the causal role of left
inferotemporal cortex in reading. Neuron 50 : 191204.
Gauthier , I., P. Skudlarski, J. C. Gore , and A. W. Anderson. 2000 . Expertise
for cars and birds recruits brain areas involved in face recognition.
Nature Neuroscience 3 : 1917.
Goodale , M. A. , and A. D. Milner. 1992 . Separate visual pathways for per-
ception and action. Trends in Neurosciences 15 : 205.
Grill-Spector , K. , and R. Malach. 2004 . Te human visual cortex. Annual
Review of Neuroscience 27 : 64977.
Hashimoto , R ., and K. L. Sakai. 2004 . Learning letters in adult-
hood: Direct visualization of cortical plasticity for forming a new link
between orthography and phonology. Neuron 42 : 31122.
Haxby , J. V., et al. 2001 . Distributed and overlapping representa-
tions of faces and objects in ventral temporal cortex . Science
293 : 242530.
Occipital Lobe Optimality Theory
567
competitors only the most harmonic of which are grammatical.
Te competition (via H-Eval ) evaluates each pair of candidates
against the universal constraint set Con , which is ordered into a
language-particular domination hierarchy or ranking ; C
1
>> C
2
means that constraint C
1
dominates constraint C
2
. In OT, con-
straint domination is strict : One violation of any constraint C is
always worse than violating constraints ranked lower than C
regardless of how many lower-ranked constraints are violated
and regardless of how severe the violations are of those lower-
ranked constraints . Given two candidate structural descriptions
p and q for an input I , p has higher harmony ( p q ) if p is pre-
ferred by the highest-ranked constraint that does not evaluate p
and q as equal. A candidate p is optimal if there is no other candi-
date q with higher harmony. If p is optimal, then p cannot violate
any constraint C unless every competing candidate that is pre-
ferred to p by C is dispreferred to p by a constraint higher ranked
than C. In this sense, violations incurred by an optimal candidate
structure are minimal . In sum, for every input I , harmony opti-
mization over the candidates in Gen ( I ) determines (at least) one
optimal, though not necessarily perfect, structural description of
I , which is ipso facto declared to be a grammatical output for that
input.
Te only cross-linguistically varying property of the grammar
is the relative ranking of the universal constraints in Con : the
set of all possible grammars then is exactly the set of all rank-
ings of this fxed set of constraints. Typically, any given empirical
pattern that is predicted to be part of the universal typology is
generated by many diferent (but typologically equivalent) rank-
ings, and the number of predicted possible typological patterns
is vastly smaller than the number of all possible rankings (e.g.,
13 patterns vs. 40,320 rankings in Smolensky and Legendre 2006 ,
Chapter 15).
Employing violable, conficting constraints often means that
the universal constraints can be more simply stated: Complexity
emerges primarily from the interaction of simple constraints;
this reduces the need for hedges or disjunctive principles arising
when universal constraints are construed as inviolable (Speas
1997 ). Constraint ranking naturally captures the common situ-
ation in which some phenomenon widely observed in other
languages is seen in one context only in language L (e.g., null
subjects in main clauses only as in Old French). Stipulating the
limited distribution is unnecessary: In L , a lower-ranked con-
straint C (e.g., against null elements) often violated in gram-
matical forms of L makes itself felt in those special contexts
where dominating constraints do not contravene. In other lan-
guages, where C is more highly ranked, the phenomenon is seen
widely .
FAITHFULNESS. Te most familiar grammatical constraints are
markedness constraints , which demand that the output struc-
ture meet some well-formedness condition (e.g., a subject must
be an agent; Smolensky and Legendre 2006 , Chapter 15). In an
optimizing grammar, unless there is pressure for the output to
contain all and only the elements contained in the input, the
optimizing system would always simply return the best of all
structures (under the given ranking). Grammars must therefore
include ( inputoutput ) faithfulness constraints, which require
minimal structural distance between the output and the input.
OT in Theoretical Linguistics
Characterizing the set of possible natural languages minimally
requires specifying i) the mental representations that are char-
acteristic of language, ii) the constraints that distinguish possible
from impossible linguistic systems, and iii) the formal mode of
interaction among these constraints. Concerning mental rep-
resentations, OT imposes no restrictions beyond requiring that
specifcations of phonological, syntactic, or semantic structure
be explicit in the sense of generative grammar. Tis makes OT
compatible with alternative substantive theories of particu-
lar grammar components; in syntax, for example, OT versions
of government and binding (Grimshaw 1997 ; Legendre,
Smolensky, and Wilson 1998 ), lexical-functional gram-
mar (Bresnan 2000 ), and the minimalist program (Mller
1997 ) are fourishing. OTs main contribution concerns con-
straint interaction (iii) and, as a consequence, the proper formal
characterization of the constraints themselves (ii). Hence, OT
is best characterized as a meta-theory of grammatical structure
compatible with any explicit theory of linguistic representation.
It is therefore applicable to all linguistic levels, and has been
applied to phonology (McCarthy and Prince 1993 ; Prince and
Smolensky [ 1993 ] 2004), syntax (Legendre, Grimshaw and
Vikner 2001 ), semantics (Hendriks and de Hoop 2001 ), and
pragmatics (Blutner and Zeevat 2004 ; Blutner, De Hoop, and
Hendriks 2006 ).
According to OT, all grammatical constraints are universal
and violable (or soft) a claim that represents a major depar-
ture from previous approaches to the characterization of phono-
logical and syntactic knowledge via language-particular rewrite
rules written in a universal notation, or as universal, inviolable
constraints supplemented by additional principles subject to
language-particular parameterization.
ARCHITECTURE. An OT grammar maps an input specifcation
onto an output structure. In phonology, the input is typically
an underlying form and the output the corresponding surface
form. In syntax the input is a proposition and the output is the
grammatical form that expresses that meaning (excepting inef-
fability: see the section on Faithfulness). Gen ( generator ) is a
mechanism for producing candidate outputs for any input and
freely generates all of the types of structures that are present
in any of the worlds languages (McCarthy and Prince 1993 ).
Pruning this enormous set down to the grammatical forms is the
job of H-Eval ( harmony evaluator ), a procedure for evaluating
the relative well-formedness (or harmony ) of candidate struc-
tural descriptions. H-Eval depends on a set of universal well-
formedness constraints Con .
COMPETITION AND CONFLICT. Te fact that well-formedness con-
straints are violable in OT means that two constraints often con-
fict : Satisfying one requires violating the other. Grammaticality
is therefore not equated with satisfaction of all grammatical
constraints. Grammatical structures are simply structures that
sufer less severe constraint violations than their ungrammatical
counterparts. Tis means that the evaluation of grammaticality is
inherently comparative . At any level of description (phonological,
syntactic, etc.) the universal set of possible structural descrip-
tions of an input I , Gen ( I ), forms a candidate set , a collection of
Optimality Theory
568
probability of a candidate is proportional to the exponential of its
harmony. Gaja Jarosz 2006 defnes a version of OT phonology,
maximum likelihood learning of lexicons and grammars (MLG),
in which underlying forms as well as rankings have probability
distributions, defning a lexicon + grammar. Te relative prob-
ability of a form has been used to model its gradient acceptability
(Boersma and Hayes 2001 ; Hayes and Wilson 2008) .
OT and Grammar Use
OT is well suited for theories of performance. With no additional
machinery, standard OT grammars assign a structural descrip-
tion to all inputs, including loanword inputs violating the pho-
notactics of the borrowing language (Yip 1993 ; Davidson, Jusczyk,
and Smolensky 2006 ) or the initial fragment of a sentence being
processed word by word (Gibson and Broihier 1998 ; Stevenson
and Smolensky 2006 ). In the latter case, processing dif culty is
predicted to occur if the optimal parse of the initial portion of
a sentence changes substantially when a new word arrives. On
a formal level, the computational complexity of the problem
of computing optimal outputs is well studied (e.g., Tesar 1996 ;
J. Eisner 1997 ; Frank and Satta 1998 ; Idsardi 2006 )
OT and Grammar Acquisition
OTs account of variation and change provides a natural extension
to the analysis of acquisition of phonology and syntax (Legendre
et al. 2002 ; Kager, Pater, and Zonneveld 2004 ; Legendre et al.
2004 ). In early child syntax, for example, a childs competence
may be characterized by a partial ranking P between faithful-
ness constraints and constraints penalizing syntactic structure
(Legendre et al. 2002 ). In some of the full rankings consistent
with P , higher-ranked faithfulness constraints lead to optimal
clauses with functional projections; in other rankings consistent
with P , lower ranking of these faithfulness constraints entail that
optimal outputs lack some or all functional projections. Te vari-
ation in child production of tense and agreement marking is thus
given a principled grammatical account.
Formal and computational studies of the problem of learning
OT grammars have been extensive, including constraint learn-
ing (Hayes and Wilson 2008), ranking learning (Tesar 1998 ; Tesar
and Smolensky 1998 ; Jason Eisner 2000 ; Boersma and Hayes
2001 ; Hayes 2004 ; Prince and Tesar 2004 ), simultaneous learning
of a probabilistic phonological lexicon and a ranking by maxi-
mum likelihood estimation (Jarosz 2006 ), and the mathematical
logic of OT learning (Prince 2002 , 2006 ) .
OTs Neural Realization
OT has historical roots in debates (Pinker and Prince
1988 ; Smolensky 1988 ) concerning neural network (or
connectionist ) cognitive models (Rumelhart, McClelland,
and the PDP Research Group 1986 ). In these models, networks
of abstract model neurons excite and inhibit one another as acti-
vation spreads from input neurons to output neurons. Formal
analysis of a class of networks reveals that they perform optimi-
zation: Tey compute mental representations (activation pat-
terns) that maximize a numerical measure of self-consistency
or well-formedness: harmony (Smolensky 1986 ). Mathematical
analysis makes precise the following general picture: At a
lower level of description, spreading activation among abstract
Faithfulness constraints, unique to OT, have been shown to
operate at all levels of linguistic description. In syntax, faithful-
ness constraints play a crucial role in accounting for language-
particular inefability, that is, syntactic structures that are simply
impossible in certain languages, for example, multiple wh-ques-
tions in Irish (Legendre, Smolensky, and Wilson 1998 ; Legendre
2009). In most languages, faithfulness to question operators in a
semantic input force the optimal syntactic output to contain mul-
tiple wh-phrases. In languages like Irish, however, such faithful-
ness is ranked below the syntactic constraints violated by clauses
containing multiple wh-phrases, and thus no optimal syntactic
structure contains multiple wh-phrases. Multiple wh-questions
are therefore inexpressible (in a single clause) .
VARIATION. OT is naturally extensible to unstable states of lan-
guage, such as free variation (Anttila 1998 ), dialectal variation,
and diachronic change (Nagy and Reynolds 1997 ). Relaxing the
requirement of a complete ranking of the universal constraints, a
language L may be characterized by a single partial ranking P ; L
is generated by the set of grammars S full rankings consistent
with P . Tese generate a set of outputs for each input. According
to Joan Bresnan, Ashwini Deo, and Devyani Sharma (2007),
intraspeaker variation in the British paradigm for be arises from
a partial ranking P among faithfulness constraints, requiring an
output to express input agreement features and markedness
constraints penalizing all features in the output. A total ranking
consistent with P having more highly ranked faithfulness yields a
more highly infected paradigm .
FURTHER DEVELOPMENTS. OT is an evolving theory; major devel-
opments include the following. Faithfulness: Output-output
faithfulness requires identity of a morphemes exponent across
its paradigm (Burzio 1994 ; Benua 1995 ); output-output anti-faith-
fulness constraints achieve morphophonological alternations by
demanding nonidentical surface forms for distinct underlying
forms (Alderete 2001 ); sympathy theory demands faithfulness
to suboptimal candidates (McCarthy 1999b ) . Harmonic evalu-
ation: Comparative markedness evaluation distinguishes con-
straint violations that are shared with the most faithful output
from those that are not (McCarthy 2003 ); targeted constraints
only compare candidates that difer only in a specifed way
(Wilson 2001b ) . Architecture: Stratal OT assumes difering rank-
ings in a series of lexical levels (Kiparsky 2006); harmonic serial-
ism derives the surface form from a series of small alterations,
each optimal at its point in the derivation (Prince and Smolensky
[ 1993 ] 2004; McCarthy 1999a ); candidate chain theory evalu-
ates entire derivations (McCarthy 2007 ); bidirectional optimi-
zation adds competition of interpretations/underlying forms to
the competition of expressions/surface forms that is standard in
OT (Smolensky 1996 ; Blutner 2000 ; Wilson 2001a ) . Probabilistic
formulations: In stochastic OT (Boersma 1998 ), each optimiza-
tion ranks the constraints according to relative numerical val-
ues randomly selected for that optimization from a probability
distribution for each constraint, with a mean value determined
by that constraints strength in the grammar. In the maximum
entropy formulation of OT (Hayes and Wilson 2008), harmony is
numerical: Each constraint has a numerical strength determin-
ing the size of the penalties it assesses to violating candidates; the
Optimality Theory
569
Kager , Joe Pater, and Wim Zonneveld. Cambridge : Cambridge
University Press.
Hayes , Bruce , and Colin Wilson . 2008 . A maximum entropy model of pho-
notactics and phonotactic learning . Linguistic Inquiry 39: 379440.
Hendriks , Petra , and Helen de Hoop . 2001 . Optimality theoretic seman-
tics . Linguistics and Philosophy 24 : 132.
Idsardi , William J. 2006 . A simple proof that optimality theory is compu-
tationally intractable. Linguistic Inquir y 37: 2715.
Jarosz , Gaja . 2006 . Rich lexicons and restrictive grammars maximum
likelihood learning in optimality theory. Ph.D. thesis, Johns Hopkins
University.
Kager , Ren. 1999 . Optimality Teory . Cambridge : Cambridge University
Press .
Kager , Ren, Joe Pater, and Wim Zonneveld , eds. 2004. Constraints in
Phonological Acquisition. Cambridge : Cambridge University Press.
Kiparsky , Paul. 2006 . Paradigms and Opacity. Stanford, CA : CSLI
Publications.
Legendre , Graldine. 2009 . Inefability in syntax. In Modeling
Ungrammaticality in Optimality Teory, ed. Curt Rice , 23766 .
London : Equinox.
Legendre , Graldine, Jane Grimshaw, and Sten Vikner eds. 2001.
Optimality-Teoretic Syntax. Cambridge, MA : MIT Press.
Legendre , Graldine, Paul Hagstrom, Joan Chen-Main, Liang Tao , and
Paul Smolensky . 2004 . Deriving output probabilities in child manda-
rin from a dual-optimization grammar. Lingua 114 : 114785.
Legendre , Graldine, Paul Hagstrom, Anne Vainikka , and Marina
Todorova. 2002 . Partial constraint ordering in child French syntax .
Language Acquisition 10 : 189227.
Legendre , Graldine, Paul Smolensky , and Colin Wilson . 1998 . When
is less more? Faithfulness and minimal links in wh- chains. In Is the
Best Good Enough? Optimality and Competition in Syntax, ed. Pilar
Barbosa , Danny Fox, Paul Hagstrom, Martha McGinnis, and David
Pesetsky, 24989. MIT Working Papers in Linguistics. Cambridge,
MA : MIT Press.
McCarthy , John J. 1999 a. Harmonic serialism and parallelism. NELS 30:
50124.
. 1999 b. Sympathy and phonological opacity . Phonology
16 : 33199.
. 2002 . A Tematic Guide to Optimality Teory . Cambridge :
Cambridge University Press.
. 2003 . Comparative markedness . Teoretical Linguistics 29 :
151.
. 2007 . Hidden Generalizations: Phonological Opacity in Optimality
Teory. London : Equinox.
McCarthy , John J. , and Alan Prince. 1993 . Prosodic morphology
I: Constraint interaction and satisfaction. Technical Report RuCCS-
TR-3. Rutgers Center for Cognitive Science, Rutgers University, and
University of Massachusetts at Amherst .
Mller , Gereon. 1997 . Partial wh-movement and optimality theory .
Linguistic Review 14 : 249306 .
Nagy , Naomi , and William Reynolds . 1997 . Optimality theory and vari-
able word-fnal deletion in Ftar . Language Variation and Change
9 : 3755.
Pinker , Steven , and Alan Prince. 1988 . On language and connection-
ism: Analysis of a parallel distributed processing model of language
acquisition. Cognition 28 : 73193.
Prince , Alan. 2002 . Entailed ranking arguments . Manuscript, Rutgers
University, New Brunswick, NJ.
. 2006 . Implication and impossibility in grammatical sys-
tems: What it is and how to fnd it. Manuscript, Rutgers University,
New Brunswick, NJ.
Prince , Alan , and Paul Smolensky. [1993] 2004. Optimality
Teory: Constraint Interaction in Generative Grammar. Malden,
neurons maximizes numerical harmony; at a higher level of
description, the same system computes the symbolic structural
description of the input that optimizes the harmony of an OT
grammar. Construed in these terms, the study of grammar is
fully integrated into the contemporary science of the mind/brain
(Smolensky and Legendre 2006 ) .
Geraldine Legendre and Paul Smolensky
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Alderete , John. 2001 . Dominance efects as transderivational anti-
faithfulness. Phonology 18 : 20153.
Anttila , Arto. 1998 . Deriving variation from grammar. In Variation,
Change, and Phonological Teory, ed. F. Hinskens , R. van Hout, and
W. L. Wetzel , 35 68. Amsterdam : Benjamins .
Benua , Laura . 1995 . Output-output faithfulness. In University of
Massachusetts Occasional Papers in Linguistics 18: Papers in Optimality
Teory, ed. Jill Beckman , Laura Walsh Dickey, and Suzanne Urbanczyk ,
77 136. Amherst : University of Massachusetts at Amherst, GLSA.
Blutner , Reinhard. 2000 . Some aspects of optimality in natural language
interpretation . Journal of Semantics 17 : 189216.
Blutner , Reinhard, Helen De Hoop , and Petra Hendriks. 2006 . Optimal
Communication . Stanford, CA : CSLI Publications.
Blutner , Reinhard, and Henk Zeevat , eds. 2004. Pragmatics in Optimality
Teory . London : Palgrave Macmillan.
Boersma , Paul. 1998 . Functional Phonology: Formalizing the Interactions
between Articulatory and Perceptual Drive s . Te Hague: Holland
Academic Graphics.
Boersma , Paul , and Bruce Hayes. 2001 . Empirical tests of the gradual
learning algorithm. Linguistic Inquiry 32 : 4586.
Bresnan , Joan. 2000 . Optimal syntax. In Optimality Teory: Phonology,
Syntax and Acquisition, ed. Joost Dekkers , Frank van der Leeuw, and
Jeroen van de Weijer , 334 85. Oxford : Oxford University Press.
Bresnan , Joan, Ashwini Deo , and Devyani Sharma. 2007 . Typology in
variation: A probabilistic approach to be and nt in the survey of
English dialects . English Language and Linguistics 11: 30146 .
Burzio , Luigi. 1994 . Principles of English Stress. Cambridge: Cambridge
University Press.
Davidson , Lisa, Peter W. Jusczyk , and Paul Smolensky. 2006 . Optimality
in language acquisition I: Te initial and fnal states of the phonolog-
ical grammar. In Te Harmonic Mind: From Neural Computation
to Optimality-Teoretic Grammar. Vol. 2. Ed. Paul Smolensky and
Graldine Legendre , 793839. Cambridge, MA: MIT Press.
Eisner , J. 1997 . Ef cient generation in primitive optimality theory. Annual
Meeting of the Association for Computational Linguistics 35: 31320.
Eisner , Jason. 2000 . Easy and hard constraint ranking in optimality the-
ory: Algorithms and complexity. In Finite-State Phonology: Proceedings
of the Fifth Workshop of the ACL Special Interest Group in Computational
Phonology (Sigphon), ed. Jason Eisner , Lauri Karttunen and A. Triault ,
2233. Morristown, NJ: Association for Computational Linguistics.
Frank , Robert , and Giorgio Satta . 1998 . Optimality theory and the gen-
erative complexity of constraint violability. Computational Linguistics
24: 30715.
Gibson , Edward , and Kevin Broihier . 1998 . Optimality theory and human
sentence processing. In Is the Best Good Enough? Optimality and
Competition in Syntax, ed. Pilar Barbosa , Danny Fox, Paul Hagstrom,
Martha McGinnis, and David Pesetsky , 157 91. MIT Working Papers
in Linguistics. Cambridge, MA: MIT Press.
Grimshaw , Jane. 1997 . Projection, heads, and optimality . Linguistic
Inquiry 28: 373422.
Hayes , Bruce . 2004 . Phonological acquisition in optimality theory: Te
early stages. In Constraints in Phonological Acquisition, ed. Ren
Optimality Theory
570
can come into being without writing a puzzle for those steeped
in literate traditions that assume the centrality of the written
word; and second, how this relates to performance (for perfor-
mance is arguably how an oral creation exists).
Earlier approaches focused largely on nonliterate settings,
especially those characterized as primitive or traditional . One
model was of spontaneous improvization by the unself-con-
scious child of nature, unfettered (and unhelped) by recog-
nized artistic conventions. Another was of unchanging tradition
from the far-distant past, not composed by living creators but
stored in the communal tribal memory.
Tese models were largely superseded by the infuential oral-
formulaic approach (also known as the oral theory ) which came
to the fore in the mid-twentieth century. Te concept of oral
composition acquired a specifc meaning and became a key term
of analysis and explanation. Its classic statement in Albert Lords
seminal Te Singer of Tales ( 1960 ) used feldwork in the 1930s in
Yugoslavia to demonstrate how lengthy oral poems were com-
posed during performance: Te singers drew on a traditional
store of formulaic phrases and themes, which enabled them,
without writing or verbatim memorization, to pour forth long
epic songs in uninterrupted fow. Variations around such formu-
laic phrases as, for example, By Allah, he said, and mounted his
white horse recur throughout the poems, providing a parallel to
the Homeric epithets like feet-footed Achilles found in early,
putatively oral, Greek epics. Tis special technique of composi-
tion (Lord 1960 , 17) relied not on preplanned, memorized texts
but on composition- in -performance. Contrary to literate expec-
tations, there was no fxed correct version: each performance was
authentic in its own right, a unique product composed and per-
formed on one occasion. Oral-formulaic composing was linked
to a traditional, oral mindset incompatible with literacy and the
literate mind, and once singers became literate, it was posited,
they lost the power to compose orally.
Te oral-formulaic theory was enormously infuential
throughout much of the later twentieth century and across a
wide span of disciplines, providing, as it apparently did, an
answer to the puzzle of verbal composition without writing.
Examples of comparable formulaic expression and hence, it
seemed, of oral composition were identifed throughout the
globe, from early Greek epic, Old English texts, or the Bible to
living examples recorded from the feld, soon also extending to
the full range of poetic genres and to prose-like forms, such as
sermons or storytelling .
Tough still regarded as a classic approach, oral-formulaic
theory has been both modifed and challenged, especially dur-
ing the last two decades. First, it has become apparent that not
all genres of unwritten verbal art follow the oral-formulaic com-
position-in-performance mode, nor, as implied by the classic
oral-formulaic analysts, is oral composition a single identifable
process. Teir often somewhat generalized conclusions have not
been fully supported by the empirical evidence, for oral forms
turn out to be created in diverse ways. Some are composed before
and separated from performance. Some do, after all, involve
memorization. One much-quoted case is of the Somali poets
who spend hours, sometimes days, composing elaborate genres
of oral poetry, later delivered word for word either by themselves
or by reciters who are able to memorize poems and, without
MA: Blackwell. Original version was a technical report, Rutgers
University and University of Colorado at Boulder..
Prince , Alan, and Bruce B. Tesar. 2004 . Learning phonotactic distribu-
tions. In Constraints in Phonological Acquisition, ed. Ren Kager , Joe
Pater, and Wim Zonneveld. Cambridge: Cambridge University Press .
Rumelhart , David E., James L. McClelland, and the PDP Research Group.
1986 . Parallel Distributed Processing: Explorations in the Microstructure
of Cognition. 2 vols. Cambridge, MA: MIT Press.
Smolensky , Paul. 1986 . Information processing in dynamical sys-
tems: Foundations of harmony theory. In David E. Rumelhart , James
L. McClelland, and the PDP Research Group , Parallel Distributed
Processing: Explorations in the Microstructure of Cognition, I: 194281.
Cambridge, MA : MIT Press.
. 1988 . On the proper treatment of connectionism . Behavioral
and Brain Sciences 11 : 174.
. 1996 . On the comprehension/production dilemma in child lan-
guage . Linguistic Inquiry 27 : 72031.
Smolensky , Paul , and Graldine Legendre. 2006 . Te Harmonic
Mind: From Neural Computation to Optimality-Teoretic Grammar.
Vol. 1. Cognitive Architecture. Vol. 2. Linguistic and Philosophical
Implications. Cambridge, MA : MIT Press.
Speas , Margaret. 1997 . Optimality theory and syntax: Null pronouns and
control. In Optimality Teory: An Overview, ed. Diana Archangeli and
D.Terrence Langendoen , 171 99. Malden, MA : Blackwell.
Stevenson , Suzanne , and Paul Smolensky. 2006 . Optimality in sen-
tence processing. In Te Harmonic Mind: From Neural Computation
to Optimality-Teoretic Grammar. Vol. 2 . Ed. Paul Smolensky and
Graldine Legendre , 307 38. Cambridge, MA : MIT Press.
Tesar , Bruce B. 1996 . Computing optimal descriptions for optimality
theory grammars with context-free position structures. Proceedings of
the Tirty-Fourth Annual Meeting of the Association for Computational
Linguistics, 1017. Morristown, NJ: Association for Computational
Linguistics.
. 1998 . Error-driven learning in optimality theory via the ef -
cient computation of optimal forms. In Is the Best Good Enough?
Optimality and Competition in Syntax , ed. Pilar Barbosa , Danny Fox,
Paul Hagstrom, Martha McGinnis, and David Pesetsky , 42135. MIT
Working Papers in Linguistics. Cambridge, MA : MIT Press.
Tesar , Bruce B. , and Paul Smolensky. 1998 . Learnability in optimality
theory . Linguistic Inquiry 29 : 22968.
Wilson , Colin. 2001 a. Bidirectional Optimization and the Teory of
Anaphora. In Optimality-Teoretic Syntax, eds. Geraldine Legendre ,
Sten Vikner and Jane Grimshaw. Cambridge, MA : MIT Press.
. 2001 b. Consonant cluster neutralisation and targeted con-
straints. Phonology 18 : 14797.
Yip , Moira. 1993 . Cantonese loanword phonology and optimality the-
ory. Journal of East Asian Linguistics 2 : 26191.
ORAL COMPOSITION
Oral composition broadly refers to the creation of organized ver-
bal formulations without reliance on writing. Tough in essence
a familiar process in everyday speech, it has become a quasi-
technical and debated term, applied especially to bringing into
being relatively sustained examples of entextualized verbal art,
both ancient and recent. It has thus been of interest to linguists,
anthropologists, folklorists, psychologists, historians, and spe-
cialists in specifc languages and cultures, also linking to work
on oral culture, performance, story, literacy , and
memory .
Te central issues have been, frst and most directly, how
lengthy oral poems, narratives, and other sustained verbal forms
Oral Composition
571
better, as a multifaceted spectrum of overlaps, interpenetrations,
and diversities. Te now-infuential concepts of entextualization
and of dialogism , here applied in particular by linguistic and
literary anthropologists, have also bridged the once-accepted
chasm between oral and literate and illumined the multiple ways
in which people construct, assemble, and interact with texts
(Barber 2007 ; Silverstein and Urban 1996 ) .
Te meaning of oral itself has also been enlarged and prob-
lematized. Most oral compositions, it is now increasingly noted,
are realized not just through words but through a constellation
of multimodal resources. Te act of performance may include,
for example, movement, bodily enactment, visual devices, and
the variegated arts of the voice (volume, intonation , speed,
silence, timbre, atmosphere, and much else): A musical element
is essential in certain genres, an aspect often neglected in Western
scholars propensity to privilege the verbal component. Although
music and words are in some cultures and genres taken as dis-
tinct, composed by diferent people, this is not always so, and
some scholars argue that language and music form a continuum
rather than a dichotomy (see Banti and Giannatasio 2004 ) . Te
substantial recent work on gesture (McNeill 2000 ; Kendon
2004 ) has also elucidated the integral relation between gesturing
and speaking. Even if below our explicit consciousness, gesture,
it seems, is a planned and patterned activity, a dimension there-
fore that, like music, must arguably enter into a full understand-
ing of oral composition and performance .
Recent approaches to memory are also relevant. Historians,
anthropologists, and psychologists have drawn attention to the
frames within which remembering is actively recreated and
to the diverse social mechanisms for organizing and manipu-
lating memory. Some cultures or genres prioritize word-for-
word memorization and organize formal or informal training
in this skill; in others, diferent arts are emphasized, including
improvization. Generic conventions themselves provide sche-
mas for organizing and activating memory, ofering constraints
and opportunities for the creative fow of language, not only
through larger frames such as narrative , praising, or lamenting
but also by memory-enhancing devices like imageries, rhythm ,
and audience (and chorus) participation and by sound-pattern
repetitions and sequences, such as rhyme, alliteration,
parallelism, or melody. In some contexts, memory is seen as
itself an aspect of creativity , eroding its apparent opposition
to composition (see further Rubin 1995 ; Carruthers 1990 ).
Te upshot is that oral composition has somewhat dissolved
as a distinctive topic for analysis. It no longer stands out as some-
thing self-evidently special or puzzling but as an aspect of pro-
cesses being studied from other viewpoints and as taking place
in many diferent forms, settings, and modalities from lengthy
art genres to the creativity of everyday conversation; from long
preplanned and rehearsed performances to extemporized
speeches; from live delivery to multimedia enactments. It is now
tied less to theories of the primitive, traditional, or, indeed, the
oral as such as to ongoing issues related to language or creativ-
ity more generally, analyzed both comparatively and in cultural
specifcities.
While in one way this has undermined the idea of oral com-
position as a subject for direct scrutiny in its own right, in another
way this broader cross-cultural approach and the empirical
writing, store large and exactly reproducible repertoires in their
memory over many years. Elsewhere, too, prior composition is
sometimes a long-drawn-out and carefully considered proce-
dure, in some cases involving multiple authors and/or rehears-
als before being performed. Certain womens personal songs in
mid-twentieth-century Zambia, for example, were thought out
by one woman, elaborated with her friends, worked over for days
by an expert composer, then rehearsed and memorized before
fnal performance. In other cases, a composer may speak aloud
words of rapid inspiration designed for later performance, to
be captured by listeners on the spot through memorizing, tape
recording, or writing (further details and discussion in Finnegan
1992 , 5287; 2007 , 96113, 179200). Contrary to the classic oral-
formulaic model, oral composing varies in diferent cultures,
genres, and circumstances .
Second, the assumption that literacy and orality are mutually
incompatible has been extensively challenged. By now, many
empirical examples of their interaction in both historical and
more recent times have been noted and investigated. At a more
theoretical level, there are also the current transdisciplinary cri-
tiques of the West-centered binary dichotomizing between primi-
tive/civilized, non-Western/Western, traditional/modern, and,
alongside these, oral/literate, together with parallel challenges
to the arguably ethnocentric and ideological presuppositions of
a simple and necessary link between literacy and modernity. In
practice, it appears, there are multiple forms of literacy, interact-
ing, therefore, in multiple ways with oral modes .
Despite challenges to some of its central presuppositions, the
legacy of the oral-formulaic school lives on. It rightly unsettled
the (literate) concept of fxed correct text, highlighted the signif-
icance of performance and audience, and, if in the (arguably)
somewhat elusive terminology of formulae, pointed up the
importance for composition of conventionalized verbal formu-
lations in generic settings. Scholars identifying themselves with
that tradition have continued their (largely textual) examina-
tions of oral and oral-derived texts while also reconfguring
their approaches by attention to the specifcities of aesthetic and
cultural traditions, interacting fruitfully with trends elsewhere to
produce sophisticated analyses of the complex interrelations of
oral with written composition (Amodio 2005 ; Foley 2002 ) .
Although there is currently no one dominant approach to
complement the earlier oral theory, the topic of composition
without writing (or anyway, without central reliance on writing)
has continued to attract interdisciplinary interest. Te focus is
now less on attempting to delineate oral composition as a single
process, or as pertaining to some special kind of culture or men-
tality, and more on complexity and plurality.
Oral composition is thus no longer conceptualized as primar-
ily confned to traditional, historic, or non-Western settings
but as also including such examples as contemporary popular
songs or the spoken oratory of modern statesmen and publi-
cists. It has also been noted how readily some long-established
oral genres are exploited in new settings, like the South African
praise poems now composed for Nelson Mandela , the national
football team, or university graduation ceremonies, and circu-
lated not only in live performance but in writing and on radio,
CD-ROMs, and the Web. Te relation between oral and literate is
now more often envisaged as continuum than as opposition or,
Oral Composition
572
culture to certain degrees, but without ever eliminating it or
totally superseding it. In the world today, an estimated one bil-
lion people do not know how to read or write any language, and
so they live in a residual form of primary oral culture. In addi-
tion, certain cultures in the world today remain highly oral, just
as Western culture did for centuries before print culture helped
usher in what is commonly referred to as modern culture and
modernity modern science, modern capitalism, modern
democracy, the Industrial Revolution, and the Romantic move-
ment. Te common distinction between modern culture, promi-
nent in the West, and premodern cultures in many other areas
of the world (e.g., Turner 1969 ) can be understood in terms of
Ongs account of Western cultural history. Premodern cultures
are examples of what Ong has referred to as primary oral cultures
and as residual forms of primary oral cultures.
When alphabetic writing was introduced, Ong claims (and so
do McLuhan and Havelock), it did not change everything over-
night. As a result, early writing such as most of the Bible (except
for the prologue to the Gospel of John) and the Homeric epics
can be seen as providing transcripts of primary oral thought
and expression. But distinctively literate forms of thought and
expression emerged in the pre-Socratics and Plato, as Havelock
explains in detail ( 1963 , 1978 , 1982 ). Perhaps more than anything
else, Ong sees the formal study of logic initiated by Aristotle as
involving distinctively literate thought; Ong has traced the his-
tory of the formal study of logic in his 1958 masterwork Ramus,
Method, and the Decay of Dialogue: From the Art of Discourse to
the Art of Reason (3d ed. 2004 ) . Within the Aristotelian tradition
of medieval logic, Ong notes, new developments emerged that he
styles the quantifcation of thought (see esp. [1958] 2004, 5391).
In a subsequent essay, he points out how these new develop-
ments contributed to the emergence of a new state of mind as
found in modern science (Ong 1962 , 72). Neither this new state
of mind nor modern science emerged in oral culture, just as the
formal study of logic developed by Aristotle has no counterpart
in oral culture .
In Ramus, Method, and the Decay of Dialogue , Ong also calls
attention to the visualist tendencies of Western philosophic
thought, which were advanced further by the development of
printed books. Te visualist tendencies of ancient Greek philo-
sophic thought have been further amplifed recently by Andrea
Nightingale ( 2004 ). But such visualist tendencies do not charac-
terize oral culture. Ong ([ 1969 ] 1995) also describes oral culture
as based on an oral-aural sense of the world as event, which he
contrasts with the visual sense of the world as something seen
(as in the expression worldview ). In World as Event: Aspects of
Chipewyan Ontology (1997), anthropologist David M. Smith
borrows Ongs expression world as event to help elucidate cer-
tain aspects of Chipewyan thought.
Ong associates oral culture with the cyclic forms of thought
that Mircea Eliade describes in Te Myth of the Eternal Return
([1949] 2005). Lynne Ballew describes further examples of cyclic
thought in Straight and Circular: A Study of Imagery in Greek
Philosophy (1979). Ong sees the recycling of souls in the story
of Er recounted by Socrates in Platos Republic as an instance
of cyclic thought in Greek philosophy, which is to say a residual
form of oral thought in Greek philosophic thought. Conversely,
Ong (see, for example, 1967a, 6182, 8398, 99126) associates
investigations it has stimulated have enabled a frmer grasp on
the complexity of the processes by which, without much or any
direct recourse to writing, people can and do produce verbal
formulations both lengthy and short, aesthetically marked and
everyday. Further, all of this has helped to challenge traditional
models of language as realized preeminently either, on the one
hand, in stable written texts or, on the other, in relatively uncon-
strained and perhaps trivial everyday speech. A consideration of
oral composing highlights the sustained and creative marshal-
ing of language in situations where writing does not necessarily
lie at the core: verbal genres that are by no means outdated or
peculiar but have had a wide spread in the world, both yesterday
and today .
Ruth Finnegan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Amodio , Mark C. , ed. 2005. New Directions in Oral Teory. Tempe : Arizona
Center for Medieval and Renaissance Studies.
Banti , G., and F. Giannatasio. 2004 . Poetry. In A Companion to Linguistic
Anthropology, ed. Alessandro Duranti , 290 320. Oxford: Blackwell.
Barber , Karin. 2007 . Texts, Persons and Publics in Africa and Beyond.
Cambridge : Cambridge University Press.
Carruthers , Mary. 1990 . Te Book of Memory . Cambridge : Cambridge
University Press.
Finnegan , Ruth. 1992 . Oral Poetry: Its Nature, Signifcance and Social
Context. 2d ed. Bloomington : Indiana University Press.
. 2007 . Te Oral and Beyond: Doing Tings with Words in Africa .
Oxford : James Currey. Chicago: Chicago University Press.
Foley , John Miles. 1988 . Te Teory of Oral Composition: History and
Methodology . Bloomington : Indiana University Press.
. 2002 . How to Read an Oral Poem. Urbana : University of Illinois
Press.
Kendon , Adam. 2004 Gesture: Visible Action as Utterance.
Cambridge : Cambridge University Press.
Lord , Albert B. 1960 . Te Singer of Tales. Cambridge: Harvard University
Press.
McNeill , David , ed. 2000. Language and Gesture . Cambridge : Cambridge
University Press.
Rubin , David C . 1995 . Memory in Oral Traditions: Te Cognitive
Psychology of Epic, Ballads, and Counting-Out Rhymes. Oxford : Oxford
University Press.
Silverstein , Michael , and Greg Urban , eds. 1996. Natural Histories of
Discourse . Chicago: Chicago University Press.
ORAL CULTURE
Oral culture is a conceptual construct associated primarily with
the work of Walter J. Ong , S.J., Marshall McLuhan , and Eric A.
Havelock , whereas the term oral tradition , which they also use,
is more often associated with the work of Milman Parry , Albert B.
Lord , and their many followers (see Foley 1985 ). Ong, McLuhan,
and Havelock use the term oral culture to refer primarily to pre-
literate cultures but also to characterize the thought and expres-
sion that carry over into manuscript culture and even into print
culture . Moreover, oral culture, which Ong also refers to as
primary oral culture, endures in the sense that people continue
to talk with one another . Te later subsequent cultural devel-
opments in manuscript culture and print culture may be seen
as cultural overlays that infuence and transform the base oral
Oral Culture
573
Brakke , David. 2006 . Demons and the Making of the Monk: Spiritual
Combat in Early Christianity. Cambridge: Harvard University Press.
Cary , Phillip. 2000 . Augustines Invention of the Inner Self: Te Legacy of a
Christian Platonist . New York : Oxford University Press.
Connor , James L. 2006 . The Dynamism of Desire: Bernard J. F.
Lonergan, S.J., on The Spiritual Exercises of Saint Ignatius of Loyola.
St. Louis, MO: Institute of Jesuit Sources.
Daniell , Beth. 1986 . Against the great leap theory of literacy. Pre/Text
7 .3/4: 18193. Also see Ong 1987.
Draper, Jonathan A. 2004. Orality, Literacy, and Colonialism in Antiquity.
Leiden : Brill.
Draper , Jonathan A., ed. 2003. Orality, Literacy, and Colonialism in
Southern Africa. Leiden : Brill.
Eliade , Mircea. [1949] 2005. Te Myth of the Eternal Return. 2d ed. Trans.
Willard R. Trask , new introduction by Jonathan Z. Smith. Princeton,
NJ : Princeton University Press.
Farrell , Tomas J. 2000 . Walter Ongs Contributions to Cultural
Studies: Te Phenomenology of the Word and I-Tou Communication.
Cresskill, NJ : Hampton.
Fixico , Donald L. 2003. Te American Indian Mind in a Linear
World: American Indian Studies and Traditional Knowledge. New
York : Routledge.
Foley , John Miles. 1999 . Homers Traditional Art. University
Park : Pennsylvania State University Press.
Foley, John Miles, ed. 1985. Oral-Formulaic Teory and Research: An
Introduction and Annotated Bibliography. New York : Garland.
Havelock , Eric A. 1963 . Preface to Plato. Cambridge : Belknap Press/
Harvard University Press.
. 1978 . Te Greek Concept of Justice: From Its Shadow in Homer to Its
Substance in Plato. Cambridge : Harvard University Press .
. 1982 . Te Literate Revolution in Greece and Its Cultural
Consequences. Princeton, NJ : Princeton University Press.
Horsley , Richard A., Jonathan A. Draper, and John Miles Foley , eds.
2006. Performing the Gospel: Orality, Memory, and Mark. Minneapolis,
MN : Fortress.
Jousse , Marcel. 1990 . Te Oral Style. Trans. Edgard Sienaert and Richard
Whitaker. New York: Garland.
Kahler , Erich. 1973 . Te Inward Turn of Narrative. Trans. Richard
Winston and Clara Winston , foreword by Joseph Frank . Princeton,
NJ : Princeton University Press.
Kelber , Werner H. 1997 . Te Oral and the Written Gospel: Te Hermeneutics
of Speaking and Writing in the Synoptic Tradition: Mark, Paul, and Q .
2d ed. Foreword by Walter J. Ong , S.J., new introduction by Werner H.
Kelber. Bloomington : Indiana University Press.
Lee , Dorothy. 1987 . Freedom and Culture. Prospect Heights,
IL : Waveland.
Lord , Albert B. 1960 . Te Singer of Tales. Cambridge : Harvard University
Press.
Low , Anthony. 2003 . Aspects of Subjectivity: Society and Individuality from
the Middle Ages to Shakespeare and Milton. Pittsburgh, PA : Duquesne
University Press.
Mansfeld , Harvey C. 2006 . Manliness. New Haven, CT : Yale University
Press.
McLuhan , Marshall. 1962 . Te Gutenberg Galaxy: Te Making of
Typographic Man. Toronto : University of Toronto Press.
McNamee , Maurice B. 1960 . Honor and the Epic Hero: A Study of the
Shifting Concept of Magnanimity in Philosophy and Epic Poetry.
New York : Holt, Rinehart and Winston.
Morris , Jan , and Barry Powell , eds. 1997. A New Companion to Homer.
Leiden : Brill.
Nightingale , Andrea . 2004 . Spectacles of Truth in Classical Greek
Philosophy: Teoria in Its Cultural Context. Cambridge : Cambridge
University Press.
the linear accounts of time in the Bible with literacy ; he likes
to style linear conceptions of history as evolutionary thought,
thereby rooting later forms of evolutionary thought in Darwin
and others within the biblical cultural tradition in Western tra-
dition. Independently of Ong, Donald L. Fixico, who is himself
of American Indian descent, works comfortably with these con-
trasts in Te American Indian Mind in a Linear World: American
Indian Studies and Traditional Knowledge (2003; also see Lee
1987 , 10520) .
In Manliness , Harvey C. Mansfeld does not happen to refer
explicitly to oral culture, but he refers to Achilles frequently to
illustrate certain points regarding manliness ( 2006 , 558, 601),
an ambivalent quality that he sees as needing to be disciplined
toward socially constructive ends. Male puberty rites, for exam-
ple, have long been used in oral cultures to help discipline and
orient young men in socially constructive ways (see van Gennep
1960 ; Ong 1971 , 11341). Te kind of socially constructive war-
rior manliness that Achilles and Agamemnon and Hector and
Odysseus represent is a necessity in oral cultures: Te entire
enterprise of modernity, however, could be understood as a proj-
ect to keep manliness unemployed (Mansfeld 2006 , 230). In
David Riesmans terminology, oral culture is tradition directed,
whereas modernity is dominated by inner-directedness (1950).
(For further recent studies of the historical development of inner-
directedness, see Williams 1993 ; Brakke 2006 ; Cary 2000 ; van t
Spijker 2004 ; Renevey 2001 ; Low 2003 ; Connor 2006 ; Bloom 1998 ;
Kahler 1973 ). In Honor and the Epic Hero , Maurice B. McNamee
(1960) shows that concepts about heroic and great-spirited per-
sons have shifted from time to time. Even though the concepts of
a heroic and magnanimous person in oral culture no longer work
for modernity, we do need to formulate some concepts of heroic
and magnanimous persons that will work for modernity .
Before concluding, we should note the critique that some
authors have made of Ongs work and related work regarding
oral culture. Te critique alleges that Ong has set forth a great
divide theory in which there is a great divide with oral culture
when literacy emerges (see, for example, Daniell 1986 ; but also
see Ongs 1987 letter about her article). Beth Daniell and others
who advance this critique do not accurately summarize what
Ong has said, and so their supposed critique amounts to little
more than knocking down a straw man named by them Ong.
(For a more detailed response to this alleged line of critique, see
Farrell 2000 , 1626, 2004).
In conclusion, in oral culture, people are culturally condi-
tioned so that they tend to favor cyclic patterns of thought and
expression, to have a world-as-event sense of life, to put manli-
ness to work in socially constructive ways, to use oral stories of
heroes as ways to help orient and put manliness to work, and to
use ritual process very efectively to promote and support socially
constructive behavior .
Tomas J. Farrell
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ballew , Lynne. 1979 . Straight and Circular: A Study of Imagery in Greek
Philosophy . Assen, the Netherlands : Van Gorcum.
Bloom , Harold. 1998 . Shakespeare: Te Invention of the Human. New
York : Riverhead Books.
Oral Culture
574
necessary for the elimination of the confusing ambiguities of
ordinary language. Advocates of the ordinary language approach
may recognize the power, the frequent utility, and the intellec-
tually admirable parsimony of the formalist, but they also insist
that ordinary linguistic practice as it stands is generally appro-
priate for our use without any preemptory need for a compre-
hensive and indispensable reform, and, moreover, that ordinary
usage contains helpful distinctions and nuances that would be
hurtfully eliminated if the rigors of a formalist system were to be
imposed as the ultimate standard. Peter Strawson , J. L. Austin ,
and John Searle have been, in varied ways, advocates of an ordi-
nary language approach, and undoubtedly Ludwig Wittgensteins
move from the rigors of the Tractatus to the complexities of the
Philosophical Investigations is seminal to the entire movement.
A short entry cannot survey all the philosophers and topics
of importance in this ordinary language approach, but as prime
examples this entry will focus on one philosopher, Peter Strawson,
and on one topic, reference. We will consider the way in which
Strawson makes reference his starting point, how this single topic is
embedded in contemporary debates within the philosophies of lan-
guage and logic, Strawsons own signifcant contributions to those
debates, and some replies he makes to his critics. Also noted will be
Bertrand Russell and Wittgenstein as prime movers in Strawsons
thought and W. V. O. Quine as a stern and characteristic critic.
Te best known of all Strawsons writings is probably his early
article On referring ( 1950 ) in which he addresses the issue of
singular reference and predication and their objects, a matter with
which he was concerned throughout his working life. Tis article
was written in response to Bertrand Russells theory of defnite
descriptions contained in On denoting ( 1905 ) . For Strawson,
we use a variety of expressions to refer to some individual person,
object, or event. We use singular descriptive pronouns ( this and
that ), proper names ( Winston Churchill ), and singular pronouns
( I , you , it ), and for what are called defnite descriptions, we use
the defnite article followed by a noun in the singular, e.g., the
king of France . Suppose someone at present utters the sentence
Te king of France is wise (S). For Bertrand Russell, S is signif-
cant; that is, it may be true or false. But he claims that to show the
true logical form of S, it needs to be rewritten as
(1) Tere is a king of France.
(2) Tere is not more than one king of France.
(3) Tere is nothing that is the king of France and is not
wise.
Tus, someone uttering S today would be saying something
signifcant but false. For Russell, we must distinguish defnite
descriptions such as the king of France from logically proper
names, for example, Winston Churchill. Te latter alone can be
subjects of sentences of a genuine subject-predicate form and
have some single object for which they stand.
Strawson thinks that Russell is wrong in this, since his account
of sentences 13 is neither completely nor even partially correct.
A correct account must begin by distinguishing among
a sentence
a use of a sentence
an utterance of a sentence
Ong , Walter J. [1958] 2004. Ramus, Method, and the Decay of
Dialogue: From the Art of Discourse to the Art of Reason. 3d ed. New
foreword by Adrian Johns. Chicago : University of Chicago Press.
.1962. Te Barbarian Within: And Other Fugitive Essays and Studies.
New York : Macmillan.
. 1967 a. In the Human Grain: Further Explorations of Contemporary
Culture . New York : Macmillan.
. 1967 b. Te Presence of the Word: Some Prolegomena for Cultural
and Religious History. New Haven, CT : Yale University Press.
. [1969] 1995. World as view and world as event. In Faith and
Contexts , III: 6990. Atlanta: Scholars Press.Originally printed in
American Anthropologist 71 (August): 63447.
. 1971 . Rhetoric, Romance, and Technology: Studies in the Interaction
of Expression and Culture. Ithaca, NY : Cornell University Press.
. 1977 . Interfaces of the Word: Studies in the Evolution of
Consciousness and Culture . Ithaca, NY : Cornell University Press.
. 1981 . Fighting for Life: Contest, Sexuality, and Consciousness.
Ithaca, NY : Cornell University Press.
. 1986 . Hopkins, the Self, and God. Toronto : University of Toronto
Press.
. 1987 . Letter to the Editor. Pre/Text 8 .1/2: 155. Comments on
Daniell 1986.
. 1992 99. Faith and Contexts. 4 vols. Ed. Tomas J. Farrell and
Paul A. Soukup. Atlanta: Scholars Press. Volumes now distributed by
Rowman & Littlefeld.
. 2002 a. An Ong Reader: Challenges for Further Inquiry. Ed. Tomas
J. Farrell and Paul A. Soukup . Cresskill, NJ : Hampton.
. 2002 b. Orality and Literacy: Te Technologizing of the Word. 2d ed.
New York : Routledge.
Opland , Jef. 1983 . Xhosa Oral Poetry. Cambridge : Cambridge University
Press.
Parry , Milman. 1971 . Te Making of Homeric Verse: Te Collected Papers
of Milman Parry. Ed. Adam Parry. New York : Oxford University Press.
Renevey , Denis. 2001 . Language, Self, and Love: Hermeneutics in the
Writings of Richard Rolle and the Commentaries on the Song of Songs.
Cardif: University of Wales Press.
Riesman, David, with Rueul Denny and Nathan Glazer . 1950 . Te Lonely
Crowd: A Study of the Changing American Character. New Haven,
CT : Yale University Press.
Scholes , Robert , and Robert Kellogg , with a chapter by James Phelan.
2006 . Te Nature of Narrative. 2d ed. New York : Oxford University
Press.
Smith , David M . 1997 . World as event: Aspects of Chipewyan ontology.
In Circumpolar Animism and Shamanism, ed. Takako Yamada and
Takashi Irimoto , 67 91. Sapporo, Japan : Hokkaido University Press.
Turner , Victor. 1969 . Te Ritual Process: Structure and Anti-Structure.
Chicago : Aldine.
van Gennep , Arnold. 1960 . Te Rites of Passage. Trans. Monika B.
Vizedom and Gabrielle L. Cafee , introduction by Solon T. Kimball .
Chicago : University of Chicago Press.
v an t Spijker , Ineke . 2004 . Fictions of the Inner Life: Religious Literature
and Formation of the Self in the Eleventh and Twelfth Centuries.
Turnhout, Belgium : Brepols.
Walker , Jefrey . 2000 . Rhetoric and Poetics in Antiquity. New York: Oxford
University Press.
Williams , Bernard. 1993 . Shame and Necessity. Berkeley and Los
Angeles : University of California Press.
ORDINARY LANGUAGE PHILOSOPHY
Within the analytic tradition of contemporary Anglophone phi-
losophy, ordinary language philosophy is set in contrast to the
view that the prescriptions of formal logic provide the means
Ordinary Language Philosophy
575
distinction when we say that a statement is untrue or that it is
inconsistent. If a deductive argument is valid, if the premises are
true, then the conclusion is judged necessarily true under pain of
inconsistency or self-contradiction. But in all of this we must also
consider the context of statements that are made. Asked if the
results of the recent election pleased me, I may signifcantly reply
that they did and they didnt. Words such as vehicle and enter-
tainment have only approximate boundaries for their appropri-
ate use. Te uses and therefore the meanings of various words
and expressions are subject to expansion and contraction.
Tus, as in On referring, logical appraisal is properly applied
to statements, not sentences. We need, therefore, to approach
the relation between formal logic and ordinary language with
caution. In formal logic, a formula is an expression such that by
substituting words or phrases for the variables we can obtain
sentences that could be used to make statements. In the formula
x is a younger son, to substitute Tom for x would yield a
sentence that would have meaning, while to substitute Te
square root of 2 would not. Tus, some variables would yield
sentences, but not signifcant statements. We can talk about the
range of admissible values for a variable, but, unlike formal logic,
in ordinary language there are no precise rules for what is admis-
sible. Once again, statements have a contextual component, and
that goes beyond the reach of formal logic .
Te limits of formal logic are also manifest in its use of sym-
bols for truth- functional connectors. Consider particularly the
logical symbol . Consider If it rains, the party will be a failure.
Tat suggests conditions that are neither logical nor linguis-
tic, but are instead discovered in our experience of the world.
Compare the function of the connector in that sentence with the
function of the same connector in If he is a younger son, then he
has a brother. Similar limitations are evident in the use of other
connectors. Consider the question of the connector that is the
appropriate one for unless .
Again, when logicians choose the pattern for their represen-
tative rules, they employ common uses drawn from ordinary lan-
guage, and then proceed to make standard what is common. In
this way, a rigidity is imposed that is foreign to the uses of ordi-
nary language.
Te logician is not a lexicographer but is concerned only with
general principles that are indiferent to subject matter. Te dif-
fculty here is that sometimes diferent expressions may have the
same uses; all , the , and a may have the same use in describing
the basic move of the pawn in chess. Similarly, the same expres-
sion may have diferent uses; not and not may be used as a
double negation, to emphasize, or to show necessity. Te logi-
cian would eliminate this complexity and clutter and impose a
systems rules to cure the perceived defciencies of ordinary lan-
guage. Te logician is content with all , some , and no but has no
use for most or few , despite the usefulness and common employ-
ment of those terms in our ordinary reasoning.
Tere are further challenges to any claim for a sovereignty of
formal logic over the workings of ordinary language. Consider
the notion of logical form as a sort of verbal skeleton that remains
when all expressions, except the selected logical constants, are
eliminated from a sentence that might be used as a statement
and are replaced by variables. For Strawson, this notion of logi-
cal form is viable, but it may lead us to the mistaken conclusions
Te sentence Te king of France is wise can be uttered at
various times and for various purposes. We cannot say that the
sentence is true or false, only that it may be used to make a true
or false assertion. At the heart of Strawsons position is the claim
that referring is not something that an expression such as the
king of France does. Referring is, instead, characteristic of the
use of an expression. Meaning is a function of the sentence or
expression, but mentioning and referring and truth and falsity
are functions of the use of the sentence or expression.
Russells claim is that someone at present uttering Te king
of France is wise (S) would a) be making a true or false state-
ment, and b) be asserting that there exists at present one and
only one king of France. Strawson fnds Russell wrong on both
counts. For Strawson, the sentence is signifcant, since it could
be true or false, and it could refer to a particular person. But that
does not mean that any particular use of the sentence must be
either true or false. Ordinarily, a person uttering S presupposes
the existence of the king, and his uttering S neither asserts nor
entails the kings existence. Tus, presupposition must be care-
fully distinguished from both assertion and entailment.
Tere is, moreover, a need to distinguish rules for referring
from rules for ascribing and attributing. Tat distinction roughly
corresponds to the grammatical distinction between subject and
predicate. For Strawson, that irreducible distinction has been
blurred by logicians in their desire to reduce or to eliminate
altogether the referring use. He fnds a prime example of that
attempted elimination in Gottfried Leibnizs efort to establish
individual identity through the use of complete individual con-
cepts done in exclusively general terms. Strawson thinks that
Russell also strives to make logic in a narrow sense adequate for
referring to individuals.
It is particularly noteworthy that Strawsons fundamental dis-
tinctions between sentence and utterance, and between referring
and describing, are a challenge to the votaries of modern logic.
Consider such nonuniquely referring expressions as all , no , some ,
and some are not , that is, the four types of standard form cat-
egorical propositions: A, E, I, and O. For the modern, only I and O
propositions have existential import. In consequence, the mod-
ern must deny some traditional doctrines, such as the square of
opposition and the validity of some forms of the syllogism. Te
moderns dilemma is for Strawson a bogus one. We may sim-
ply say that the question of whether or not the quantifcational
expressions are being used to make true or false statements just
do not arise except that when the existential conditional is ful-
flled for the subject term, then all of the laws of traditional logic
hold good. If we ask a literal-minded and childless man if all of
his children are asleep, he will not answer either yes or no
because the question simply does not arise .
For Strawson, neither Aristotelian nor Russellian rules give
the exact language of any expression of ordinary language, since
ordinary language has no exact logic.
In light of On referring, Strawson sets out in Introduction
to Logical Teory ( 1952 ) to remedy the failures of modern logi-
cians to address adequately the relationships between formal
logic and the logical features of ordinary language. He begins by
noting diferences among the various ways we make judgments
about what someone says. To say that a statement is logical is
ordinarily a commendation. Tere is a further and more complex
Ordinary Language Philosophy
576
Not all have agreed. Quine is notable for his long-standing
disagreement on the issue of singular terms and reference, the
issue which is central in Strawsons On referring and perva-
sive in all of his later writings. For Quine, singular terms are at
best superfuous, to be eliminated without loss; to be is to be the
value of a variable ( 1972 , 234). Here is the great divide that, on
Strawsons account, separates him from both Quine and Russell.
Quines concern that singular terms are ambiguous in their ref-
erence is set aside by Strawson on the ground that such terms do
not refer at all; they are, instead, used by persons to make refer-
ence. If the reference is ambiguous, the responsibility rests with
the statement maker, not the term. For that matter, ambiguity
has its own uses and, indeed, its own occasional sweetness in the
ordinary language of daily life.
Strawson found strong support for his views on ordinary lan-
guage in Wittgensteins transition from his positions in Tractatus
to those in the Philosophical Investigations. Tis is manifest in
Strawsons review of the latter work in Mind ( 1954 ). In sections
38137 of the Investigations , Strawson fnds an evident rejection
of the logical atomism that characterizes the Tractatus. In that
earlier work, Wittgenstein had been concerned with the idea of
the genuine names of a language, and with the idea of the simple
indestructible elements of reality that are only to be named, not
described or defned, and which are the meanings of those genu-
ine names. Tese primary elements are Russells individuals,
and the objects of the Tractatus . Tose elements are connected
to the belief that the clarifcation of ordinary language depends
on an analysis in which ambiguous sentences are replaced by
ones that refect exactly the logical form of the fact under con-
sideration. Logic then seems to be pure, exact, and general, the
essence of the thoughts that mirror the structure of the empirical
world. Tat leads us to the illusion that this process of analysis
is fnite, that there is a single completely resolved form for every
expression.
For Wittgenstein in the Philosophical Investigations , and for
Strawson, the cure for this illusion is to give up the search for the
very essence of language and to direct our attention, instead, to
the various ways in which language actually functions, In a well-
known example, Wittgenstein asks us what is common to all of
the proceedings we call games. We cannot say that they must all
have something in common, an essence, simply because they
have a common name. Tere is no single element they all share.
Tere are only family resemblances, a network of overlappings
and crisscrosses .
What is true of games is true of linguistic activity; there is no
single use, only family resemblances. Tere is no exact bound-
ary of use, although a fxed boundary could be set to serve some
particular purpose. A word or a linguistic practice need not be
exact in order to be understood and acted upon; stand roughly
here may be serviceable enough. To say in dispraise that it is
inexact misses the mark. Te demand for absolute and fxed
meanings is senseless. Whether or not there is enough preci-
sion is determined by whether the concept is used with general
agreement.
Te consequence is that we are not to provide ordinary lan-
guage with a necessary revision and reduction; we are simply to
describe the ways it works. If we do so, we eliminate the puzzles
that arise when language goes on holiday, when we consider
that a statement must have just one logical form, or that logical
form makes the work of the lexicographer superfuous, or that
logical features need not take into account the relevant subject
matter, or that validity depends upon form, rather than the other
way about.
Te claim may be made that appropriate caution will enable
us to avoid such mistakes, and that the relation between ordin-
ary language and formal logic might be seen minimally as peace-
ful coexistence, and more truly as a separation of powers that is
necessary and useful to both sides. For Strawson, the ongoing
dif culty here is that the logician is not content with being con-
sistent but seeks the completeness of a system. Tat ideal is com-
promised by the fact that the typical truth-functional connectors
defy a single ordinary use, and the attendant complexities run
counter to a mathematical model taken as the paradigm for the
whole of logic. Tat paradigm appeals, but its seduction mis-
leads, with profound consequences for the study of metaphysics
and epistemology.
Of all of the identifcations between the truth-functional con-
nectors of formal logic and ordinary words, Strawson fnds con-
junction and negation least troublesome, but even here there are
limitations. By the laws of formal logic, pq and qp are equiva-
lent, but in ordinary language, the order may be essential to the
meaning. Most troublesome in the identifcation of logical con-
nectors and ordinary words is . Te falsity of the antecedent
suf ces in material implication for the truth of the statement, but
not in the corresponding hypothetical statement.
Te workings of the class system of modern quantifcational
logic further compound these dif culties about the relation
between ordinary language and truth-functional logic. Modern
orthodoxy claims that once the older Aristotelian system is
cleaned up, it is simply a small part of todays quantifcational
logic. Conversely, Strawson contends that with only a few res-
ervations, the traditional rules dating from Aristotle conform to
the use of words in ordinary language, and indeed avoid some
of the incongruities of the moderns practice. Standard criticism
of tradition rests on the question of existential import, that is, for
the four moods A (all), E (no), I (some) and O (some are not ),
whether there is a commitment in the tradition to the actual
existence of the members of the terms. Te moderns assump-
tion is that only I and O have such import. But consider some-
ones saying All Johns children are asleep. Again, if John is
fatherless, the existential import question simply does not come
up. Te existence of those children is a necessary precondition of
the statement being either true or false. Te modern goes wrong
in failing to distinguish sentence from statement. Te sentence
may be true or false, that is, meaningful, but in its use as a state-
ment, the question of existential import is determined by the
context.
In sum, for Strawson there are two kinds of logic, the entail-
ment rules of formal logic, which abstract from the time and
place of utterance, and the referring rules , which lay down the
contextual requirements of what a sentence presupposes. In the
study of those referring rules, we do not fnd the elegance and
system of formal logic, but Strawson does fnd a feld of intellec-
tual study unsurpassed in richness, complexity, and the power
to absorb. Te two kinds of logic are interrelated, and both are
necessary in human communication .
Ordinary Language Philosophy
577
Direct Versus Indirect Selection
Te notion that language constituted an evolutionary adaptation
sensu stricto in other words, that it arose through some selec-
tive pressure acting directly upon pre-existing genetic material
had been around since Darwin but is most cogently expressed
by Steven Pinker and Paul Bloom ( 1990 ). Arguing against the
suggestion, made by Stephen Jay Gould among others, that lan-
guage could be a spandrel an accidental by-product of other
evolutionary developments these authors pointed out that the
intimate interconnections among the various parts of language
parallel a similar interconnectivity in the eye, an object univer-
sally agreed to have evolved through natural selection. Although
their approach entailed a gradual process of evolution, they did
not address the initial stage of that process nor discuss in detail
possible adaptive pressures (beyond suggesting that competi-
tion among humans was probably more infuential than envi-
ronmental factors).
While few if any scholars would deny that selective pressures
have played a role in the development of many prerequisites for
language, some still suggest that the emergence of language itself
was not specifcally selected for. Te notion that language was an
invention by human ancestors with expanded brains is still held
by some (e.g., Donald 1991 ). Others propose that laws of form
afecting brain structure and growth played a more signifcant
role than natural selection (Jenkins 2000 ; see also biolinguis-
tics ). Alternatively, a mutation or the modifcation of some prior
nonlinguistic faculty might have yielded recursion, the capac-
ity to generate infnite structures from fnite materials (Hauser,
Chomsky, and Fitch 2002 ), and recursion added to prehuman
conceptual structure might have suf ced to produce language.
Tese last two proposals imply that language emerged in more
or less its current form, without any intermediate stage between
animal communication and true language. Approaches of
this type would be strengthened if the required laws of form,
mutations, or changes of function could be precisely specifed;
this has not yet been done .
Selective Pressures
Among those who see language as an adaptation, explanations
for the selective pressure involved have changed over time. Until
the 1980s, it was widely assumed that language arose for pur-
poses of tool making and/or cooperative hunting. However, eco-
logical studies revealed complex cooperative hunting patterns in
nonhuman species, while anthropological studies showed that
preliterate peoples made tools and taught tool making largely
without using words. Moreover, ethological studies of ape spe-
cies showed highly complex societies in which individuals com-
peted with and sought to deceive and outwit one another (Byrne
and Whiten 1992 ). An infuential essay (Humphrey 1976 ) had
already suggested that higher cognitive faculties, including lan-
guage, had most likely been generated through intense within-
group competition .
Te view that language arose from social intelligence is
nowadays shared by a majority, but it has problems. Social
competitiveness is far from unique to humans; so why has no
form of language, however rudimentary, evolved in other pri-
mate species? A unique adaptation suggests a unique pres-
sure. Furthermore, there must surely have been a stage when
words and sentences in abstraction from their ordinary uses. For
Wittgenstein, philosophys proper task is simply the assembling
of a series of reminders of actual uses, with the purpose of dispel-
ling confusions that arise in specifc contexts.
Tat conception of philosophy is one that Strawson shares up
to a point, but he also fnds that an appropriate philosophy of lan-
guage provides the basis for a descriptive metaphysics, one that
is content to give an account of the actual structure of the world
of our experience. Tis is set in contrast to a revisionary meta-
physics that vainly strives to do better. Descartes, Leibniz, and
George Berkeley are revisionary; Kant and Aristotle are descrip-
tive. Tat contrast in many ways mirrors the distinction between
ordinary language philosophy and those formalist attempts that
only mar whats well .
Cliford Brown
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Austin , J. L. 1962 . How to Do Tings with Words . Oxford : Clarendon.
Brown , C. 2006 . Peter Strawson. Montreal : McGill-Queens University
Press.
Quine , W. V. O. 1972 . Methods of Logic. New York: Holt, Rinehart and
Winston.
Russell , B. 1905 . On denoting . Mind 14.4 : 47993 .
Strawson , P. 1950 . On referring . Mind 59: 2152.
. 1952 . Introduction to Logical Teory . London: Methuen.
. 1954 . Wittgensteins Philosophical Investigations . Mind , n.s.,
63 .249: 7099.
. 1974 . Subjects and Predicates in Logic and Grammar. London:
Methuen.
ORIGINS OF LANGUAGE
Tis term and language evolution are sometimes used inter-
changeably. Here, origins of language will mean the earliest
emergence of a system structurally distinct from the communi-
cation systems of other animals and having at least some of the
attributes of human language; the term will not include further
evolutionary developments leading to the emergence of fully
modern language or the subsequent diversifcation of human
languages (see historical linguistics ). Although the topic
is one that has engaged the human imagination throughout his-
tory, an orgy of armchair speculation following Darwin caused
it to fall into disrepute. However, now that advances in a vari-
ety of sciences have made possible more informed (if still inevi-
tably speculative) approaches, the attention devoted to it has
increased annually, with perhaps an overly ebullient prolifera-
tion of theories.
Te issues may be more sharply defned by considering sepa-
rately three major questions to which the topic gives rise:
(1) Was language directly selected for, or an emergent prod-
uct of other faculties?
(2) If it was selected for, what pressure(s) selected for it?
(3) What form did its earliest emergence take?
Other issues involve the timing of language emergence, the
modality it originally employed, and whether or not language
evolved directly from prior means of animal communication.
Origins of Language
578
A synthetic protolanguage faces many dif culties, however
(Tallerman 2007 ). Whereas a compositional protolanguage
enables basic functions of language, such as creating new infor-
mation, asking questions, and negating statements, a synthetic
protolanguage allows for none of these. Predication and dis-
placement are equally impossible. Other problems arise at the
stage of reanalysis into a compositional system. For instance,
unless a given holophrase is equivalent to just one sentence in a
compositional language, no two people would necessarily agree
as to the meanings of its analyzed segments; yet if such equiva-
lence exists, a compositional language must already exist, at least
mind-internally so why is a holophrastic stage necessary? Te
precise nature of protolanguage has been, and will doubtless
continue to be, hotly debated, a debate to which experimental
evidence will hopefully contribute (Bowie 2006 ) .
Other Issues
A further controversy revolves around whether language was orig-
inally spoken or signed. Given that sign languages develop as
naturally among the deaf as do spoken ones among the hearing,
and that the hands of our closest primate relatives are more agile
and under more volitional control than their vocal organs, the
notion of a signed protolanguage is not unreasonable and has
been vigorously defended (Corballis 2002 ) . However, even if the
original modality could be determined (and for all we know, pro-
tolanguage could originally have mixed signs and vocalizations
indiscriminately), this would not answer the questions discussed
here or tell us how language came to acquire the properties that
distinguish it from other modes of communication.
Another unresolved issue concerns the timing of emergence.
None of the evidence from the fossil record is unambiguous.
Endocasts of Homo habilis suggest a developed brocas area,
and this has been taken to indicate an early (~2.5 million years
ago) beginning for language. But since, even today, Brocas area
subserves both linguistic and nonlinguistic functions, we can-
not know what functions it performed in antecedent species.
Symbolic artifacts are sometimes used to date language origins,
but while these indicate that language already existed, they can-
not tell how long before their appearance it began. Absent reli-
able evidence, estimates of when language originated tend to
be determined by researchers positions on other issues. For
instance, those who believe that language emerged abruptly
more or less in its present state favor a recent date coinciden-
tal with the emergence of anatomically modern humans (~140
thousand years ago), or even later. Conversely, those who take
an adaptationist approach argue for a much earlier date, any-
thing up to a few million years ago. Te origin of language is
probably associated with some speciation event, but this issue,
like most others, is unlikely to be resolved without new sources
of evidence.
Te question of continuity with prelinguistic systems is
somewhat clearer. Tat language evolved from some prior com-
municative system was, to Darwin, an article of faith, and some
subsequent authors have assumed that a commitment to grad-
ual evolution entails such continuity, discounting the possible
capacity of mutations, changes in function, and interactions
between diferent faculties to produce evolutionary novelties.
But the only plausible continuist scenario is the holophrastic ,
language was limited to a handful of symbols with which it
would have been impossible to express any socially signifcant
meaning. What, in such a situation, would have reinforced lan-
guage use?
Advocates of some form of social adaptive pressure whether
for gossip (see grooming, gossip, and language ), sexual
display, or social manipulation have so far failed to address
such problems adequately. An alternative proposal is that some
primitive form of language developed for exchanging informa-
tion about food sources among small groups of extractive for-
agers (Bickerton 2002 ). Carcasses of megafauna, in particular,
would have required the rapid recruitment of signifcant num-
bers for ef cient exploitation. Nobody doubts that language,
once it had emerged, would have been used for a variety of social
functions; such functions, in turn, would have expanded lan-
guage. Te real, and still unanswered, question is exactly what
led to its initial emergence.
Te issue is rendered still more problematic by the fact that
words are cheap tokens (Zahavi 1975 ). Since they take so little
efort to produce, and since primate species constantly engage in
deception, why would anyone have believed them, and if no one
believed them, who would have persevered in their use?
Initial Structure
While some (as noted) believe that language has always pos-
sessed its present structure, most researchers would probably
agree that some simpler form developed frst a stage generally
termed protolanguage (Bickerton 1990 ) and subsequently grew
more complex. Until recently, it was assumed that protolanguage ,
like early-stage pidgins , consisted of a small quantity of units
(roughly equivalent in semantic coverage to modern words)
that could be concatenated, without any consistent grammatical
structure, to form brief propositions ; in other words, proto-
language was compositional . Tis view is now challenged by
the proposal that protolanguage was synthetic, with holophrastic
units (like the units of animal communication systems) roughly
the semantic equivalents of complete propositions and not divis-
ible into smaller meaningful units the whole thing means the
whole thing (Wray 2002 , 118) .
Defenders of a synthetic system note that (in contrast with
a compositional system) there would nowhere be any break in
continuity between language and the prelinguistic communica-
tion system of hominids (assumed to be similar to those of other
primates; see primate vocalizations ), which it would at frst
resemble except for productivity (holophrastic units could be
multiplied indefnitely). At a subsequent stage, chance phonetic
similarities between portions of holophrases would cause the lat-
ter to be reanalyzed into wordlike segments; these could then be
recombined to form a modern, compositional language.
It is claimed that a synthetic protolanguage would be less sub-
ject to ambiguities than a compositional one and would be bet-
ter adapted for manipulation of other group members. Support
has come from computational linguists , many of whose
simulations of language evolution begin with units that repre-
sent propositional rather than lexical units (Briscoe 2002 ). Tose
who, following Darwin and Otto Jespersen, assume a common
origin for language and music are more or less obliged to adopt
some form of the synthetic hypothesis .
Origins of Language
579
the irregular form, and that experience eventually settles in favor
of the irregular form.
In a crucial 1986 paper, however, D. E. Rumelhart and J. L.
McClelland showed that newly developed connectionist
networks could simulate the rise and decline of overregulariza-
tions in childrens speech without the use of general rules or syn-
tactic symbols like verb . Briefy stated, connectionist networks
hypothesize connective paths between the constituent features
of present and past forms. Trough feedback about correctness,
eventually the right feature-to-feature connections get sorted
out. With skillful design, such networks can simulate the tem-
poral courses through which children pass without ever using
any general rule statement at all. In such a model, there is an
implicit competition between irregular and past forms, but the
competition is really among connections of features. Tere is no
general rule, no general reference to verb as a category. Irregular
and regular forms are produced by a single overall network pro-
cess. So these are called single process models versus dual process
(general rule vs. individual lexical entry) .
Rumelhart and McClellands paper instigated a series of
simulations, arguments, criticisms, and new simulations that
continue to this day (e.g., McClelland and Patterson 2002 ;
Plunkett and Marchman 1991 , 1993 ; Pinker 1999 ). Perhaps the
most prominent empirical data were introduced by G. Marcus
and colleagues ( 1992 ), who analyzed longitudinal studies of four
children and cross-sectional studies of hundreds more. Tey
argued that in any competition account, one would expect that
overregularizations would originally occur at a high rate before
experience wore them down. But in their analysis of the longitu-
dinal and cross-sectional subjects, they found that overall pre-
school year rates seemed very low, around .04 to .06 (or .02 to .10,
depending on the method).
Tis means, they argue, that in actuality, children probably
know the irregular form is correct as soon as they learn it. Tis
knowledge is available because an innate general heuristic
called blocking tells children that if two forms are possible but
only one is heard, choose the heard one. Te actually heard
irregular form thus has an innate heuristic preferred status.
Overregularizations only occur if the child does not know the
irregular form, or if the child temporarily cannot remember the
irregular form and the regular rule intrudes itself. Such retrieval
errors are posited to be inherently rare, for some unstated
reason .
Te blocking hypothesis requires a general reference to alter-
native rules and, in practice, to regular rule versus irregular indi-
vidual lexical patterns. So blocking contradicts connectionist
formulations in many ways. If there is no competition, connec-
tionist models cannot be correct, as they presuppose competi-
tion of some sort. If blocking in particular is correct, it requires
statement at general symbolic and rule levels, and so network
formulations are inadequate.
M. Maratsos ( 2000 ), however, has argued against the empir-
ical conclusions of Marcus and colleagues ( 1992 ). Using sam-
pling arguments, he notes that for frequent irregular verbs,
which dominate overall tabulations, even in a competition
model children would probably hear hundreds of correct inputs
within a week or few weeks after the competition started, and
so overregularizations would fall to near zero very quickly; the
synthetic model of Wray, discussed previously. If objections to
this are overcome, the case for continuity could be maintained;
otherwise, the diferences between language or even protolan-
guage and any nonlinguistic system suggest a sharp discontinu-
ity between the two .
Derek Bickerton
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bickerton, Derek. 1990. Language and Species . Chicago : University of
Chicago Press.
. 2002. Foraging versus social intelligence in the evolution of pro-
tolanguage. In Te Transition to Language, ed. Alison Wray , 207 26.
Oxford: Oxford University Press .
Bowie, Jill. 2006. Te evolution of meaningful combinatoriality . Paper
presented at the Sixth International Conference on the Evolution of
Language, Rome.
Briscoe, Ted , ed. 2002. Linguistic Evolution through Language
Acquisition: Formal and Computational Models. Cambridge : Cambridge
University Press.
Byrne, Frank, and Andrew Whiten. 1992. Cognitive evolution in pri-
mates. Man 27 : 60927 .
Corballis, Michael. 2002. From Hand to Mouth: Te Origins of Language .
Princeton, NJ : Princeton University Press.
Donald, Merlin. 1991. Origins of the Modern Mind . Cambridge : Harvard
University Press.
Hauser, Marc, Noam Chomsky, and Tecumseh Fitch. 2002. Te fac-
ulty of language: What is it, who has it, and how did it evolve ? Science
298 : 156979.
Humphrey, Nicholas K. 1976. Te social function of intellect. In
Growing Points in Ethology, ed . P. P. G. Bateson and R. A. Hinde ,
30317. Cambridge : Cambridge University Press .
Jenkins, Lyle. 2000. Biolinguistics: Exploring the Biology of Language .
Cambridge : Cambridge University Press.
Kirby , Simon, and Morton H. Christiansen , eds. 2003 . Language Evolution .
Oxford : Oxford University Press . A collection of position papers by
leading scholars in the feld.
Pinker, Steven, and Paul Bloom. 1990. Natural language and natural
selection . Behavioral and Brain Sciences 13 : 70726.
Tallerman, Maggie. 2007. Did our ancestors speak a holistic protolan-
guage ? Lingua 117 : 579604.
Wray, Alison. 2002. Dual processing in protolanguage: Performance
without competence. In Te Transition to Language, ed. A. Wray ,
113 37. Oxford : Oxford University Press.
Zahavi, Amotz. 1975. Mate selection a selection for a handicap .
Journal of Teoretical Biology 53 : 20514.
OVERREGULARIZATIONS
Overregularizations like runned and mans have played a major
role in the language development literature for more than 40 years
(see also childrens grammatical errors; morphology,
acquisition of; syntax, acquisition of ). Once brought
into focus, overregularizations comprised prime examples of
the way in which childrens use of general grammatical rule
knowledge (the regular past tense rule of adding - ed ) could pro-
ductively overwhelm the word-specifc knowledge they gained
form actual input, a prime example of using rules to go beyond
the input . Less attention was paid to the way in which children
would get rid of overregularizations, but results indicated that
there is a verb-by-verb competition between the regular rule and
Overregularizations
580
Rumelhart , D. E. , and J. L. McClelland . 1986 . On learning the past tenses
of English verbs. In Parallel Distributed Processing: Explorations in
the Microstructure of Cognition. Vol. 2: Psychological and Biological
Models. Ed. J. L. McClelland , D. E. Rumelhart , and the PDP Research
Group , 21671. Cambridge, MA : Bradford Books/MIT Press .
P
PARABLE
Standard defnitions, such as the one given in the Oxford English
Dictionary , conceive of parable as a literary term; it is said to be
the expression of one story through another. Literary historians
have modifed this conception by placing limits on the kind of
story that counts as parable, attempting to distinguish it from, for
example, fable or allegory.
Tere are, however, even among literary scholars, some who
see parable as a much larger phenomenon, belonging not merely
to expression and not exclusively to historical genres but, rather,
as C. S. Lewis ( 1936 , 44) observed, to mind in general . (See also
Louis MacNeices discussion of literary critical perspectives on
parable in MacNiece [1963] 1965 , 5.)
For the language sciences, parable is not only, or even chiefy,
a kind of story; it is not an expression at all but, rather, a men-
tal faculty that allows the human mind to integrate two concep-
tual stories or narratives into a third story, thereby creating a
conceptual blending network that has emergent meaning .
Straight history, or the observation of human interaction,
often can serve as the material for such parabolic blending. For
example, Sun Tzus Te Art of War treats 13 aspects of warfare.
It has been studied in the West by military strategists since the
eighteenth century. Written in the sixth century b.c. in China, it
precedes by a couple of millennia the origin of modern business
management. But in the 1980s, it underwent extensive para-
bolic rendering in numerous books and articles for the purpose
of ofering guidance to twentieth-century graduate students of
business and investment on how to conduct their professional
lives.
Parable frequently blends two stories that have strong con-
ficts in their content. It is a scientifc riddle why human beings
should be able to activate two conficting stories simultaneously,
given the evident risks of mental confusion, distraction, and
error. Yet, uniquely among species, human beings can evidently
not only activate fundamentally conficting stories simultane-
ously and construct connections between them but also blend
them to create emergent meaning. Tis ability to blend two con-
ceptual arrays with strong conficts in their framing structure is
central to higher-order human cognition and is a hallmark of the
cognitively modern human mind. It is known as double-scope
blending (Fauconnier and Turner 2002 ) .
Consider a parable from the Fourth Gospel. In John 10:1118,
Jesus presents Himself as the good shepherd, who lays down
His life for the sheep, in contrast to the hired hand, who does
not care for the sheep and fees in the face of the wolf. He says
the Father loves Him because He lays down his life and that no
one takes it from Him. Rather, He has the power to lay it down
result would be an overall rate of near zero in a sample of two to
three years. Our samples are so small (usually an hour a week)
that they would fail to catch these occurrences. Only less fre-
quent verbs, discounted by Marcus and colleagues, might show
evidence of strong overregularization. In fact, that they do was
shown for R. Browns two low-overregularizing subjects Adam
and Sarah (see, e.g., 1973 ). For Adam, for example, the average
overregularization rate was a strong 55 percent for his 21 lower-
frequency verbs. Te same rate was found even in samples after
the child frst produced the correct irregular form of a verb.
Arguments from sampling considerations indicated that such
overregularizations were still persisting after tens or even hun-
dreds of uses. Recent work from a more intensively recorded
subject, Peter (Maslen et al. 2004 ), has strongly supported these
analyses and extended them to noun plurals. Tese data indi-
cate that overregularizations do often appear frequently after
the irregular past is known, contrary to blocking. Our samples
restrictions just make it dif cult to capture them for the more
frequent irregular verbs whose numbers dominate overall
rates.
Suppose these analyses do indeed indicate that the low-rate
blocking account is incorrect. Do they also show that the connec-
tionist account is therefore correct? Actually, they only indicate
that a competition process of some sort is involved. As noted,
older rule-based models also assumed a competition between
regular rule and individual entry. Te current association of
competition with connectionism and with non-rule models thus
refects current disputes, not the basic analytic problem. Te
confict between connectionist and rule-based approaches will
thus have to be resolved ultimately, if it can be, using other data
and arguments .
Michael Maratsos
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brown, R. 1973. A First Language: Te Early Stages . Cambridge : Harvard
University Press.
Maratsos , M. 2000 . More overregularizations after all: New data and dis-
cussion on Marcus, Pinker, Ullman, Hollander, Rosen & Xu . Journal
of Child Language 27 : 183212 .
Marchman , V. , and E. Bates . 1994 . Continuity in lexical and morpho-
logical development: A test of the critical mass hypothesis . Journal of
Child Language 21 : 31718 .
Marcus , G., S. Pinker, M. Ullman, M. Hollander, T. Rosen , and F. Xu .
1992 . Overregularizations in language acquisition . Monographs of
the Society for Research in Child Development 57, serial no. 228.
Maslen , R. J., A. L. Teakston, E. V. Lieven , and M. Tomasello . 2004 .
A dense corpus study of past tense and plural overregulariza-
tion in English . Journal of Speech, Language, and Hearing Research
47: 131933.
McClelland , James , and Karalyn Patterson . 2002 . Rules or connections
in past-tense infection: What does the evidence rule out? Trends in
Cognitive Science 6 : 46572 .
Pinker , S . 1999 . Words and Rules: Te Ingredients of Language . New
York : Basic Books.
Plunkett , K., and B. Marchman . 1991 . U-shaped learning and frequency
efects in a multi-layered perceptron: Implications for language acqui-
sition . Cognitio n 28 : 73193.
. 1993 . From rote learning to system building: Acquiring verb mor-
phology in children and connectionist nets . Cognition 48 : 3559 .
Overregularizations Parable
581
. 2001 . Cognitive Dimensions of Social Science: Te Way We Tink
About Politics, Economics, Law, and Society . New York : Oxford
University Press .
PARALANGUAGE
Nuances, connotations, and innuendos, which are integral char-
acteristics of verbal communication, are given the vague term
paralanguage . Tese meanings arise from sources both within
and outside of standard linguistic structure. Linguistic elements
words, word order, semantics , grammar can be utilized
for paralinguistic communication. Tese combine with variations
of speech melody in ways that often defy structural description.
Paralanguage (as the term implies) both draws on and lies over
the known and describable ortholinguistic levels of phonetics,
phonology, morphology and lexicon, syntax , and seman-
tics. All of these elements can be harnessed for paralinguistic
communication, as is well known from baby talk, from connota-
tional meaning diferences in terms such as skinny , slim , slender ,
and from word-order choices such as Herman Melvilles Tat
inscrutable thing is chiefy what I hate. In addition, emotion,
attitude, intention, mood, psychological state, personality,
and personal identity can be communicated without referring
to words. Because of the power of the intonational contribu-
tion to paralanguage, the notion of two channels in the speech
signal has been invoked, but their intimate interplay has been
emphasized (Bolinger 1964 ). Words can communicate emotions,
but when prosody does so using a diferent channel, the paralin-
guistic intent overrides the ortholinguistic content, as in Im not
angry! spoken with increased pitch , amplitude, and rate.
Much of paralanguage is carried over longer stretches of
utterance than the phonetic element or the single word. Lexical
and syntactic choices may interact with intonational features
with a cumulative efect. Formulaic and nonliteral expressions
may be called into play. It has come to my attention that your
stonewalling is holding up the works contains conventional and
metaphoric utterances that build to a message more fraught
with paralinguistic content than Ive learned that your hesita-
tion is contributing to a delay. Although subtle contrasts can be
conveyed on short utterances (see Nine ways of saying yes in
Crystal 1995 ), paralanguage prefers a larger canvas. Repetition of
words (Shakespeares a little, little grave) may have a powerful
paralinguistic efect. Movement from low to high pitch across an
intonational unit displays surprise or amazement; temporal units
are stretched to express sadness or disappointment; increased
intensity signals aggression or thematic emphasis; voice quality
becomes creaky to communicate victimization or breathy to
signal excitement.
Prosody, a major vehicle of paralanguage, can be decom-
posed into measurable elements: timing, pitch, amplitude, and
voice quality. Tese measures combine into complex patterns,
such that associating acoustic cues with paralinguistic meanings
is far from straightforward. John didnt drive the car can be
intoned with sadness, happiness, fear, or disgust, and may enfold
attitudes such as incredulity, relief, perplexity, or amusement.
Contradiction or denial, and conversational presumptions, such
as sincerity and truthfulness, are carried by phrasal intonation.
Take a common paralinguistic trope, sarcasm (see irony ), in
and take it up again. Te clash between the story of the shep-
herd and the blend Jesus proposes is astonishing. It is quite
implausible that a shepherd would choose to die defending the
sheep, because then the sheep would be without a defender.
Yet this consequence is not projected to the blend: Te actual
shepherd cannot return after being killed to look out for the
fock, but in Jesuss blend, He can. Te emergent struc-
ture in the blend is crucial: Jesuss narrative blends dying
with physical manipulation of an object. (Physical manipula-
tion is at the root of human understanding. See Chapter 4 of
Turner 1996 , Actors Are Movers and Manipulators.) In the
story of manipulation, we can lay down an object and pick it
back up. Blending manipulation of a physical object with the
state of being alive or dead, Jesus achieves the remarkable
ability of self-revival .
As discussed in Chapter 4 (Analogy) of Cognitive Dimensions
of Social Science (Turner 2001 ), almost all the mental achieve-
ments analyzed by analogy theorists as analogy involve con-
siderable unrecognized blending. In general, analogy involves
dynamic forging of mental spaces , construction of connec-
tions between them, and blending of the mental spaces to create
a conceptual integration network of spreading coherence, whose
fnal version contains a set of what are recognized, after the fact
in the rearview mirror, as systematic, even obvious analogical
connections. But those analogical connections are more often
the outcome of conceptual blending than its preconditions. Put
diferently, what is commonly discussed as analogy manifests
the faculty for parable .
It is also important to recognize that a parable is not, in gen-
eral, a conceptual metaphor for understanding one con-
ceptual domain in terms of another. Consider 2 Samuel 12. Te
prophet Nathan creates an elaborate blend in which a rich man
is blended with King David, a poor man is blended with Uriah
the Hittite, Uriahs wife Bathsheba is blended with a favored ewe
lamb, and there is a traveler who comes to dinner. Te point of
the complex blend is that David has done wrong. Te source
and target are complicated, drawing on many conceptual
domains, and the principal connection is that in both of them,
one man abuses another and deserves punishment. No general
conceptual metaphor provides this set of cross-space connec-
tions. Most of them are not metaphoric .
Parable as a form of literary expression might be of interest
to historians, anthropologists, and critics. But parable as a spe-
cies-specifc mental faculty that can activate, connect, and blend
sharply conficting stories to produce new emergent meaning is
a far larger and more fundamental topic, posing one of the cen-
tral riddles of the cognitive and language sciences .
Mark Turner
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Fauconnier , Gilles , and Mark Turner . 2002 . Te Way We Tink: Conceptual
Blending and the Minds Hidden Complexities . New York : Basic
Books.
Lewis , C. S . 1936 . Te Allegory of Love . Oxford : Oxford University Press .
MacNeice , Louis . [1963] 1965. Te Varieties of Parable [Te Clark
Lectures]. Cambridge : Cambridge University Press .
Turner , Mark . 1996 . Te Literary Mind: Te Origins of Tought and
Language . New York : Oxford University Press .
Parable Paralanguage
582
PARAMETERS
Te term parameter is used in linguistics on analogy with its
usage in mathematics and engineering. In mathematics, the
parameters of a function are those aspects of the function that
are held constant when defning a particular function, but which
can vary in a larger context so as to characterize a family of sim-
ilar functions. For example, the function for a line in analytic
geometry is f(x) = mx + b, with x the variable and m and b param-
eters of the function (the slope and the y -intercept). In the def-
nition of any one line, the parameters m and b are held constant,
while the value of x varies, giving the diferent points on the same
line. In a broader context, however, the parameters m and b can
vary so as to defne a family of similar functions: the set of all
lines. In the same way, parameters in linguistics are properties
of a grammatical system that are held constant when character-
izing one particular human language, but which are allowed to
take diferent values in a broader context so as to characterize
a whole family of possible human languages. Te idea that the
observed variation in human languages can be understood as
the fxing of certain parameters within an otherwise innate and
invariant system of principles ( universal grammar ) is most
commonly associated with the Chomskyan approach to formal
generative linguistics (see generative grammar) . As a result,
this approach is sometimes called the principles and param-
eters theory . Te idea is, however, a very general one, and it
can also be used in the context of other views about the nature of
the human language faculty .
Tis notion of linguistic parameters was introduced into lin-
guistic theory by Chomsky ( 1981 ) and Rizzi ( 1982 ), during the
government and binding period. Te paradigmatic case
was the pro-drop parameter (or null subject parameter ). It was
observed that languages like Spanish and Italian difer from
French and English in several ways that appear to be interre-
lated. First, Spanish and Italian allow the subject pronoun of a
fnite clause to be omitted, whereas French and English do not:
(1) a. Verr. (Italian: He/she will come)
b. *Came. (English)
Second, Spanish and Italian allow the subject to come after the
verb as well as before it, whereas French and English generally
do not:
(2) a. Verr Gianni (Italian, Will-come Gianni)
b. *Came John (English)
Tird, the subject of an embedded sentence in Spanish and
Italian can be moved to the beginning of the sentence as a whole,
even when there is an overt complementizer, whereas in French
and English some sort of accommodation is needed in sentences
like these:
(3) a. Chi credi che verr? (Italian)
Who you-think that will-come
b. *Who do you think that came? (English)
Although these are clearly three distinct properties of the lan-
guages in question, they have a common theme: Informally put,
French and English require that there be an overt noun phrase
in the canonical subject position immediately before the fnite
Tat was a good efort. We know sarcasm when we hear it, but
exactly what in the signal conveys that the speaker is intending
to communicate the opposite of the usual lexical meanings is dif-
fcult to specify. In one version, the sarcastic utterance utilizes
higher pitch and greater amplitude on the frst word followed
by falling intonation, pharyngeal voice quality, tensed vocal
tract, and spread lips. While morphological, lexical, and syntac-
tic meanings can be structurally analyzed using units, features,
and rules, paralinguistic meanings constitute a brew of unstable,
feeting, and subjective qualities. Tese paralinguistic qualities
shade into one another, and they impinge on purely linguistic
uses of prosodic contrasts, as in question and statement intona-
tion. Te auditory-acoustic cues that comprise paralanguage are
graded, in that they are not perceptually allocated by the listener
into discrete, contrastive categories as are the acoustic signals
for phonetic and lexical elements. Using deft combinations and
placements of prosodic cues, a speaker can communicate more
or less fear, gradations of perplexity, and degrees of denial .
Te development of the pragmatics of communication ,
a branch of linguistics that studies language use in conversation
(see conversation analysis ), jokes (see verbal humor ),
and storytelling, has advanced understanding of paralanguage.
Communicative elements such as turn-taking, inference, and
theme (topic of the discourse), and how they are signaled by the
speaker and comprehended by the listener, are investigated. Te
felds of prosody and pragmatics have provided another valuable
impetus for the productive study of paralanguage: investigation
of the communicative competence of right hemisphere lan-
guage processing in humans. While it has long been known
that the left hemisphere modulates language processing,
studies of pragmatics and prosody indicate involvement of the
right hemisphere in processing emotions and attitudes, infer-
ence and theme. Te notion of two channels, ortholinguistic and
paralinguistic, is supported by the model that allocates process-
ing to left and right hemispheres, respectively. Paralinguistic
nuances are intimately woven into the propositional message, so
much so that synthesized speech is often judged as unpleasant.
A goal of speech synthesis is to produce more natural-sounding
speech, which means infusing paralanguage, a challenging but
worthy goal .
Diana Van Lancker Sidtis
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bolinger , Dwight . 1964 . Around the edge of language: Intonation. In
Intonation , ed. D. Bolinger . Harmondsworth, UK: Penguin Books.
Crystal , David . 1995 . Nine ways of saying yes. Te Cambridge
Encyclopedia of the English Language, ed. David Crystal, 248.
Cambridge: Cambridge University Press.
Kreiman , Jody , Diana Van Lancker Sidtis , and Bruce Gerratt . 2005 .
Perception of voice quality. In Handbook of Speech Perception , ed.
David Pisoni and Robert Remez , 33862. Maldon, MA : Blackwell .
Van Lancker Sidtis , Diana . 2007 . Te relation of human language to
human emotion. In Handbook of the Neuroscience of Language ,
ed. Brigitte Stemmer and Harry Whitaker . San Diego, CA : Academic
Press .
Williams , C ., and K. Stevens . 1972 . Emotions and speech: Some
acoustical correlates . Journal of the Acoustical Society of America
52. 4B: 1238 50.
Paralanguage Parameters
583
parameters that are intended to account for the large-scale dif-
ferences among the major classes of languages discovered by
typology . Te head directionality parameter is a parameter of
this sort. Another early example was Ken Hales ( 1983 ) noncon-
fgurationality parameter , which was designed to explain why
Australian languages like Warlpiri tolerate free word order and
discontinuous phrases, whereas languages like English do not.
Similarly, parameters have been proposed to capture the difer-
ence between ergative languages (like Basque and Eskimo), in
which the object of a transitive clause is treated in some respects
like the subject of an intransitive clause, and accusative lan-
guages (like English and most Indo-European languages), in
which all subjects are treated similarly. Tese proposals range
from radical diferences in how syntactic structure is initially
constructed (Marantz 1984 ) to relatively minor diferences in
how case and agreement morphology are assigned in a simple
sentence (Bittner and Hale 1996 ). Mark Baker ( 1996 ) proposes
a polysynthesis parameter that attempts to give a unifed charac-
terization of the diference between many native American lan-
guages, in which a large part of the expressive burden is placed
on verbal morphology, and languages like English, in which the
primary expressive burden is borne by syntactic combination.
Taken together, some set of parameters such as these might
characterize the major linguistic types we observe .
Other parameters operate on a smaller scale, defning the dif-
ferences between historically-related languages or dialects .
Te pro-drop parameter was a parameter of this sort, distinguish-
ing French from Italian. Another example is the parameter that
determines whether the subject of a clause moves from its origi-
nal position inside the verb phrase to the highest position in the
clause or not; this accounts for the diference between English,
which has subject-fnite verb-object word order, and Celtic lan-
guages like Welsh, which have fnite verb-subject-object word
order (Koopman and Sportiche 1991 ). Jean-Yves Pollock ( 1989 )
argues that there is a parameter that says that verbs move to a
higher position in French than they do in English; this accounts
for a cluster of subtle word-order diferences having to do with the
placement of verbs, negation, and adverbs in the two languages
(e.g., John kisses often Mary is normal French but bad English).
A third example is Jonathan Bobaljik and Dianne Jonass ( 1996 )
proposal that some Germanic languages have an extra posi-
tion available for subjects that other Germanic languages dont
have; this makes sentences like Tere have some trolls eaten
Christmas pudding possible in some Germanic languages but
not others, among other things. (See Baker 2001 for a general
overview of these parameters and several others.)
In the early days of parametric theory, it was thought that
virtually any syntactic principle could be parametrized, and
parameters were proposed that were relevant not only to X-bar
theory but also to movement , the theory of binding , and even
the projection principle. On that view, there would be a mod-
est number of parameters (dozens or perhaps hundreds), each
of which would have a relatively large impact on the language
generated. But this view has been questioned in more recent
work. Hagit Borer ( 1984 ) proposed almost immediately that the
syntactic principles themselves are invariant, and what is param-
eterized is the features associated with individual lexical items.
Rather than saying that the syntax of French is diferent from
verb, whereas Spanish and Italian do not. Tis diference in the
syntax of subjects was also related to a morphological dif-
ference: Te agreement morphology on the fnite verb is rich
enough to uniquely identify which pronoun would be in the
subject position in Spanish and Italian, whereas in French and
English it is not. Te universal syntactic condition, then, is that
fnite clauses require subjects (the extended projection prin-
ciple ); the parameter concerns exactly what kind of subject is
necessary to fulfll this condition. In Italian and Spanish, the rich
agreement on the verb means that null or displaced subjects are
permissible because (roughly) much of the information con-
cerning the sort of subject it was is locally available on the fnite
verb. In French and English, the agreement on the verb is of little
help, and so an overt subject in the canonical subject position
is required. A parameter, then, is a way of attributing a unifed
theoretical account of the systematic diferences that distinguish
one class of languages from another.
While the pro-drop parameter was the frst important param-
eter to be proposed, it is by now not considered the best case. A
look at a wider range of languages both nonstandard dialects
of the Romance languages and languages from other families
quickly showed that the properties in (1)(3) do not correlate
with one another as closely as was thought (Jaeggli and Safr
1989 ). Tis implies that the pro-drop parameter as it was origin-
ally conceived is either false or highly oversimplifed .
Tat does not mean that the idea of a parameter was ill-con-
ceived, however. Te current paradigmatic example is what is
sometimes called the head directionality parameter (terminolo-
gies vary). Tis can be stated as an open factor in the principles
of phrase structure (see x-bar theory ). Roughly put,
when a word -level category x merges with a phrase Y to create
a phrase of type X, there are two ways that the elements can be
ordered: Te order can be X-Y within XP, or it can be Y-X. Setting
the parameter in the frst way gives head-initial languages like
English, in which complementizers come before embedded
clauses, tense particles come before verb phrases, verbs come
before their objects, prepositions come before their objects, and
so on:
(4) John will think that Mary showed a picture to Sue.
Setting the parameter in the second way gives head-fnal lan-
guages like Japanese, in which complementizers come after
embedded clauses, tense particles come after verb phrases, verbs
come after their objects, prepositions come after their objects,
and so on:
(5) Taroo-ga Hiro-ga Hanako-ni syasin-o miseta to omotte iru .
Taro SUBJ Hiro SUBJ Hanako to picture OBJ show that think-
ing be
Taro is thinking that Hiro showed a picture to Hanako.
In this way, a parametric theory can account for many of the
most robust Greenbergian universals (Greenberg 1963 ; Dryer
1992 ) concerning word order in an elegant way. Tese two
very common and stable language types fall out of a simple and
unity choice that is made in the precise formulation of a univer-
sal principle of language .
Parameters vary widely in the range and scope of the efects
that they are supposed to capture. Some theorists have proposed
Parameters
584
Jaeggli , Osvaldo , and Kenneth Safr . 1989 . Te null subject parameter
and parametric theory. In Te Null Subject Parameter , ed. Osvaldo
Jaeggli and Kenneth Safr, 144. Dordrecht, the Netherlands : Kluwer .
Kayne , Richard . 2005 . Some notes on comparative syntax, with spe-
cial reference to English and French. In Te Oxford Handbook of
Comparative Syntax , ed. Guglielmo Cinque and Richard Kayne, 369.
New York : Oxford University Press . A detailed discussion of general
considerations and very small-scale parameters.
Koopman , Hilda , and Dominique Sportiche . 1991 . Te position of sub-
jects. Lingua 85 : 211 58.
Marantz , Alec . 1984 . On the Nature of Grammatical Relations . Cambridge,
MA : MIT Press .
Pollock , Jean-Yves . 1989 . Verb movement, universal grammar, and the
structure of IP . Linguistic Inquiry 20 : 365 424.
Rizzi , Luigi . 1982 . Issues in Italian Syntax . Dordrecht, the
Netherlands : Foris .
PARIETAL LOBE
Anatomy
Te parietal lobe is situated superior to the occipital lobe and
posterior to the frontal lobe. More specifcally, it extends from the
central sulcus anteriorly, to the imaginary boundary of the pari-
etal- occipital fssure posteriorly, to the sylvian fssure ( peri-
sylvian cortex ) inferiorly. Te parietal lobe(s) can be further
subdivided into three main areas. Tese include: 1) the soma-
tosensory strip, also known as the postcentral gyrus (Brodmanns
area [BA] 1, 2, 3, 43), 2) the superior parietal lobule (BA 5), and 3)
inferior parietal lobule (includes BA 39-angular gyrus and 40-su-
pramarginal gyrus). Te latter two areas are separated by the
intraparietal sulcus (see Figure 1). Medially, the parietal lobe(s)
comprises the postcentral gyrus extension of the paracentral lob-
ule, the precuneus, and part of the cingulate gyrus (see Figure 2).
Physiology
Tere are two parietal lobes, one in each hemisphere, which are
divided functionally on the basis of dominance. Te dominant
lobe is typically the left one and the nondominant the right.
Tere are many diferent non-language functions performed
by the parietal lobe, for example, perception and localization of
touch, pressure, pain, and temperature on the opposite side of
the body, and visuospatial processing. Te variety of language-
related functions associated with the parietal lobe will be espe-
cially highlighted in the context of non-language functions.
Te dominant parietal lobe is involved primarily in integrat-
ing sensory information to create a particular perception. Te
inferior portion of this lobe, particularly the supramarginal gyrus
and angular gyrus, is involved in structuring information for
reading and writing (see writing and reading, neurobiol-
ogy of ) , performing mathematical calculations, and perceiving
objects normally.
Damage to the dominant lobe can result in apraxia (motor
planning defcit) aphasia (language disorder), agnosia (abnormal
perception of objects), and sensory impairment (e.g., touch, pain).
Lesions to the inferior portion of the dominant lobe involving the
angular gyrus can result in Gerstmanns syndrome, which is char-
acterized by leftright confusion, dif culty pointing to named
fngers (fnger agnosia), impaired writing ability (agraphia), and
inability to perform mathematical calculations (acalculia) .
the syntax of English in that verbs raise to the tense/inf node in
French, this view says that the lexicon of French is diferent from
the lexicon of English in that French has tenses that require the
verb to move into them, whereas English does not.
Borers view has the conceptual advantage that it largely
reduces the learning of syntax to the learning of individual lexical
items. It also suggests that there might be thousands of param-
eters, rather than dozens, because each distinct lexical item is
a possible locus of parametric variation (see especially Kayne
2005 ). Each individual parameter, however, will afect only a
relatively narrow part of the grammar since it is limited to those
structures in which a particular item appears. Tis view is com-
patible with the fragmentation of the pro-drop parameter, which
is now seen as a cluster of small-scale distinctions, each of which
can vary independently of the others, giving one the fexibility to
describe the various intermediate patterns found in the dialects
of southern France and northern Italy. As a result, Borers view
has been championed by Richard Kayne ( 2005 ) as the one that
is supported by his methodology of comparing closely related
dialects .
Baker ( 1996 , 2008), however, argues that there may also be
syntactic parameters in more or less the original sense, in addi-
tion to the fne-grained lexical parameters. Taken strictly, Borers
view does not really account for the unity of the head directional-
ity parameter. Even the smaller-scale parameters do not seem to
vary lexical item by lexical item. For example, it is not the case
that some tenses trigger verb-adverb-object order in French and
others do not; rather, all the diferent tenses trigger that order in
French, whereas none of the English tenses do. Perhaps, then, the
proper locus of much parameterization is neither the individual
lexical item nor the syntactic principle, but rather a natural class
of lexical items. How to state this and what its implications are
continue to be topics of discussion .
Mark C. Baker
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baker , Mark . 1996 . Te Polysynthesis Parameter . New York : Oxford
University Press .
. 2001 . Te Atoms of Language . New York : Basic Books. A broad
overview of the notion of a parameter in linguistic theory, written for
a general audience.
. 2008 . Te macroparameter in a microparametric world. In
Te Limits of Syntactic Variation , ed. Teresa Biberauer, 35174.
Amsterdam : John Benjamins . Provides an overview and argument for
large-scale parameters.
Bittner , Maria , and Kenneth Hale . 1996 . Ergativity: Toward a theory of a
heterogeneous class . Linguistic Inquiry 27 : 531 604.
Bobaljik , Jonathan , and Dianne Jonas . 1996 . Subject positions and the
roles of TP . Linguistic Inquiry 27 : 195 236.
Borer , Hagit . 1984 . Parametric Syntax: Case Studies in Semitic and
Romance Languages . Dordrecht, the Netherlands : Foris .
Chomsky , Noam . 1981 . Lectures on Government and Binding . Dordrecht,
the Netherlands: Foris . Te reference that started it all.
Dryer , Matthew . 1992 . Te Greenbergian word order correlations .
Language 68 : 81 138.
Greenberg , Joseph . 1963 . Universals of Language . Cambridge, MA : MIT
Press .
Hale , Kenneth . 1983 . Warlpiri and the grammar of nonconfgurational
languages . Natural Language and Linguistic Teory 1 : 5 49.
Parameters Parietal Lobe
585
of the research involving humans with brain damage and animal
studies using rhesus monkeys. Sir William Turner ( 1873 ) is con-
sidered the frst to describe in detail the intraparietal sulcus (BA
40). Before being scientifcally discredited, phrenologists pro-
posed that damage or disease to the parietal lobe(s) was a major
cause of melancholia (depression), and parietal eminence was
believed to relate to cautiousness (Hollander 1902 ). Due to the
wide variety of symptoms reported from brain damage studies,
the parietal lobes were accurately but vaguely thought to be a
general association area combining all the information from
various functions, specifcally visuospatial and attention; how-
ever, details as to how this function occurred physiologically
were lacking until modern times.
Josef Gerstmann ([ 1924 ] 1971) frst described fnger agnosia
in a patient with a left parietal stroke, and the efects of various
lesions on the parietal cortex were identifed and cataloged in
detail by John McFie and Oliver L. Zangwill ( 1960 ). Much of the
early parietal research was pioneered by scientists Macdonald
Critchley and later Juhani Hyvarinen in their respective works
Te Parietal Lobes (1953) and Te Parietal Cortex of Man and
Monkey (1982) .
A great deal of neurological investigation has been conducted
on rhesus monkeys, and there appears to be signifcant overlap
between the human and monkey parietal lobe in both function
Te nondominant parietal lobe, however, is involved in a dif-
ferent set of functions that are mostly non-language related. In
particular, this region is responsible for visuospatial functions as
it receives and integrates input from the visual system (occipital
lobe) to make sense of the spatial order of the world around us.
M. A. Eckert and colleagues ( 2005 ) found that Williams syndrome,
whose phenotype (visuospatial defcits) and genotype (deletion
on chromosome 7) are well characterized, is linked to superior
parietal impairment. Williams syndrome, thus, may provide a
valuable system for understanding parietal lobe function .
Damage to the right parietal lobe can result in a constellation
of defcits involving spatial and body relations. Bilateral lesions
may result in Balints syndrome , which afects both visual atten-
tion and motor skills. If both the parietal and temporal lobes are
damaged, memory impairments and personality changes may
result. Specifcally, if this damage occurs on the dominant (left)
side, it may result in verbal memory defcits and dif culty in the
retrieval of strings of numbers. If the damage is on the right side,
it will afect nonverbal memory functions and will signifcantly
impair personality .
History
For more than a century, the exact role of the parietal lobe has
been debated by neuroanatomists and psychologists, with much
Figure 1.
Figure 2.
Parietal Lobe
586
however, recent evidence (Wagner et al. 2005 ) suggests that the
parietal lobes may have a role to play in it as well. Te role of
declarative memory in language has been attributed to word
learning or vocabulary storage. Wagner and his colleagues sug-
gest three theories explaining the contributions of the parietal
lobe in episodic memory retrieval. Tey highlight that, indeed,
the parietal lobe does not play an independent role in this
retrieval; rather, it mediates the major pathways in which the
MTL subserves episodic memory .
In sum, a number of neurolinguistic positions have been
taken from available neuropsychological and brain-mapping
data (Stein 1989). Tese include the parietal lobe as a) a senso-
rimotor association area such that the posterior parietal cortex
(PPC) becomes a junction of somaesthetic and visual informa-
tion that interacts in a complex fashion, b) a sensorimotor inte-
gration area, which is very similar to the previous theory except
for the addition of an actual integral function, and c) a com-
mand apparatus that is actually able to initiate a motor activity
from the accumulated sensory information. Te authors propose
that although it is possible that the parietal lobe is involved in
some motor processes, it is more likely that the process is one
of maintenance than of initiation and is d) a region for directing
attention to stimuli of interest. Here, the PPC and the pathways
it receives are postulated to direct attentional focus to the tar-
get stimulus while coordinating and communicating with the
inferotemporal cortex. J. Stein advocates that the PPC does not
have a single narrow neurocognitive focus; nevertheless, it could
have a common underlying function that integrates its multifac-
eted involvement in cognitive as well as automatic linguistic and
extralinguistic processes.
Overall, the parietal lobe is crucial for several language func-
tions, most importantly, naming, semantic processing, and
phonological shaping of words, as well as reading and writing.
In addition, it mediates attention and memory, both essential at
diferent levels of language processing .
Yael Neumann , Hia Datta and Daniel P. Rubino
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baddeley , Alan , Susan Gathercole , and Costanza Papagno . 1998 . Te
phonological loop as a language learning device . Psychological Review
105 : 158 73.
Catani , Marco, Derek K. Jones , and H. Dominic . 2005 . Perisylvian lan-
guage networks of the human brain . Annals of Neurology 57 : 8 16.
Critchley , Macdonald . 1953 . Te Parietal Lobes . London : Edward Arnold .
Eckert , M. A ., B. S. Hu , S. Eliez , U. Bellugi , A. Galaburda , J. Korenberg , D.
Mills , and A. L. Reiss . 2005 . Evidence for superior parietal impairment
in Williams syndrome . Neurology 64 : 152 3.
Gerstmann , Josef . [1924] 1971. Fingeragnosie: Eine umschriebene
Strung der Orientierung am eigenen Krper. Wiener Klinische
Wochenschrift 37 : 101012. Trans. in Archives of Neurology
24 : 4756.
Geschwind , Norman . 1965. Disconnection syndromes in animals and
man . Brain 88 : 237 94.
Hickok , Gregory . 2000 . Speech perception, conduction aphasia, and
the functional neuroanatomy of language. In Language and the
Brain: Representation and Processing , ed. Y. Grodzinsky , L. Shapiro,
and D. Swinney, 87104. San Diego, CA : Academic Press
Hollander , Bernard . 1902 . Scientifc Phrenology . London : Grant
Richards .
and form, though it is noteworthy that diferences have been
identifed, such as larger parietal cortex, asymmetry of the lobes,
and more neural subdivisions in humans (Kolb and Whishaw
1990 ) .
Language
Continuing the classical connectionist tradition of Hugo
Liepmann , Norman Geschwind ( 1965 ) championed the simpli-
fed yet controversial position that the parietal lobe acts as the
association area of association areas. Neural tissue damage to
this area often results in the classical disconnection syndromes ,
for example, apraxia and others.
Aleksandr Romanovich Luria ( 1973 ) considered the parietal
cortex one piece in his two-part model of mental activity, stat-
ing that it was important for understanding reception, analysis,
and storage of information. Lesions to the left parietal lobe were
understood to result in aferent motor aphasias (dif culty in
fnding the correct articulatory positions for specifc phonemes),
particularly lesioned primary and secondary sensory areas
afecting speech motor control and lesioned tertiary sensory
area resulting in aphasia (the loss of speech production and/or
comprehension).
Recent research, such as that of Gregory Hickok ( 2000 ), sug-
gests that the inferior parietal lobe serves as the connection
between phonological representations and motor control for
those representations, that is, the auditory-motor interface,
which is part of a larger network of interfaces and systems
subserving language function. Marco Catani , D. Jones, and
H. Dominic ( 2005 ) in a signifcant paper, confrm the analysis
that includes the inferior parietal lobe in the use and possibly
the acquisition of language via a new circuit connecting the
traditional language areas of broca and wernicke . It has
been labeled Geschwinds territory in honor of Geschwinds
original proposal that the parietal lobe is critical to language
function.
In sum, the left parietal cortex has particular areas that are
responsible for various linguistic functions. However, there are
other extralinguistic processes that the parietal lobe is known for
as well .
Extralinguistic Processes
ATTENTION. Te function of the parietal lobe in attention mecha-
nisms has been discussed over a period of time. Michael Posner
and Steven E. Peterson ( 1990 ) outlined the diferent subsystems
of attention: a) orientation to sensory events (not conscious), b)
signal detection for focal processing (conscious), and c) main-
tenance of a vigilant state (conscious). From the available neu-
rocognitve evidence, the researchers assert that the posterior
parietal lobe plays an important role in attention mechanisms,
specifcally in orientation and signal detection that are essential
for linguistic processing. Earlier, Luria ( 1973 ) identifed this par-
ietal region that mediates attention as an involuntary orienting
system. However the posterior parietal attentional mechanisms
are greatly impacted by the frontal regions that subserve alerting
mechanisms as well .
MEMORY. Traditionally, episodic memory, or declarative mem-
ory, has been attributed to the medial temporal lobe (MTL);
Parietal Lobe
587
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Battison , Robbin . 1978 . Lexical Borrowing in American Sign Language .
Silver Spring, MD : Linstok .
Bresnan , Joan . 1982 . Te Mental Representation of Grammatical Relations .
Cambridge, MA : MIT Press.
Browman , Catherine P ., and Louis Goldstein . 1990 . Articulatory gestures
as phonological units. Phonology 6 : 20151.
Charniak , Eugene . 1993 . Statistical Language Learning . Cambridge,
MA : MIT Press .
Chomsky , Noam . 1965 . Aspects of the Teory of Syntax . Cambridge,
MA : MIT Press .
Fowler , Carol A ., and Julie M. Brown . 2000 . Perceptual parsing of acous-
tic consequences of velum lowering from information for vowels .
Perception and Psychophysics 62 : 2132.
Frazier , Lyn . 1987 . Sentence processing: A tutorial review. In Attention
and Performance. Vol. 12: Te Psychology of Reading . Ed. M. Coltheart,
55986. Hillsdale, NJ : Lawrence Erlbaum .
Frazier , Lyn , and Janet Dean Fodor . 1978 . Te sausage machine: A new
two-stage parsing model . Cognition 6 : 291325.
Hale , John . 2006 . Uncertainty about the rest of sentences . Cognitive
Science 40 : 64372.
Jurafsky , Daniel , and James H. Martin . 2000 . Speech and Language
Processing: An Introduction to Natural Language Processing,
Computational Linguistics, and Speech Recognition . Upper Saddle
River, NJ : Prentice Hall .
McClelland , James , and Mark St. John . 1989 . Sentence comprehen-
sion: A PDP approach . Language and Cognitive Processes 4 : 287336.
Steedman , M . 1999 . Connectionist sentence processing in perspective .
Cognitive Science 23 : 61534.
Tabor , W ., and M. Tanenhaus . 1998 . Dynamical models of sentence pro-
cessing . Cognition 23 : 491515.
PARSING, MACHINE
Te query Over which strait in North Wales did Tomas Telford
build a suspension bridge? illustrates the fact that natural lan-
guages have complex syntactic structures. Comparison of the
question with the answer He built a suspension bridge over the
Menai Strait reveals that the phrases including strait, occur in
diferent positions in the two utterances, and that the verb posi-
tions are quite diferent, leading linguists to propose a constit-
uent structure like (1) for the question:
(1) [[ over/Preposition [which/Determiner [strait/Noun [in/
Preposition [North/Noun Wales/Noun ]NN ]PP ]N1 ]NP ]
PP [did/V
aux
Tomas_Telford/NP [build/Verb [a/
Determiner [suspension/Noun bridge/Noun ]Noun ]
NP t/PP ]VP ]S
inv
]S
q
A parser is a program that analyzes sentences in order to fgure
out their structure, using a list of rules describing the structure of
the language, such as:
(2) S NP VP
S
q
PP S
inv
S
inv
V
aux
NP PP
PP Preposition NP
NP Determiner N1
etc .
Most parsers begin by determining the part of speech of each
word (see word classes ). A bottom-up parser then attempts to
Kandel , Eric R ., James H. Schwartz , and Tomas M. Jessell . 1991 .
Principles of Neural Science , 3d ed. New York : Elsevier.
Kolb , Bryan , and Ian Q. Whishaw . 1990 . Fundamentals of Human
Neuropsychology . New York : Freeman .
Joseph , Rhawn . 2000 . Neuropsychiatry, Neuropsychology, Clinical
Neuroscience . New York : Academic Press.
Juhani Hyvarinen . 1982 . Te Parietal Cortex of Monkey and Man: Studies
of Brain Function . Berlin : Springer Verlag .
Luria , Aleksandr Romanovich . 1973 . Te Working Brain: An Introduction
to Neuropsychology . New York : Basic Books .
McFie John , and Oliver L. Zangwill . 1960 . Visual-constructive disabili-
ties associated with lesions of the left cerebral hemisphere . Brain
83 : 243 60.
Posner , Michael , and Steven E. Peterson . 1990 . Attention system of the
human brain . Annual Review of Neuroscience 13 : 25 42.
Stein, J. F. 1989. Representation of egocentric space in the posterior pari-
etal cortex. Experimental Physiology 74: 583606.
Turner , William . 1873 . Te Convolutions of the Brain in Relation to
Intelligence . Yorkshire : Te West Riding Lunatic Asylum Medical
Reports .
Wagner , Anthony D. , Benjamin J. Shannon , Itamar Kahn , and Randy
L. Buckner . 2005 . Parietal lobe contributions to episodic memory
retrieval. Trends in Cognitive Sciences 9.9 : 445 53.
PARSING, HUMAN
In general, parsing refers to breaking something into its
constituent parts. Tus, machines (see parsing, machine ) and
humans can decompose a message (such as print or spoken lan-
guage) into phrases, words , and morphemes . Most commonly
human parsing has been considered in the context of sentence
processing, particularly its syntactic and semantic aspects.
Language spoken, written, and signed can also be described
in terms of smaller functional units, including syllables, pho-
nemes , features, and gestures .
An understanding of grammatical constraints has guided the
development of descriptive representations (e.g., sentence dia-
grams) and formal systems of language structure and use. Parsing
models also have been infuenced by linguistic, psycholinguistic ,
and cognitive theory and by techniques used in computational
linguistics , natural language processing, and speech recogni-
tion (Chomsky 1965 ; Bresnan 1982 ; Jurafsky and Martin 2000 ).
Representative approaches include linguistic, statistical, connec-
tionist, and dynamical systems models (Charniak 1993 ; Hale
2006 ; McClelland and St. John 1989 ; Steedman 1999 ; Tabor and
Tanenhaus 1998 ; see also self-organizing systems ).
Te scale at which we can break the signal into pieces depends
upon both our attention to detail and to our descriptive goals, as
can be seen in numerous psychological studies that range from
ambiguity resolution (Frazier and Fodor 1978 ; Frazier 1987 ) to
assessment of our ability to perceive, produce, and use informa-
tion at various levels of description. Parsing linguistic information
is not restricted to sound and print but can include a consider-
ation of the gestures underlying the production of language by
voice (the coordinated movement of speech articulators, such as
the tongue body, tongue tip, jaw, and lips; see speech produc-
tion ) and sign (manual, facial, and body orientation) (Battison
1978 ; Browman and Goldstein 1990 ; Fowler and Brown 2000 ) .
Philip Rubin
Parsing, Human Parsing, Machine
588
frst, meaning. A nice derangement of epitaphs thus literally
means a nice arrangement of epithets if uttered with the relevant
semantic intention and understood accordingly. According to
Davidson, this does not obliterate the distinction between lit-
eral meaning and speakers meaning; speakers meaning for
instance, metaphorical meaning always comes later in the
order of intentions.
According to Davidson, Tarski-style theories of truth
(T-theories) can be used as formal semantic theories. To spe-
cify the literal meaning of any utterance, be it ever so idiosyn-
cratic, a full T-theory is required. In the case of malapropisms
and other novel or idiosyncratic use, these theories will be of a
transient, passing character; they might not hold for more than
a single utterance. If they hold for a certain utterance, Davidson
speaks of speaker and hearer sharing a passing theory (for
that utterance). Prior theories, on the other hand, specify the
interpretations speakers expect hearers to make, and hearers are
prepared to make, prior to actual utterances (cf. Davidson [ 1986 ]
2005, 101f).
Davidson then uses the terminology of prior and passing
theories to renew his argument against any account of linguistic
competence essentially involving the prior mastery of a system
of shared semantic and syntactic conventions or rules: [S]har-
ing such a previously mastered ability [is] neither necessary nor
suf cient for successful linguistic communication (Davidson
[ 1994 ] 2005, 110; cf. also [ 1982 ] 1984). To model successful lin-
guistic communication, systematic semantic theories of pass-
ing and prior nature are required, but sharing of prior theories
is not suf cient for successful linguistic communication. Even if
speaker and hearer share a prior theory, the ability to interpret
in accordance with that theory does not account for those cases
of successful communication where words are used in novel or
idiosyncratic ways. Nor is a shared prior theory necessary for
communication to succeed all that is necessary is that the pass-
ing theory be shared. Sharing passing theories, however, does
not amount to sharing a previously mastered ability: In conclu-
sion, then, I want to urge that linguistic communication does not
require, though it very often makes use of, rule-governed rep-
etition; and in that case, convention does not help explain what
is basic to linguistic communication, though it may describe a
usual, though contingent feature (Davidson [ 1982 ] 1984, 280).
Davidsons 1986 paper has been heavily criticized, among
others by Michael Dummett. Part of the criticism is due to the
provocative formulation Davidson gives there to his conclu-
sion: [T]here is no such thing as a language, not if a language
is anything like what many philosophers and linguists have sup-
posed ([1986] 2005, 107). A controversy between Davidson and
Dummett ensued regarding the questions of whether the notion
of an idiolect is to be explained in terms of a communal language
or the other way around, and whether meaning is essentially nor-
mative or prescriptive. Davidson argues that any obligation we
owe to conformity is contingent on the desire to be understood
([1994] 2005, 118), and he explicitly opposes those forms of social
meaning externalism (such as Tyler Burges), according to
which the literal meaning of words is essentially a matter of the
linguistic practices of the community surrounding the speaker
(Davidson [ 1994 ] 2005, 119). Just as for Gricean accounts of
meaning, there are also issues of psychological realism that arise
group the words into phrases and phrases into clauses, according
to the grammar rules, keeping track of multiple possible analyses
because of the extensive ambiguity of natural languages. Top-
down parsers, though perhaps less intuitive, are more frequently
used: Tey essentially work by attempting to generate the input
sentence .
Standard parsing algorithms for analyzing any artificial
language (such as programming languages) have been devel-
oped and can be used with any context-free grammar. Tus,
linguists can write the grammar rules: Tey do not need to be
programmers. But linguists grammars of natural languages
often make use of additional devices, such as agreement or
subcategory features (as with S
inv
and V
aux
in (1) and (2), denot-
ing inverted sentences and auxiliary verbs). generative
grammars, therefore, usually augment constituent structure
with additional information: Tere may be labels to uniquely
identify individuals or additional levels of information, such as
meanings. Work on feature-based frameworks such as lexical-
functional grammar and head-driven phrase struc-
ture grammar has gone hand in hand with the development
of complementary parsing methods .
John Coleman
SUGGESTION FOR FURTHER READING
Jurafsky , Daniel , and James H. Martin . 2000 . Speech and Language
Processing . Upper Saddle River, NJ : Prentice Hall .
PASSING THEORIES
Passing theories are utterance-specifc formal semantic theo-
ries; they specify the correct interpretation, or literal meaning,
of particular linguistic utterances: sentences uttered by par-
ticular speakers at particular times. Te expression passing
theory was coined by Donald Davidson in his 1986 paper A
Nice Derangement of Epitaphs, which was part of his attack on
accounts of linguistic communication essentially involving con-
ventionally determined, shared meanings ([1986] 2005).
According to Davidson, expressions like language , meaning ,
or sentence are theoretical terms used for describing, or explain-
ing, successful linguistic communication (cf. [1992] 2001, 108f).
For communicative success, regular or conventional use of
words is not necessary; what is necessary is only that the hearer
understand what the speaker intends to mean. For instance, if by
the words a nice derangement of epitaphs the speaker intends
to mean a nice arrangement of epithets and the hearer under-
stands that, we have a case of successful linguistic communica-
tion. Davidson suggests characterizing communicative success
in terms of the semantic intentions of the speaker. Tese he con-
strues as intentions to be interpreted in a particular way on a par-
ticular occasion and by a particular hearer. Moreover, they are of
a Gricean, self-referential form (see communicative inten-
tion ): A semantic intention is an intention to achieve the end
of being interpreted in a certain way by means of the intentions
being recognized by the hearer (Davidson [ 1986 ] 2005, 92 f).
Any utterance is made with a number of intentions that can be
ordered in terms of means to ends; the frst intention in such
a sequence (as ordered by in order to) specifes its literal, or
Passing Theories
589
speech community, defned as an organization of diversity that
had to be constituted and managed via performances, rather
than as a preexisting homogeneous entity.
In his concern with how language functions in society,
Hymes was inspired by the work of the prewar Prague School
(192938) and, in particular, by Roman Jakobson (18961982).
Working against Russian Formalisms emphasis on the inner
laws and formal structure of text without regard for context,
the Prague School focused attention on the multifunctionality
of language. Jakobson (1960), building on work by Karl Bhler
and Jan Mukaovsk, identifed six constitutive factors of a com-
municative event and postulated that each factor was associated
with a particular language function. Jakobsons constitutive fac-
tors include addresser, addressee, context, message, contact,
and code; he termed their associated functions expressive (or
emotive), conative, referential, poetic, phatic, and metalingual.
Tus, for instance, an utterance (such as eee-gads!) that directs
attention to the addresser (speaker) would be associated with
the expressive function, and so on. Tis model provided a basis
from which scholars could investigate the relationships among
form, function, and meaning .
In attending to speaking as a social accomplishment, the eth-
nography of speaking opened the way for studies of language
as an arena for the performance of social identities (see iden-
tity, language and ). Earlier studies tended to focus on the
organization of communicative life in small, often face-to-face
communities, highlighting the diferential distribution of lin-
guistic resources by age, gender, ethnicity, or other status
markers (see Bauman and Sherzer 1974 ). Later works consider
how particular linguistic performances are both embedded in
and help to shape wider political or cultural formations, such
as race relations, subcultural or national identities, multicul-
turalism, secularism, and the like. Linguistic anthropologys
historical emphasis on ways of speaking, strategies of voicing,
participation structures, and orientation to audiences made the
feld especially amenable to the approach of the Russian liter-
ary theorist Mikhail Bakhtin ( 1981 ), whose work on dialogism
and heteroglossia inspired studies in areas including lan-
guage ideology , genre, and intertextuality (Silverstein
and Urban 1996 ).
By foregrounding speaking as a social performance, the eth-
nography of speaking countered an alternative use of the term
performance proposed by Noam Chomsky ( 1965 ). Drawing
on a distinction made by Ferdinand de Saussure ([1907] 1959 )
between language ( langue ) and speech ( parole ), Chomsky
defned performance as the incomplete and imperfect realization
of language by particular speakers. He opposed performance to
competence , an internalized set of general rules that constitute
ones knowledge of a language, abstracted from particularities of
performance. In contrast, theorists of performance, along with
many linguistic anthropologists and sociolinguists , empha-
size communicative competence, understood not as a hypothet-
ical capacity for language but as the contextually grounded and
culturally acquired ability to speak in socially appropriate ways
(Bauman 1977 , 11). Here, speaking is understood as a creative
and emergent act through which social life is accomplished. As
such, speaking is inherently risky; it involves skill and account-
ability and is subject to critical evaluation .
for Davidsons account of successful linguistic communication
in terms of the complicated semantic intentions of the speaker .
Kathrin Gler
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bar-On , Dorit , and M. Risjord . 1992 . Is there such a thing as a language?
Canadian Journal of Philosophy 22 : 163 90.
Davidson , Donald . [1982] 1984. Communication and convention .
Inquiries into Truth and Interpretation , 26580. Oxford: Clarendon.
. [1986] 2005. A nice derangement of epitaphs. In Truth, Language,
and History , 89108. Oxford : Clarendon .
. [1992] 2001. Te second person. In Subjective, Intersubjective,
Objective : 10722. Oxford : Clarendon .
. [1994] 2005. Te social aspect of language. In Truth, Language,
and History , 10926. Oxford : Clarendon .
Dummett , Michael . A nice derangement of epitaphs: Some comments
on Davidson and Hacking. In Truth and Interpretation: Perspectives
on the Philosophy of Donald Davidson , ed. E. Lepore , 45976.
Oxford : Blackwell .
Gler , Kathrin . 2001 . Dreams and nightmares: Conventions, Norms,
and Meaning in Davidsons Philosophy of Language. In Interpreting
Davidson , ed. P. Kotatko , P. Pagin , and G. Segal , 5374. Stanford,
CA : CSLI .
Pietroski , Paul . 1994 . A defense of derangement . Canadian Journal of
Philosophy 24 : 95 118.
PERFORMANCE
Te study of performance investigates communicative practices
in their sociocultural contexts from three perspectives. First, it
foregrounds the performativity of communicative forms and
practices as modes of action or means of accomplishing social
ends. Second, it directs attention to the poetics of communica-
tive practice or to the forms of verbal artistry through which com-
municative acts are crafted and communicative skill is displayed.
Tird, it focuses attention on performances as a special class of
events, such as rituals, spectacles, festivals, or fairs, in which a
societys symbols and values are publicly displayed, interpreted,
and transformed. Within language study, the frst and second
perspectives have been foregrounded .
Te contemporary focus on the poetics and performance of
communicative practice emerged in the subdiscipline of linguis-
tic anthropology from a line of inquiry called the ethnography of
speaking . Developed by Dell Hymes and his students during the
1960s and 1970s, the ethnography of speaking highlights perfor-
mance in two linked ways: as speaking practice and as artfully
marked ways of speaking (Bauman and Sherzer 1975 ). Its cen-
terpiece is what Hymes called the speech event or communica-
tive event, a framework that allowed scholars to analyze multiple
components of language in use, including setting , participants ,
ends (goals, purposes), act sequences , key (tone, tenor), instru-
mentalities (channel, code), norms , and genres (the SPEAKING
acronym provides a mnemonic) (Hymes 1967 ). Te interest was
not simply in cataloging these components but, rather, in under-
standing how speakers use language within the conduct of social
life. In highlighting the emergent and creative nature of speech
performance, the ethnography of speaking focused attention
on linguistic forms as resources for living, in Kenneth Burkes
([1941] 1973 ) sense. Further, it proposed a new unit of study, the
Performance
590
to forms of social relationship aligned with the democratic
aspirations of their movement (the Berber Cultural Movement
was a minority ethnolinguistic , subnational, and secular
opposition movement in a majority Arabo-Islamist nation).
A chorus ofered a way of teaching children new gender roles
while displaying new modes of gender interaction to the wider
community. To accomplish this, the young men created a new,
highly marked event within the already marked wedding: Tey
mounted a stage, rented microphones, hung lights, and thus
confgured an entirely new relationship between performers
and audience, placing the guests in an unfamiliar spectator
role. Te children sang political songs that, while well known,
were not typically associated with weddings. Tis repertoire
provided the backdrop for yet a third performance: An ado-
lescent girl recited a poem on gender relations written by her
brother (the chorus director) a novel form of verbal art that
until then had no possibility of public performance in the vil-
lage. Yet the girl appeared to be only partially invested in serv-
ing as a spokesperson for her brothers text (she animated the
text, in Gofmans sense); at one point, she stumbled over the
words, and her brother prompted her, mouthing the words
from the sidelines. Te event culminated in a rousing dance in
which the children spontaneously organized themselves into
malefemale couples, a transformation of gender roles in dance
that galvanized the audience for nearly an hour.
Tis multilayered performance highlights the use of verbal
art (songs, poems) alongside other performance modes to efect
a transformation of the social relations of gender. It also illus-
trates diferential relations to linguistic resources and linguis-
tic authority (a concern of the ethnography of speaking): Te
childrens chorus had access to political repertoires but not to
womens traditional songs or henna poems. A young man could
fashion himself as the author of a poem; a young woman could
only animate it, and was subjected to her brothers corrective
voicing from the sidelines. Further, it shows how the participant
structure (frst made salient in Hymess SPEAKING model) was
both creatively altered for political ends and amenable to mul-
tiple interpretations. Putting girls on a public stage constituted
a display of political commitment to democracy for the young
men; for the girls, in contrast, their appearance on stage was a
highly controversial and far more ambivalent deviation from the
social norms of female performance. Beyond gender consider-
ations, this performance clearly reoriented what was typically
framed as a purely local event to wider ethnolinguistic and sub-
national concerns. Yet embedding this political orientation into
the already sanctioned frame of the wedding entailed less risk
(and ensured greater audience) than mounting a stand-alone
political event might have done.
In sum, the study of performance provides a point of entry
for research into social life as it is constituted, critiqued, and
transformed through communicative practices. It highlights the
emergent, creative, and transformative nature of language use in
a sociocultural context. Finally, performance ofers a compelling
vantage point on the mutually constitutive relationship between
seemingly microlevel practices and wider processes, ideologies,
and political formations .
Jane E. Goodman
Richard Bauman highlights the dimensions of risk, respon-
sibility, and accountability in what has become a classic defni-
tion of performance: Performance as a mode of spoken verbal
communication consists in the assumption of responsibility to
an audience for a display of communicative competence (1977,
11). Inspired by Hymes, Bauman has been particularly inter-
ested in the forms of verbal artistry through which communica-
tive skill is put on display. His work generated a pivotal shift in
folklore studies from a classifcatory concern with texts indepen-
dent of their contexts of use to an interest in the performance of
verbal art as a constitutive ingredient of social life. Performance
in this sense may range from sustained, full performance to a
feeting breakthrough into performance, with hedged or negoti-
ated performance lying somewhere in between (Bauman 2004 ,
110; the phrase breakthrough into performance comes from
Hymes 1981 ). Both Bauman and the interactional sociologist
Erving Gofman have been interested in how performances are
framed or keyed, but whereas Gofmans approach is dramatur-
gical, highlighting how social actors move from back stage
regions to perform the face work associated with an array of
social roles (Gofman 1959 ), Baumans interest lies in poetics ,
voice , and genre as verbal resources for the accomplishment of
social ends.
Tus far, performance has been considered from two related
vantage points, each grounded in particular disciplinary per-
spectives: Performance as speaking practice has been a focus
of linguistic anthropology and sociolinguistics; performance
as verbal art has been highlighted in folkloristics and linguistic
anthropology. A third approach views performance as a special
class of marked events in which a societys symbols are dis-
played for commentary, interpretation, or transformation. Tis
approach, pioneered by Victor Turner ( 1967 , 1969 ), is less con-
cerned with language per se. Trough its focus on collective rep-
resentations, cultural symbolism, and collective efervescence,
or communitas, it is located in a Durkheimian paradigm, with
inspiration from Arnold Van Genneps work on rites of passage.
Increasingly, however, scholars are drawing on aspects of all
three approaches. One example of how the three approaches
may be productively considered together is Jane Goodmans
analysis of a childrens performance in the Kabyle Berber region
of Algeria ( 2005 ) .
Te performance in question took place at a wedding,
understood as a festive occasion in which villagers suspended
interpersonal or political conficts and came together to collec-
tively celebrate the new union. Te wedding was set apart from
everyday life by various formal markers: location (an outdoor
public square), timing (late evening), dress, music (traditional
band), and activities (dance). Special forms of verbal art also
marked the occasion: A hired poet recited a poem after henna
was applied to the grooms hand; older village women sang
traditional songs to mark transitions. Wedding guests danced
to show support for the new couple. In this village, men and
women shared the same dancing space but typically danced
sequentially rather than concurrently; in no case did they
dance as couples. One summer, however, village youth active
in the national Berber Cultural Movement formed a mixed-
gender childrens chorus as a way of changing gender relations
in the community and, more broadly, fostering a commitment
Performance
591
perform an action, as opposed to constatives, which describe a
state of afairs.
Te seminal source for this distinction is Austin ( 1962 ), pub-
lished posthumously as a written record of lectures delivered in
1955 (and based on earlier, largely unpublished ideas). Tis is
important for two reasons. First of all, much of Austins thinking
is actually contemporary with (though probably largely uninfu-
enced by) Ludwig Wittgensteins ideas on language-games .
Secondly, the 1962 monograph records an evolution in Austins
thinking, in which he starts from a distinction between two utter-
ance classes and ends up drawing the conclusion that this dis-
tinction is untenable and that all utterances perform. In order
to understand this major shift, it is necessary to trace the evolu-
tion in his model in some detail.
Constatives are defned as utterances that have truth condi-
tions, the prototype case being desciptions of states of afairs
(e.g., Te sun comes up in the East ). In contrast, Austins origi-
nal performatives do not have truth conditions, in that they do
not commit the speakers beliefs to the proposition expressed.
Utterances such as I hereby baptize this child John Doe do
things: Tey perform actions, in that they change reality from
one in which a child named John Doe does not exist to one in
which such a child does exist. Performatives do not have truth
conditions but, rather, felicity conditions ; that is, they are
only performed successfully (or happily, to use Austins term) in
specifc circumstances. For instance, baptizing is only performed
happily if the speaker has the proper authority to perform the
procedure (e.g., a priest), if the procedure is carried out correctly
and completely (using the appropriate verbal format), and if the
parties involved carry out any necessary subsequent conduct .
Te question arises whether performatives and constat-
ives have any formal identifying features, such as grammatical
or lexical devices, that provide cues for the hearer about their
pragmatic status. Austin originally thought that so-called per-
formative verbs might be a good candidate; in an utterance like
I (hereby) promise Ill fnish the essay on time , the matrix verb
promise marks the performance of the action of promising.
Constatives appear to lack such a marker. Since performative
verbs are easily identifed (they are frst person present tense
indicative, they collocate with hereby , etc), they may function as
powerful cues and can be said (following Searle 1969 ) to function
as illocutionary force indicating devices (or IFIDs) .
Te assumption that performatives must have performative
verbs proves untenable, however. A slightly variant formula-
tion of the aforementioned promise, such as Ill fnish the essay
on time , is functionally very similar, if not identical, to the ver-
sion with promise . Te problem is that the second version does
not contain a performative verb; therefore, if one assumes that
both versions perform the same action, it must necessarily follow
that performative utterances do not need to have explicit perfor-
mative verbs. Tis leads Austin to posit a distinction between
explicit and implicit (or, as he called them, primary performa-
tives) performatives.
We have now lost any kind of formal marking of performatives
since both constatives and implicit performatives lack overt per-
formative verbs. In fact, once one posits the existence of implicit
performatives, the possibility is raised that a constative such as
Te sun rises in the East is, in fact, an implicit version of the more
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bakhtin , Mikhail . 1981 . Te Dialogic Imagination . Ed. Michael Holquist ,
trans. Caryl Emerson and Michael Holquist. Austin: University of Texas
Press.
Bauman , Richard . 1977 . Verbal Art as Performance . Prospect Heights,
IL : Waveland .
. 2004 . A World of Others Words: Cross-Cultural Perspectives on
Intertextuality. Malden, MA, and Oxford : Blackwell .
Bauman , Richard , and Joel Sherzer , eds. 1974. Explorations in the
Ethnography of Speaking . London and New York : Cambridge University
Press .
. 1975 . Te ethnography of speaking . Annual Review of
Anthropology 4: 95 119.
Burke , Kenneth . [1941] 1973. Te Philosophy of Literary Form: Studies in
Symbolic Action . 3d ed. Berkeley and London : University of California
Press .
Chomsky , Noam . 1965 . Aspects of the Teory of Syntax . Cambridge,
MA : MIT Press .
Hymes , Dell . 1967 . Models of the interaction of language and social set-
ting . Journal of Social Issues 23 .2: 8 28.
. 1981 . In vain I tried to tell you: Essays in Native American ethno-
poetics . Philadelphia : University of Pennsylvania Press .
Gofman , Erving . 1959 . Te Presentation of Self in Everyday Life. New
York : Doubleday.
Goodman , Jane . 2005 . Berber Culture on the World Stage: From Village to
Video. Bloomington : Indiana University Press .
Gumperz , John J. , and Dell Hymes , eds. 1986. Directions in
Sociolinguistics: Te Ethnography of Communication. Oxford and New
York : Blackwell.
Jakobson , Roman . 1960 . Concluding statement: Linguistics and poetics.
In Style in Language , ed. T. A. Sebeok , 35077. Cambridge, MA : MIT
Press.
. 1990 . On Language . Ed. Linda R. Waugh and Monique Monville-
Burston . Cambridge : Harvard University Press.
Saussure , Ferdinand de . [1907] 1959. Course in General Linguistics .
Ed. Charles Bally and Albert Sechehaye , trans. Wade Baskin. New
York : Philosophical Library .
Silverstein , Michael , and Greg Urban . 1996 . Natural Histories of Discourse .
Chicago : University of Chicago Press .
Turner , Victor . 1967 . Te Forest of Symbols: Aspects of Ndembu Ritual .
Ithaca, NY : Cornell University Press .
. 1969 . Te Ritual Process: Structure and Antistructure .
Chicago : Aldine.
PERFORMATIVE AND CONSTATIVE
Te distinction between performative and constative utterances
was frst introduced by J. L. Austin , and is illustrative of a reaction
within language philosophy to the doctrine of logical posi-
tivism : Tis paradigm holds that sentence meaning can be
captured in terms of truth conditions (see truth conditional
semantics ) and logical relations, and that sentences that can-
not be thus verifed are essentially meaningless. In contrast,
ordinary language philosophy , as conceived by philoso-
phers such as J. L. Austin, Peter Strawson , and H. P. Grice (see
cooperative principle ), examines language in use, and thus
lies at the basis of the development of modern pragmatics .
Austin observes that there are utterances, such as I (hereby)
bequeath my watch to my brother , for which any evaluation in
terms of truth and falsity is irrelevant; this type of utterance he
labeled (at least at the outset) performatives, or utterances that
Performative and Constative
592
(ii) Directives (e.g., requests, suggestions, commands),
which consist of attempts by the speaker to get the hearer to
do something;
(iii) Commissives (e.g., promising, threatening, ofering),
which commit the speaker to some future course of action;
(iv) Expressives (e.g., apologizing, congratulating, thanking),
which express a psychological state;
(v) Declarations (e.g., declaring war, baptizing, chisten-
ing), which efect immediate changes in some institutional
state of afairs, typically relying on elaborate extralinguistic
institutions.
Tis classifcation, despite having been hugely infuential,
raises some serious problems. First of all, some speech-acts seem
to belong to more than one category: A complaint such as Im
upset that you forgot to put the trash out presumably expresses the
speakers psychological state (expressive) but might also be inter-
preted as an attempt to get the hearer to take the trash out (direc-
tive). Secondly, illocutionary acts refect the communicative
intention of the speaker, who hopes that this intention will be
recognized and interpreted accurately by the hearer. Again, this
raises the question as to how hearers are able to do so. Searles
answer is the performative hypothesis , whereby every utterance U
has an underlying format of the form I (hereby) Vp you (that) U ,
Vp representing the (explicit or implicit) performative verb. Tis
still begs the question how hearers know that, for instance , Te
door is standing wide open is the implicit version of I apologize for
leaving the door open , rather than of I am complaining that you
left the door open (or, for that matter, an indirect version of the
request Could you close the door? ) Te three traditional sentence
types (declarative, interrogative, imperative) potentially ofer
some help by functioning as IFIDs, as may some lexical refexes
associated with certain illocutionary acts (e.g., please appears to
co-occur exclusively with directives). However, the fact remains
that most speech-act realizations contain neither a performative
verb nor any other IFID. Such utterances, which exhibit no overt
structural marking of their speech-act status (as in Te door is
standing wide open when intended as a request), Searle labels
indirect speech-acts . However, since most usages of speech-acts
appear to be indirect rather than direct, it remains unexplained
how hearers are capable of computing the speakers intended
illocutionary force in the absence of structural signals. A possible
explanation is that people rely on contextual cues, working out
the implicit meaning by relying on Grices cooperative principle
through conversational implicatures .
Te fact remains that Searles classifcation ofers little help in
assigning speech-act status to stretches of verbal interaction in
ethnographic data. Ultimately, it could be argued, the interpreta-
tion of an utterance will depend on the speech event in which it
occurs, that is, the culturally recognized social activity in which
language plays a specifc, and often rather specialized, role
(Levinson 1983 , 279). In classroom interactions, for instance,
teacher questions regularly violate Searles sincerity condition
since the speaker already knows the answer and is thus not sin-
cere in trying to obtain a missing piece of information.
A radically diferent approach to the interpretation prob-
lem (i.e., how the speakers communicative intention is rec-
ognized in the absence of linguistic cues) is ofered by the
explicit I (hereby) state that the sun rises in the East (since state
here has all the characteristics of a performative verb). What is
more, one could claim that statements also have felicity condi-
tions in the sense that they are only uttered happily if the speaker
is reasonably sure about the truth of the proposition expressed.
Conversely, many performatives need to bear some relation to
actual facts and, thus, have at least some propositional content.
Te question thus arises whether constatives are similar to per-
formatives in that they also perform an action. Austin admits
that they do, namely, the action of committing the speaker to the
truth of the propososition: Once we realize that what we have
to study is not the sentence but the issuing of an utterance in a
speech situation, there can hardly be any longer a possibility of
not seeing that stating is performing an act (Austin 1962 , 139).
Te performativeconstative distinction thus becomes unten-
able, and one can only conclude that all utterances are actions.
Austins original two utterance classes are then merely sub-
classes of acts performed through language, or speech-acts ,
which consist of three distinct types of act: locution, illocution,
and perlocution .
Te locutionary act can be more or less equated to the
semantic meaning of the utterance and is roughly equivalent
to uttering a certain sentence with a certain sense and refer-
ence (Austin 1962 , 109). Illocutionary acts are utterances
which have a certain (conventional) force ( 1962 , 109), such
as baptizing, promising, and all of Austins original performa-
tives, but also former constatives such as informing and stating.
Perlocutionary acts, fnally, are what we bring about or achieve
by saying something ( 1962 , 109), that is, the consequences that
utterances trigger (which may, but need not be verbal), such as
convincing, deterring, or frightening. In short, all utterances per-
form three diferent acts: the act of saying something (locution),
what the speakers intention is in saying something (illocution),
and what its consequences are by saying something (perlocu-
tion). In much of the subsequent literature, the term speech-act
has become virtually synonymous with the illocutionary force of
the utterance, but it is important to stress that, for Austin, per-
forming a speech-act involves performing all three kinds of act
simultaneously .
Since all utterances are now considered to be performative
speech-acts, the question is raised as to how many diferent
classes of speech-acts can be distinguished on linguistic grounds.
Out of the three acts involved, locution does not provide any use-
ful distinguishing criteria since the same propositional content
can be employed for creating various speech-acts; neither does
perlocution since the perlocutionary efect of a speech-act is dif-
fcult to predict. However, utterances do difer systematically
with regard to their illocutionary force and, thus, presumably
have diferent felicity conditions. It should, therefore, be possible
to develop a new taxonomy of illocutionary acts based on these
felicity conditions or the linguistic realizations thereof.
Austin did, in fact, develop a rudimentary taxonomy, but it
was left to his pupil J. R. Searle to come up with a more systematic
classifcation (see Searle 1979 ). Searle distinguishes fve major
classes of illocutionary act:
(i) Representatives (e.g., stating, describing, concluding),
which commit the speaker to the truth of the expressed
proposition;
Performative and Constative
593
1861 ; Wernicke [1874] 1969 ), usually the left in right-handed
individuals (Broca 1865). Tis cortex surrounds the sylvian fs-
sure and runs from the pars triangularis and opercularis of the
inferior frontal gyrus (Brodmans areas [BA] 45, 44: Brocas area),
through the angular and supramarginal gyri (BA 39 and 40) into
the superior temporal gyrus (BA 22: Wernickes area) in the dom-
inant hemisphere ( Figure 1 ) .
Classical Clinical Models of the Functional Neuroanatomy
of Perisylvian Cortex for Language
Te frst theories of the functional neuroanatomy of language
pertained to this cortical region. Te pioneers of aphasiology
Paul Broca , Karl Wernicke , John Hughlings Jackson , and other
neurologists described patients with lesions in the left inferior
frontal lobe whose speech was hesitant and poorly articu-
lated, and other patients with lesions more posteriorly, in the
superior temporal lobe , who had disturbances of compre-
hension and fuent speech with sound and word substitutions
(see aphasia ). Tese correlations led to the theory that lan-
guage comprehension went on in unimodal auditory association
cortex ( wernickes area , BA 22) adjacent to the primary audi-
tory cortex (Heschls gyrus, BA 41), and motor speech planning
went on in unimodal motor association cortex in brocas area
(BA 44 and 45) adjacent to the primary motor cortex (BA 4).
Tese theories incorporated the only principle that has ever been
articulated regarding the localization of a language operation.
According to this principle, language operations are localized in
relation to their sensory-motor requirements. Speech planning
goes on in Brocas area because Brocas is immediately adjacent
to the motor area responsible for movement of the articulators,
and Wernickes area is involved in comprehension because it
is immediately adjacent to the primary auditory cortex. Tese
ideas and models were extended by Norman Geschwind and his
colleagues in the 1960s and 1970s. Geschwind ( 1965 ) added the
hypothesis that word meaning was localized in the inferior
parietal lobe (BA 39 and 40) because word meanings consist
of associations between sounds and properties of objects, and
the inferior parietal lobe is an area of multimodal association
cortex to which fbers from unimodal association cortex related
to audition, vision, and somasthesis project.
ethnomethodological paradigm of conversation analysis .
Consider the following exchange:
S: Another glass of wine would hit the spot.
H: I dont think so mate, youve had enough.
Ss utterance, despite being a declarative, is clearly not inter-
preted as simply stating a fact by H; rather, Hs response (a
refusal to comply) shows that it was interpreted as a request
for another glass of wine. Te basis for interpretation here lies
in the conversational sequencing of the two contributions: Tey
are conditionally reliant upon each other, by virtue of being two
parts of a request-refusal adjacency pair . Te question as to
the intended illocutionary force of Ss turn becomes moot in this
approach; what matters is that H has clearly interpreted it as
request-like, having provided an appropriate second part to the
adjacency pair. Of course, H might provide an incorrect interpre-
tation, but if this is the case, it will become appararent in the sub-
sequent interaction. Such an inductive approach avoids some of
the pitfalls inherent in attempts to classify speech-acts according
to the nonobservable, and therefore unfalsifable, intentions of
the speaker .
Ronald Geluykens
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Austin , J. L. 1962 . How to Do Tings with Words . Oxford : Oxford University
Press .
Levinson , S. C . 1983 . Pragmatics . Cambridge : Cambridge University
Press .
Searle , J. R . 1969 . Speech Acts: An Essay in the Philosophy of Language .
Cambridge : Cambridge University Press.
. 1979 . Expression and Meaning: Studies in the Teory of Speech Acts .
Cambridge : Cambridge University Press.
PERISYLVIAN CORTEX
Beginning in the late nineteenth century, the application of def-
cit-lesion correlations based on autopsy material to the problem
of the regional specialization of the brain for language yielded
the fact that human language requires parts of the association
cortex in the lateral portion of one cerebral hemisphere (Broca
Brocas area
Brocas area
Wernickes area Wernickes area
Inferior parietal lobe
Figure 1. A depiction of the left hemisphere of the brain
showing the main language areas.
Perisylvian Cortex
594
evidence that this wave originates in the inferior temporal lobe
(Nobre and McCarthy 1995 ), though perhaps more posteriorly
than the lesion studies would suggest. Other brain areas that
have been suggested as loci for semantic processing (the infe-
rior frontal lobe: Petersen et al. 1988 ; Dapretto and Bookheimer
1999 ) are much less clearly related to this function .
In the past two decades, studies of impairments of word
meaning and functional neuroimaging have suggested a fner-
grained set of distinctions within the class of objects. Both def-
cits and functional activation studies have suggested that there
are unique neural loci for the representation of categories such
as tools (frontal association cortex), animals and foods (lateral
inferior temporal lobe), and faces (medial inferior temporal lobe)
(see Caramazza and Mahon 2006 , for review). Debate continues
as to whether such divisions and localizations refect diferent
co-occurrences of properties of objects within these classes or
innate, neurally localized human capacities to divide the world
along these lines .
SYNTACTIC PROCESSING. Most researchers also subscribe to
localizationist views regarding aspects of syntactic process-
ing. A well-known hypothesis is the trace deletion hypothesis
(Grodzinsky 2000 ), which claims that patients with lesions in
Brocas area have defcits afecting certain moved constituents
( traces in Chomskys theory). Te evidence supporting these
models is based on correlating defcits in syntactic comprehen-
sion to lesions. However, there are two issues that such data must
face. First, it is often not clear whether a patient has a defcit in
a particular parsing operation or a reduction in the resources
available to accomplish syntactically based comprehension.
Second, there is virtually no consistency in an individual patients
performance across tasks, raising questions about whether a
patient who fails on a particular structure has a parsing defcit
(Caplan, DeDe, and Michaud 2006 and Caplan, Waters, Dede, et
al. 2007) .
Assuming that patients performances refect defcits in
particular parsing operations, the relation of these defcits to
lesions does not support invariant localization models. We have
recently reported the most detailed study of patients with lesions
whose syntactic comprehension has been assessed (Caplan,
Waters, Kennedy, et al. 2007). Lesion size in multiple, cytoarchi-
tectonically diferent small areas of cortex both within and out-
side the perisylvian and non-perisylvian area, not connected by
major fber tracts, predicted performance, ruling out invariant
localization as the mode of neural organization for the opera-
tions supporting this function that were assessed. At the same
time, patients who performed at similar levels behaviorally had
lesions of very diferent sizes in larger areas of the brain (such
as the perisylvian association cortex, or the entire left hemi-
spheric cortex) in which it has been suggested that syntactic
processing might be distributed, and patients with equivalent
lesion sizes in these larger areas varied greatly in their level of
performance, arguing that syntactic processing in comprehen-
sion is not distributed in these areas. Te data are consistent
with a model in which the neural tissue that is responsible for
the operations underlying sentence comprehension and syntac-
tic processing is localized in diferent neural regions in diferent
individuals .
Despite its widespread clinical use, however, this model has
serious limitations. It deals only with words, not other levels of
the language code. From a linguistic and psycholinguistic
point of view, the syndromes are all composed of many process-
ing defcits, which are diferent in diferent patients. Te syn-
dromes themselves do not provide a guide to the localization of
more specifc components of the language processing system.
As reviewed in the following, Geschwinds critical contribu-
tion regarding the role of the parietal lobe receives no empirical
support .
Linguistically Oriented Models of the Functional
Neuroanatomy of the Perisylvian Cortex for Language
Since approximately 1975, psychologists and linguists have
approached language disorders and their neural basis in a more
systematic fashion, informed by models of language structure
and function. I briefy review two areas of work that relate these
models to the functional neuroanatomy of the persiylvian cortex
and other brain regions.
LEXICAL SEMANTIC PROCESSING. As noted, traditional neu-
rological models of the neural basis for word meaning main-
tained that the meanings of words consist of sets of neural
correlates of the physical properties that are associated with a
heard word (Wernicke [1874] 1969 ), all converging in the infe-
rior parietal lobe (Geschwind 1965 ). It is now known that most
lesions in the inferior parietal lobe do not afect word meaning
(Hart and Gordon 1990 ), and functional neuroimaging stud-
ies designed to activate word meanings do not tend to activate
this region (see the following). A. Damasio ( 1989 ), therefore,
modifed this model, suggesting that the meanings of words
included retroactivation of neural patterns in unimodal
association primary sensory cortex. Evidence for this comes
from functional neuroimaging results that reveal activation
for diferent classes of words in diferent areas, each related
to the sensory-motor associations of the word (frontal cortex
for verbs and manipulable objects; inferior temporal cortex for
concrete nouns) (see Caramazza and Mahon 2006 , for review).
However, it is not clear that these activations refect the mean-
ing of words, rather than properties commonly associated with
words. Word meanings include much more than sensory and
motor associations; the essence of word meaning is itself quite
mysterious (Fodor 1998 ). In any event, word meanings are part
of a network that relates a word to a complex set of concepts
and contexts (Tulving 1972 ).
Tere is evidence that a critical part of this semantic network
is located outside the perisylvian cortex, in the anterior inferior
temporal lobes. Patients with semantic dementia, a degenera-
tive disease that afects the anterior inferior temporal lobe, and
herpes encephalitis, with somewhat more posterior lesions, have
initially selective and ongoing major problems with many aspects
of semantic memory (Davies et al. 2005 ; Gorno-Tempini et al.
2004; Warrington and Shallice 1984 ). Activation studies have
implicated the inferior temporal cortex in representing concepts
and word meanings (Caramazza and Mahon 2006 ). Some stud-
ies of the neural generators for the N400 event-related potential
(ERP) wave, which refects some aspect of semantic processing
(Kutas and Hillyard 1980 ; Holcomb and Neville 1990 ), present
Perisylvian Cortex
595
Caramazza , A. and B. Mahon . 2006 . Te organization of conceptual
knowledge in the brain: Te futures past and some future directions.
Cognitive Neuropsychology 23 : 13 38.
Damasio , A. 1989 . Time-locked multiregional retroactivation: A sys-
tems-level proposal for the neural substrates of recall and recogni-
tion . Cognition 33 : 25 62.
Dapretto , M. , and S. Y. Bookheimer . 1999 . Form and con-
tent: Dissociating syntax and semantics in sentence comprehension .
Neuron 24 : 427 32.
Davies , R. R. , J. R. Hodges , J. R. Kril , K. Patterson , G. M. Halliday , and J.
H. Xuereb . 2005 . Te pathological basis of semantic dementia . Brain
128 : 1984 95.
Fiebach , C. J. , M. Schlesewsky , and G. Lohmann . 2005 . Revisiting the
role of Brocas area in sentence processing: Syntactic integration ver-
sus syntactic working memory . Human Brain Mapping 24 : 79 91.
Fodor , J. A . 1998 . Concepts . Oxford : Oxford University Press .
Geschwind , N . 1965 . Disconnection syndromes in animals and man .
Brain 88 : 23794, 585 644.
Gorno-Tempini , M. L. , N. F. Dronkers , K. P. Rankin , J. M. Ogar ,
L. Phengrasamy , H. J. Rosen , J. K. Johnson , M. W. Weiner , and B. L.
Miller . 2004 . Cognition and anatomy in three variants of primary pro-
gressive aphasia . Annals of Neurology 55 : 335 46.
Grodzinsky , Y. 2000 . Te neurology of syntax: Language use without
Brocaa area . Behavioral and Brain Sciences 23 : 47 117.
Hart , J., Jr. , and B. Gordon . 1990 . Delineation of single-word seman-
tic comprehension defcits in aphasia, with anatomical correlation .
Annals of Neurology 27 : 226 33.
Holcomb , P. J ., and H. J. Neville . 1990 . Auditory and visual semantic
priming in lexical decision: A comparison using event-related brain
potentials . Language and Cognitive Processes 5.4 : 281 312.
Kutas , M. , and S. A. Hillyard . 1980 . Reading senseless sentences: Brain
potentials refect semantic incongruity . Science 207 : 203 4.
Nobre , A. C. , and G. McCarthy . 1995 . Language-related feld poten-
tials in the anterior-medial temporal lobe: II. Efects of word type and
semantic priming . Journal of Neuroscience 15 : 1090 8.
Petersen , S. E. , P. T. Fox , M. Posner , M. Minton , and M. Raichle . 1988 .
Positron emission tomographic studies of the cortical anatomy of
single-word processing . Nature 331 : 585 9.
Tulving , E. 1972 . Episodic and semantic memory. In Organization
of Memory , ed. E. Tulving and W. Donaldson , 381403. New
York : Academic Press .
Warrington , E. , and T. Shallice . 1984 . Category specifc semantic impair-
ments . Brain 107 : 829 53.
Wernicke , K. [1874] 1969. Te aphasic symptom complex: A psychologi-
cal study on a neurological basis. Breslau: Kohn and Wegert. Repr. in
Boston Studies in the Philosophy of Science. Vol. 4. Ed. R. S. Cohen and
M. W. Wartofsky , 3497 . Boston : Reidel .
PERLOCUTION
In pragmatics , perlocution refers to the efect speech-acts
have on the hearer (H). J. L. Austin ( 1962 ) distinguishes three
types of act that utterances perform simultaneously: locution
(roughly equivalent to the meaning in a propositional sense),
illocution (the intended force of the speech-act), and perlocu-
tion. Austin characterizes perlocution as follows: Saying some-
thing will often, or even normally, produce certain consequential
efects upon the feelings, thoughts, or actions or actions of the
audience, or of the speaker, or of other persons: and it may be
done with the design, intention, or purpose of producing them
(1962, 101). Hs reaction to an illocutionary act might be verbal
(e.g., asking a question might prompt an answer), or nonverbal
Functional neuroimaging studies have been said to provide
evidence for the localization of specifc parsing and interpretive
operations in Brocas area (Ben-Shachar et al. 2003 ; Ben-Shachar,
Palti, and Grodzinsky 2004 ; Bornkessel, Fiebach, and Friederici
2005 ; Fiebach, Schlesewsky, and Lohmann 2005 ). However,
most neuroimaging studies actually show multiple cortical areas
of activation in tasks that involve syntactic processing, and dif-
ferent areas have been activated in diferent tasks. Overall, these
data also suggest variation in the localization of the areas that
are suf cient to support syntactic processing within the language
area across the adult population, although invariant localization
models are not ruled out (Caplan, Chen, and Waters 2008) .
Overview
Te left perisylvian association cortex appears to be the most
important brain region supporting human language. However,
it is not the sole area involved in these abilities. How this area
and other brain regions act to support particular language oper-
ations is not yet understood. Tere is evidence for both locali-
zation of some functions in subparts of this region and other
brain areas, and for either multifocal or distributed involve-
ment of brain areas in other language functions. It may be that
some higher-level principles are operative in this domain. For
instance, content-addressable activation and associative opera-
tions such as those that underlie phoneme recognition, lexical
access, and lexical semantic activation, may be invariantly
localized, while combinatorial computational operations such
as those that constitute the syntax of natural language may not
be. However, many aspects of these topics remain to be studied
with tools of modern cognitive neuroscience .
David Caplan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ben-Shachar , M. , T. Hendler , I. Kahn , D . Ben-Bashat , and Y. Grodzinsky .
2003 . Te neural reality of syntactic transformations: Evidence from
fMRI . Psychology Science 14 : 433 40.
Ben-Shachar M ., D. Palti , and Y. Grodzinsky . 2004 . Te neural correlates
of syntactic movement: Converging evidence from two fMRI experi-
ments . Neuroimage 21 : 1320 36.
Bornkessel I. , C. Fiebach , and A. Friederici . 2005 . On the cost of syntactic
ambiguity in human language comprehension: An individual difer-
ences approach . Cognitive Brain Research 21 : 11 21.
Broca , P . 1861 . Remarques sur le sige de la facult du parole articul,
suivis dune observation daphmie (perte de parole) . Bulletin de l
Socit dAnatomie (Paris) 36 : 330 57.
Broca P. 1865. Sur le sige de la facult du langage articul. Bulletin de
la Socit danthropologie 6 : 33793.
Caplan, D., E. Chen, and G. Waters. 2008. Task-dependent and task-
independent neurovascular responses to syntactic processing. Cortex
44 : 25775.
Caplan, D., G. DeDe, and J. Michaud . 2006. Task-independent and task-
specifc syntactic defcits in aphasic comprehension. Aphasiology
20 : 893920.
Caplan, D., G. Waters, G. DeDe, J. Michaud, and A. Reddy. 2007. A study
of syntactic processing in Aphasia I: Behavioral (psycholinguistic)
aspects. Brain and Language 101 : 10350.
Caplan, D., G. Waters, D. Kennedy, N. Alpert, N. Makris, G. DeDe,
J. Michaud, and A. Reddy. 2007. A study of syntactic processing in
aphasia II: Neurological aspects. Brain and Language 101 : 15177.
Perisylvian Cortex Perlocution
596
groups including speakers but excluding addressees. Further
divisions include an impersonal category and a sentient/non-
sentient third person opposition. Like number and gender
marking , person can also be indicated on agreeing elements,
particularly fnite verbs. Present tense English verbs show only
third person singular agreement ( walk-s ), while agreement
on Italian indicative verbs distinguishes three persons in both
singular ( parl-o I speak, parl-i you speak, parl-a he/she/it
speaks) and plural.
Linguistic phenomena related to person include morpho-
logical categories of pronouns and agreement; partial mor-
phological syncretisms among person categories, in pronouns
or agreement; interactions of person with the ordering of pro-
nominal clitics ; interactions of person with case , agreement,
or structural position; and surprising restrictions on person com-
binations, usually involving direct and indirect objects (the *me
lui efect, or Person Case Constraint ). Such phenomena form the
empirical basis of morphosyntactic theories of person.
Tere are three principal theoretical approaches to person. A
traditional insight represents person categories within a hierar-
chy of nominals infuencing pronoun morphosyntax, for exam-
ple, case and agreement marking in transitive clauses (Dixon
1994 , 85). Cross-linguistically, third person is the least marked,
ranking below frst and second. For example, in Georgian, frst
and second person objects are indexed by verbal morphology,
while verbs with third person objects resemble intransitives. In
Dyirbal, frst and second person pronouns have nominative/
accusative case marking, while third person pronouns, proper
names, and common nouns show an ergative/absolutive oppo-
sition. Some scholars rank frst person highest (Zwicky 1977 ),
while others regard the ranking of frst and second person as
variable .
Another approach seeks to derive morphosyntactic efects by
representing person as a complex category built from elemental
features. One such feature analysis locates person features
such as [participant], [speaker], and [addressee] within a univer-
sal geometry of privative pronominal features, in which the avail-
ability of one feature may depend on the presence of another.
An infuential paper by Heidi Harley and Elizabeth Ritter (2002)
outlines this approach. Another type of analysis treats person
features as binary rather than privative; this allows the grammar
to refer to negative values, such as [speaker]. Robert Rolf Noyer
( 1997 ) makes a signifcant case for the binary-feature analysis .
A third approach, potentially compatible with the second,
associates diferent persons with diferent syntactic representa-
tions (Ritter 1995 ; Dchaine and Wiltschko 2002 ; Bejar 2003 ).
Within the featural approach, most commentators assume
the existence of features corresponding to frst and second per-
son. However, third person is widely treated as simply lacking
such features (Zwicky 1977 ; Noyer 1997 ). Tis analysis correctly
predicts certain limits on the typology of person categories
(Greenberg 1966 ). As noted, some languages have separate cat-
egories for inclusive and exclusive we , whose use depends on
whether addressees are included. Tus, [addressee] is a distinc-
tive feature; inclusive ([speaker, addressee]) has it, while frst
person ([speaker]) does not. However, there is no parallel con-
trast between categories whose use depends on whether nonpar-
ticipants are included. For example, no known languages have
(e.g., an insult may result in a slap in the face), but also an internal
psychological or emotional state (e.g., a threat might result in H
being frightened or angry).
Although Austin intended perlocution to be an integral part
of a speech-act, later developments of speech-act theory have
focused almost exclusively on illocution, that is, the speakers
mental state or intention (e.g., Searle 1969 ). As a result, the term
speech-act has become virtually synonomous with illocution-
ary force . Tis is perhaps unsurprising, given that perlocutions
do not always consist of observable behavior (and might therefore
be argued to fall outside a linguistic theory of pragmatics; but see
Gu 1993 ). Moreover, perlocutions are hard to classify: Not only
do certain illocutions allow for a range of possible perlocutions
(a request, for instance, may result in either compliance or rejec-
tion by the hearer); there is often no way of knowing whether the
achieved perlocution is actually the one the speaker (S) intended
to achieve (a warning, say, may be intended to make the hearer
(H) take evasive action but may only result in frightening him/
her). Nevertheless, it is easy to demonstrate that perlocutions are
intrinsic parts of speech-acts since their successful performance
often depends on them. As Austin points out, an utterance such
as I bet you 10 dollars the Knicks will win by 5 points is felicitous
only if it receives uptake , that is, if H acknowledges and accepts
the bet (see felicity conditions ) .
conversation analysis (Sacks 1992 ) ofers a potential
alternative, inductive approach to (verbal) perlocutions based on
local sequential organization. Consider the following exchange:
S: Have a cookie
H: ehm no thanks Ive just had dinner
In this exchange, Ss contribution can be labeled an ofer by vir-
tue of its being the frst part of an ofer-refusal (or ofer-accep-
tance) adjacency pair . If H recognizes Ss utterance as such,
he/she will have to provide a sequentially appropriate response
(or perlocution). Te second part is thus conditionally reliant on
the frst part .
Ronald Geluykens
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Austin , J. L . 1962 . How to Do Tings with Words . Oxford : Clarendon
Press .
Gu , Yueguo . 1993 . Te impasse of perlocution . Journal of Pragmatics
20 : 405 32.
Sacks , Harvey . 1992 . Lectures on Conversation . Oxford : Blackwell .
Searle , J. R . 1969 . Speech Acts: An Essay in the Philosophy of Language .
Cambridge : Cambridge University Press .
PERSON
Person is a morphosyntactic property of nominal phrases (nouns
and pronouns) used to indicate the discourse role of their ref-
erent. English personal pronouns show three person distinc-
tions: frst person, indicating speakers ( I , we ); second person,
indicating addressees ( you ); and third person, indicating dis-
course nonparticipants ( he , she , it , they ). Some languages also
distinguish inclusive and exclusive we : Ojibwa has kiinawint for
groups including speakers and addressees, and niinawint for
Person
597
Bruening , Yoonjung Kang , and Martha McGinnis , 42549. Cambridge,
MA : MITWPL .
Harley , Heidi , and Elizabeth Ritter. 2002. Person and number in pro-
nouns: A feature-geometric analysis . Language 78 : 482 526.
McGinnis , Martha . 2005 . On markedness asymmetries in person and
number . Language 81 : 699 718.
Noyer , Robert Rolf . 1997. Features, Positions, and Af xes in Autonomous
Morphological Structure . New York: Garland.
Ritter , Elizabeth . 1995 . On the syntactic category of pronouns and agree-
ment . Natural Language and Linguistic Teory 13 : 405 43.
Zwicky , Arnold M . 1977 . Hierarchies of person . Chicago Linguistic
Society 13 : 714 33.
PHILOLOGY AND HERMENEUTICS
Tis entry briefy outlines some aspects of the study of linguistics
leading up to the twentieth century. As two of the earliest, most
thoroughgoing attempts in the West to understand written texts
and spoken discourse, philology and hermeneutics represent
vital precursors of todays language science. Still synonymous
with classical studies and historical linguistics , philol-
ogy as both word and practice can be traced to ancient Greece
and Rome. While it is likewise based on a Greek word and while
the problem of interpretation engaged many ancient thinkers,
hermeneutics is often narrowly associated with vigorous philo-
sophical debates centered in late eighteenth-century Germany
and originating in Reformation treatises on the right interpreta-
tion of scripture. Today, the heritage of philology and hermeneu-
tics persists in the modern organization of university disciplines,
as well as in many indispensable scholarly monuments, such as
Te Oxford English Dictionary .
Philology implies love of language and once stood for lin-
guistics. Hermeneutics can be defned more specifcally as the
art (or science) of interpretation. Te progress from amateur art
to professional science marks the history of both. In their hey-
days, philology and hermeneutics were deemed central to all dis-
ciplines, whether scientifc or humanistic; at other times, either
discipline could also be reduced to trivial pedantry. Among their
more prescient discoveries are Sir William Joness hypothesis of a
common genetic origin for the evolution of all Indo-European
languages nearly a century before Charles Darwins On the
Origin of Species and the hermeneutic circle, the feedback-like
cycle of interpretation formulated by Friedrich Ast almost 150
years before the birth of cybernetics .
Te historical survey to follow highlights the respective ori-
gins, development, and interrelations of philology and her-
meneutics and is singularly appropriate, given the historical
predilection of both felds. Because of the limitations of space, the
focus remains on the European intellectual tradition. However,
the theme emphasized here, that early investigations of language
sometimes uncannily anticipated modern scientifc paradigms,
applies equally to non-Western traditions. In South Asia, for
instance, the classical Sanskrit grammar of Panini (ca. sixth to
ffth cent. b.c. ) strongly prefgures generative grammar .
Interest in the nature and origins of human language goes back
to the earliest Western literature, such as the Tower of Babel in
Genesis. Tere is also a fascinating folktale retold by Herodotus,
in which an Egyptian pharaoh isolates two children from birth
in order to see what language they will speak presumably the
separate categories for inclusive and exclusive plural you , whose
use depends on whether nonparticipants are included. Such
observations imply that there is no third person feature, there-
fore no categories such as [speaker, addressee, nonparticipant],
[speaker, nonparticipant], or [addressee, nonparticipant]. Tird
person pronouns thus refer to nonparticipants by default, lack-
ing the features that allow reference to discourse participants.
Nevertheless, some phenomena seem to require reference to
nonparticipants, for example, syncretism in Mam pronominal
enclitics (Noyer 1997 ) or the Spanish spurious se rule (Bonet
1991 ). An obvious solution is to permit limited reference to nega-
tive values, such as [speaker, addressee]. Te success of the
privative approach depends on identifying plausible alternative
analyses for such cases .
Although the [speaker] and [addressee] features are suf -
cient to generate the four main person categories attested cross-
linguistically, there is evidence for an additional [participant]
feature, shared by frst and second person (Farkas 1990 ; Noyer
1997; Halle 1997 ). For example, while Winnebago agreement dis-
tinguishes frst and second person, free personal pronouns only
distinguish participants from nonparticipants ( nee , I or you,
ee he/she).
Te argument against a [nonparticipant] feature also applies
to the [addressee] feature in languages without an inclusive cat-
egory (McGinnis 2005 ). Such languages treat the inclusive as frst
person, not second (Zwicky 1977 ; Noyer 1997 ). Tus, in such
languages, [addressee] is non-distinctive: Tere can only be an
opposition between [speaker] and non[speaker] participants,
not between [addressee] and non-[addressee] participants. If
[nonparticipant] is nonexistent because it is never distinctive,
then [addressee] is likewise nonexistent in languages without
an inclusive category. Tis suggests that the morphosyntactic
contrast between frst and second person is suf cient to activate
[speaker], while [addressee] can be activated only by an addi-
tional contrast between inclusive and frst person. In such cases,
[addressee] is indeed necessary to capture widespread (and non-
default) syncretisms between inclusive and second person
most famously identifed in Algonquian languages but common
among languages with an inclusive category. For example, the
inclusive pronoun in Ojibwa ( kiinawint ) shows syncretism with
both second person ( kiin , plural kiinawaa ) and frst ( niin , plural
niinawint ), but not with third ( wiin , plural wiinawaa ) .
Martha McGinnis
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bejar , Susana . 2003 . Phi-syntax: A theory of agreement. Ph.D. diss.,
University of Toronto.
Bonet , Eullia . 1991 . Morphology after syntax: Pronominal clitics in
Romance. Ph.D diss., Massachusetts Institute of Technology.
Dchaine , Rose-Marie , and Martina Wiltschko . 2002 . Decomposing
pronouns . Linguistic Inquiry 33 : 409 42.
Dixon , R. M. W . 1994 . Ergativity . Cambridge : Cambridge University
Press .
Farkas , Donka . 1990 . Two cases of underspecifcation in morphology .
Linguistic Inquiry 21 : 539 50.
Greenberg , Joseph . 1966 . Universals of Language . Te Hague : Mouton .
Halle , Morris . 1997 . Distributed morphology: Impoverishment
and fssion. In MITWPL 30: Papers at the Interface , ed. Benjamin
Person Philology and Hermeneutics
598
verbal tense and aspect . All these advances were authorita-
tively compiled by Dionysius Trax in his Treatise on Grammar
(ca. 100 b.c. ), which was so infuential that it was often called
simply Te Manual (and thereby probably subject to extensive
later revision by others). Some of this work, such as the eightfold
division of parts of speech and the treatment of Greek nominal
and verbal systems, still appears in twentieth-century textbooks.
In Rome, the Greek grammatical heritage was appropriated by
writers from Varro ( On the Latin Language , frst cent. b.c. , only
partially preserved) down to Priscian (ffth to sixth cent. a.d. ),
whose exhaustive Principles of Gramma r (ca. 500), fortuitously
designed to assist the Greek speakers of the longer-lived eastern
Roman empire, would become the ultimate authority for learn-
ing Latin throughout medieval Europe. Although (as is so often
the case) much Latin grammatical theory slavishly followed
Greek models, it was impossible to ignore obvious diferences
between the two languages (e.g., Latins lack of an article, one
past tense fewer, and one additional case ) .
Since the dominant unit of linguistic analysis of the time was
the word , and less so the sentence , the primary achievements
of classical language science lay in its descriptive and pragmatic
dimensions, particularly in linguistic pedagogy and the accurate
preservation, understanding, and annotation of written texts. For
instance, we have the Hellenistic era to thank for the invention of
such scholarly staples as footnotes, commentaries, critical edi-
tions, dictionaries, encyclopedias, and library catalogs. On the
other hand, investigations of phonetics and syntax , though
found in some early classical theorists, remained rudimentary.
And sadly, despite the story of King Mithridates of Pontus (or
Mithradates VI, 12063 b.c. ), who was fuent in all 22 languages
of his subjects, there was almost no formal ethnographic study
of the many other now-extinct languages of the Mediterranean
region; non-Greek speakers were simply barbarians ( bar-
baroi, babblers). Lexicographical work was driven by the need
to translate Greek and Latin, as well as to comprehend archaic
texts (e.g., Homer), and many word lists have been preserved as
hermeneumata translations and lists of glosses (glossaries,
from glossai, unfamiliar words). Although prodigious efort
from Socrates on was invested in etymology (the pursuit of a
words etymon , truth), this was almost a complete failure since
ancient philologists did not yet grasp how important phonol-
ogy and rules of sound change are for tracing the historical roots
of words. Te results ranged from the fanciful to the ridiculous.
Tus, Latin lignum wood hid potential ignis fre; lepus hare
was light-foot (compounding levis + pes ); and words could
stem from their opposites: bellum war was so named for being
not at all bellum beautiful. Much of this dubious heritage was
compiled by Isidore of Seville (sixth to seventh cent. a.d. ), whose
Etymologiae remained infuential throughout the medieval
period. Many such classical and medieval compilations remain
secondarily valuable, however, because they often preserve the
sole remaining fragments of hundreds of ancient texts .
As a time of consolidation and preservation of the Greco-
Roman heritage, the Middle Ages made relatively few signifcant
contributions to the study of language, as for many centuries
the Latin culture of Europe lagged behind the Greek learning of
the eastern Roman or Byzantine empire and the Arabic schol-
arship of Moorish Spain. Based on Varros lost writings on the
worlds oldest. Nevertheless, although both philology and her-
meneutics have Greek roots, neither was avowed as a primary
concern of leading classical philosophers such as Plato and
Aristotle. In classical Greek, the keyword logos signifed dis-
course in many diverse senses, including speech (both lan-
guage and oration), argument (a single proposition or an
entire line of reasoning), prose, story, history, reason,
and thought. Eventually, philologia , like philomatheia , would
imply studiousness, love of learning in general, since all learn-
ing at that time revolved around gaining written (and mathe-
matical) literacy, but Socrates could be called a philologos in the
more original sense of fond of speaking he famously refused
to write down his ideas. Plato, on the other hand, had fewer com-
punctions about writing. (In order to elevate written dialogue to
full-blown dialectic, Plato may himself have coined philosophia
philosophy as a more rigorous alternative.) Te frst classi-
cal fgure to embrace the title philologos was Eratosthenes, the
second Plato, who was one of the librarians of Alexandria and
a true philomath: He wrote on such diverse felds as geometry,
history, philosophy, poetry, and literary criticism .
In the classical era, hermeneia interpretation (sometimes
in the sense of translation) was a secondary philosophi-
cal concern, recalling the subsidiary status of the messenger
god Hermes (Roman Mercury). Today, readers of Aristotles
On Interpretation (probably not Aristotles title) may be disap-
pointed to fnd that this short treatise deals exclusively with the
logic of propositions . Similarly, Platos dialogue Cratylus is mired
in a shortsighted attempt to show that the names of things may
be both conventional and natural, as if individual letters could
somehow coherently imitate reality. (Socrates commitment to
sound symbolism is satirized in Aristophanes comedy Clouds. )
Nevertheless, the ancient world made great strides in one par-
ticular area, namely, grammar ( grammatike ), which ranged
from the teaching of literacy (including to non-native speakers),
to scholarly description and cataloging of word forms, to literary
and textual criticism, to more rarifed philosophical concerns.
Like philology , grammar could entail a very wide disciplinary
spectrum. In addition to the question of whether language was
a product of nature ( physis ) or convention ( nomos ), an equally
central and ultimately more fruitful debate among grammar-
ians revolved around whether language should be understood
in terms of analogy or anomaly: Analogia implied that lan-
guage was ultimately patterned and governed by regularity,
and anomalia that language was irreparably disorganized and
marred by exceptions. To analogy can be traced the systematicity
that still dominates language science (to say nothing of the leg-
acy of prescriptivist correctness in language use), and anomaly
can be thanked for introducing an honestly empirical dimension
to linguistic studies.
In the century after Plato, Stoic philosophers elevated the
study of language to a separate philosophical concern, but
their treatises have largely been lost. Under the Ptolemies, the
Hellenistic librarians of Alexandria refned and advanced all
earlier knowledge of language in their quest to amass, catalog,
and edit as many texts in as many felds of knowledge as pos-
sible. Tis included gathering descriptive word lists of various
Greek dialects, as well as making detailed analyses of orthogra-
phy, parts of speech (see word classes ), morphology , and
Philology and Hermeneutics
599
of languages. In 1440, Lorenzo Valla used historical-linguistic
evidence to demonstrate that Te Donation of Constantine , a
lucrative grant to the church, was a forgery, thereby founding the
forensic philology of diplomatics . Te Renaissance humanists
also revived the learning of Greek, along with Arabic and Hebrew
(considered the original human language), and in the wake of
Dante, various vernacular languages of Europe and even some
languages of foreign lands received grammars of their own. Te
languages of the world began to be surveyed, and Joseph Justus
Scaliger sharpened Dantes analysis of the language families of
Europe ( Diatriba de Europaeorum Linguis , 1599). Meanwhile, the
fundamentals of human thought explored by Ren Descartes and
John Locke also inspired such works as the Port-Royal General
and Rational Grammar (1660) and utopian attempts at invent-
ing universal communication systems, such as John Wilkinss
Essay towards a Real Character and a Philosophical Language
(1668). Te Italian rhetorician Giambattista Vico argued for what
he called Te New Science (1725; revised 1744) , an ambitious
philological recreation of the history of human mental and cul-
tural development via a succession of master tropes embodied
in ancient language, laws, and other social institutions. In short,
the Enlightenment brought a return to Eratosthenes multidis-
ciplinary philology: Te famous French Encyclopdie of Denis
Diderot and others (17512) cites philologie as a universal disci-
pline bridging the sciences and the humanities .
Te year 1768 is justly remembered as a watershed in the his-
tory of linguistics: It is the date of the famous paper of the legal
scholar William Jones to the Royal Asiatic Society in Calcutta.
Assigned as a colonial judge in the subcontinent, Jones had set
about learning Sanskrit, the ancient language in which Indias
religious and legal texts are preserved, much as Latin had done
for Europe. After only a few months of study, Joness brilliant
surmise was that certain obvious similarities among Sanskrit,
Latin, Greek, and other European languages implied a common
ancestor, which, crucially, might no longer survive. Such group-
ings had been noticed before, as by Dante and J. J. Scaliger , but
had been explained by the mechanisms of borrowing or decay,
rather than by the process of gradual and divergent evolution
from a now-dead proto-language. Te modern discipline of
historical and comparative linguistics had been born, and the
Enlightenments passionate but efete search for language ori-
gins was given a fresh scientifc direction: the problem of proto-
linguistic reconstruction.
Te year 1768 also marked the birth of Friedrich
Schleiermacher , so infuential in the feld of hermeneutics. Since
the Reformation, increasing philological concern had been
brought to bear on the text of the Bible. Tough a philological
monument in its own right, Jeromes Latin Vulgate (trans. ca.
380405) was no longer suf cient for the new commentaries and
vernacular translations desired by the Reformers who knew the
original Hebrew and Greek. Tis biblical hermeneutics would
develop into the infuential higher criticism, one of the trou-
bling scientifc advances that precipitated the Victorian crisis
of faith. Higher criticism described scripture not as an inspired
and inerrant document but as a layered tissue of competing
sources that had been edited together at some intermediate
time. Stemming from the patterns of stylistic diferences in bibli-
cal accounts (e.g., the varying names for God) frst noticed by
disciplines, Martianus Capellas Marriage of Philology and
Mercury (ffth cent. a.d. ) formalized the division of the seven lib-
eral arts (the lettered trivium of grammar, logic, and rhetoric ,
and the numeric quadrivium of geometry, arithmetic, music,
and astronomy), cementing the philological basis of Western
education for more than a millennium. Capella personifed phi-
lology as the mother of the liberal arts (from Latin ars , better
translated today as science), and the frst art was grammar, the
learning of literacy through the close study and imitation of clas-
sic texts. Te advent of Christianity did not entirely displace the
pagan past, but instead brought new urgency to the problem of
how to comprehend this legacy in the context of the new world-
view. One result was the famous multileveled system of allegory,
a hermeneutics that invited medieval thinkers to integrate three
competing cultural systems: the Hebrew Bible and the Jewish
religion; Greco-Roman mythology, literature, and history; and
orthodox Latin Christianity. Also known as typology , allegori-
cal interpretation was not limited to biblical texts but could be
extended to read types (emblems, characters) everywhere in
Gods creation, including the natural world (the second book
after the Bible).
Although it foreshadows modern linguistic procedures,
medieval allegory now seems as empty as classical etymologiz-
ing. It is not overly unfair to the philology of the Latin Middle
Ages to say that it is bracketed by its two greatest authors, its frst
and its last: Augustine and Dante. Certainly, there were import-
ant contributions to the understanding of language in between
these landmarks, such as the brilliant attempt at orthographical
reform via phonetic analysis by the so-called First Grammarian
of twelfth-century Iceland, but it seems typical that this work was
forgotten until the nineteenth century. And there were the scho-
lastic authors of speculative grammars (often under the rubric
of modi signifcandi , the means of signifying) who began the
ongoing search for universal principles in language. Yet long
before, around the ffth-century fall of Rome, Augustine bril-
liantly anticipated modern semiotics in On Christian Doctrine ,
and he was the frst ever to consider the problem of childhood
language acquisition in the autobiographical Confessions .
Augustine also helped Christianize Capellas seven pagan lib-
eral arts. Meanwhile, the Latin language itself was undergoing
change, and Augustine could no longer hear the vowel quantities
that underlaid Virgils poetic meter : the Romance languages
were slowly diferentiating across Europe. A millennium later,
Dante (who also took the theory and practice of medieval alle-
gory to new heights in his Divine Comedy and elsewhere) wrote
a milestone work on language entitled On the Eloquence of the
Vernacular (ca. 1305) . Tough necessarily and paradoxically
written in Latin, this unfnished treatise argued for the propriety
of using vernacular languages like Italian in literature, and is the
earliest mapping of European languages based on diferences
that seem to have evolved over time. It was the frst articulation
of the problem of language change .
Te fact that Johannes Gutenberg worked simultaneously
on printing his famous Bible alongside an edition of the still-
ubiquitous Latin grammar of Donatus (fourth cent. a.d. ) reveals
how the classical world still dominated the early Renaissance.
Soon, the rediscovery and promulgation of less-digested ancient
texts and ideas caused a surge in textual criticism and the study
Philology and Hermeneutics
600
Another great paradigm split was marked by the publication
of the one-time philologist Ferdinand de Saussures Course in
General Linguistics (1916) . Perhaps a victim of its own success,
diachronic philology, which so carefully traced the evolution of
parole, eventually yielded its disciplinary headship of language
study to Saussures synchronic langue (see synchrony and
diachonry and structuralism ).
Presently partitioned among various university disciplines,
philology and hermeneutics still govern the felds of medieval
and classical studies, historical linguistics, literary theory and
criticism, textual editing, lexicography, prosody and metrics,
and many others (see Cerquiglini [1989] 1999 ; Gumbrecht 2003 ).
Today, though the normal science of language emphasizes
such synchronic contexts as society, psychology, and the brain,
there is little doubt that philology and hermeneutics will persist
and reappear, like Hermes and Mercury, in many new guises in
the future .
Christopher M. Kuipers
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Boeckh , August . [1886] 1968. On Interpretation and Criticism . Ed. and
trans. John Paul Pritchard . Norman : University of Oklahoma Press .
Cerquiglini , Bernard . [1989] 1999. In Praise of the Variant: A Critical
History of Philology . Trans. Betsy Wing . Baltimore : Johns Hopkins
University Press .
Gumbrecht , Hans Ulrich . 2003 . Te Powers of Philology: Dynamics of
Textual Scholarship . Urbana : University of Illinois Press .
Mueller-Volmer , Kurt , ed. 1985. Te Hermeneutics Reader: Texts of the
German Tradition from the Enlightenment to the Present . New York :
Continuum .
Ormiston , Gayle L. , and Alan D. Shrift , eds. 1990. Te Hermeneutic
Tradition: From Ast to Ricoeur . Albany : State University of New York
Press .
Palmer , Richard E. 1969 . Hermeneutics: Interpretation Teory in
Schleiermacher, Dilthey, Heidegger, and Gadamer . Evanston,
IL : Northwestern University Press. Still the standard introduction .
Robins , R. H. 1997 . A Short History of Linguistics . 4th ed. London :
Longman .
PHONEME
Te phoneme is the smallest unit of speech that discriminates
one word from another in a particular language. Phonemes
are represented by symbols between slashes thus /p/ or /b/.
Phonemes may have alternate forms, called allophones. For
example, in English, the same phoneme /p/ is produced difer-
ently in p it and s p it . Minimal pairs are used to determine whether
two speech sounds are allophones or separate phonemes. For
example, in English, the phonemes /p/ and /b/ distinguish the
word p ull from b ull , and /t/ and /d/ distinguish the word ba t
from ba d .
Miwako Hisagi
PHONETICS
What Is Phonetics?
Phonetics is the area of language science research that studies
the articulation, acoustic properties, and auditory perception of
Reformation commentators, the documentary hypothesis sug-
gests that the canonical fve books of Moses (the Pentateuch)
are carefully patched together from a number of distinct source
texts .
Just as Dante had thought ft to apply sacred allegory to his
own secular literary production, so did Enlightenment students
of the Bible acknowledge that no special method of interpreta-
tion should be required for the word of God. As hermeneutic
theorist Johann August Ernesti put it in 1761, the verbal sense of
Scripture must be determined in the same way in which we ascer-
tain that of other books (quoted in Palmer 1969 , 38). Te paral-
lel development of secular higher criticism was also underway.
In his Introduction to the Correct Interpretation of Reasonable
Discourses and Books (1742), Johann Martin Chladensius became
the frst hermeneuticist to argue for the importance of point of
view ( Sehe-Punckt ) in interpreting historical texts. Similarly, the
classical scholar Friedrich August Wolf , who famously insisted
on taking his doctoral degree in philology, rather than philoso-
phy, published his Prolegomena to Homer (1795) , which asked the
still-vexed Homeric question: Was there really a single author
behind the Iliad and Odyssey ? Te concept of the linguistic fam-
ily tree of William Jones also found application in secular textual
editing, as Karl Lachmann (17931851) perfected the method of
stemmatics to posit nonextant archetypes from which various
groups of manuscripts descended and thus to help eliminate a
texts accumulated errors.
While Wolf and others developed Altertumswissenschaft
(classical scholarship) and biblical critics analyzed scripture,
Schleiermacher, who himself published on both classical and bib-
lical philology, elevated hermeneutics to a general practice that
would ultimately bring it far away from traditional philological
concerns. (Te primary source for Schleiermachers general her-
meneutics are detailed outlines he prepared for his university
lectures, notes partly published in 1819). Hermeneutics followed
this philosophical direction throughout the nineteenth cen-
tury; Wilhelm Dilthey , for example, located hermeneutics as the
supporting discipline for the universitys Geisteswissenshaften
(human sciences, literally sciences of the spirit). Following
the phenomenology of Edmund Husserl , the hermeneutic pro-
ject was furthered by Martin Heidegger and has continued down
to the present in a debate between Hans-Georg Gadamer and
Jrgen Habermas , with wider theoretical ripples still being felt in
the French and Anglo-American discourses of modernism and
postmodernism. General hermeneutics grew to be concerned
not only with interpretation per se but also with the very nature
of understanding, being, and reality itself. Today, hermeneut-
ics has grown more at home with the purer varieties of literary
theory and aesthetics than with traditional philologys lower
criticism.
Philology became an increasingly technical mode of histori-
cal linguistics during the nineteenth century. Comparative phi-
lologists such as Rasmus Rask , Jacob Grimm , and Franz Bopp
assembled exhaustive phonological and morphological data on
modern and ancient global languages in order to trace their devel-
opment and interrelationships (sometimes with a troublingly
Orientalist attitude; see Boeckh [1886] 1968 , 10, 44). Eventually,
philologys scientifc hypertrophy drove the foundation of sepa-
rate humanistic departments devoted to texts as literature.
Philology and Hermeneutics Phonetics
601
lawful, albeit complex, physical properties. A second puzzle that
has been much discussed in phonetic research is the puzzle of
lack of invariance . Tis refers to the dif culty of reconciling the
linguists view that language calls on a small fxed set of pho-
nological (or contrastive) units in organizing its words with the
observation that there are no invariant properties of these units
in the speech signal. Indeed, experiments using sinewave syn-
thesis have shown that even signals completely lacking nor-
mal speech cues can nevertheless be perceived as speech and
understood. One, but not the only, source of lack of segment-
ability and lack of invariance is the phenomenon of coarticula-
tion. Tis refers to the fact that neighboring speech sounds are, in
fact, articulatorily coproduced in time and thus interact with one
another and mutually shape the speech signal. Consequently,
phonological units in natural speech are realized in a highly vari-
able, context-dependent fashion.
Speech perception research in both children and adults
probes, in part, how human listeners are able to recover pho-
nological units from the speech signal (see, e.g., speech per-
ception in infants ) and engage in lexical access (word
identifcation; see word recognition, auditory ) (Pisoni
and Remez 2005 ). Tis involves understanding how listeners deal
with variability in phonetic form and how prior speech and lan-
guage experience shapes these processes. Investigation of these
puzzles has informed phoneticians theoretical views regarding
the fundamental nature of speech units .
Ways of Doing Phonetic Research
Tere are a number of areas of inquiry in the feld of phonetics,
and these generally fall under the purview of articulatory pho-
netics, acoustic phonetics, or speech perception. We encounter
some issues related to each of these areas in the following, but
frst, it is worthwhile to consider the two general approaches to
phonetic research. Te frst focuses on the description, classif-
cation, and transcription of speech sounds, the second on exper-
imental phonetics.
Traditionally, the frst approach was done by ear, thanks to
the carefully trained abilities of phoneticians, often trained in
a direct line of descent from one practitioner to another. Te
International Phonetic Association is a more than century-old
organization whose aim is to promote the scientifc study of
phonetics and its practical applications. Te association has
provided, with regular updates over the years, a consensus
International Phonetic Alphabet (referred to as the IPA, as is
the association itself) that serves as a notational standard for the
phonetic transcription of all sounds known to exist contrastively
in the worlds languages (and many noncontrastive variations of
these sounds) (IPA 1999 ). Te latest version of the IPA was pub-
lished in 2005 and is displayed in Figure 1 .
Tis transcription system is a standard reference in the feld
of phonetics and has been an important tool for description and
classifcation. Phoneticians doing work of this sort must deter-
mine what the linguistically relevant speech categories are that
is, what counts as linguistically the same and diferent and what
principled (or idiosyncratic) variation is observed among these
speech units. As can be seen from the IPA chart, phoneticians
have identifed important dimensions of variation, in particular,
for consonants:
speech units (see speech production, acoustic phonet-
ics, articulatory phonetics , and speech perception ,
respectively). More specifcally, phonetics can be understood
as linguistically informed speech science, and research phone-
ticians are generally trained linguists who bring to bear their
knowledge of the structural properties of language. Rather than
focusing only on one particular language or on universal ana-
tomical properties of hearing or articulation, a phonetician
has a special interest in understanding the full range of dis-
tinct possibilities in human speech or signed communication.
Because of the important role that linguistics plays in phonetic
study, most often phonetics fnds itself housed academically
as a linguistic discipline, though sometimes it fnds its home
in engineering, psychology, or in a language-specifc setting.
Te most prominent textbook used in educating phoneticians
is P. Ladefogeds A Course in Phonetics (2006) , now in its ffth
edition.
Within linguistics, phonetics is related to the feld of
phonology , another area of theoretical linguistic research.
Linguists vary in their opinions regarding the degree of dis-
tinctness and areas of overlap between the phenomena con-
sidered to be the objects of phonetic versus phonological
research. Both are concerned with the component speech units
or building blocks into which words can be divided. However,
the general view is that phonetics investigates measurable,
physical properties of these speech units, such as the precise
articulation of speech units, their detailed and contextually
dependent acoustic properties, and cross-linguistic variation
in these physical properties. Phonology, in contrast, is gener-
ally concerned with how these speech units are combined or
organized into acceptable word forms within a language (e.g.,
allowable sequences) and with the underlying principles of
organization shared across languages (see phonology, uni-
versals of ) . On analogy to chemistry, phonetics investigates
subatomic structure, and phonology studies the formation of
molecules out of basic atoms. Traditionally, the phonological
structure has been viewed as cognitive or grammatical, while
the phonetic structure has been viewed as purely physical and
implementational. However, the dividing line between cogni-
tive and physical has blurred or dissolved over the years (e.g.,
Browman and Goldstein 1995) .
Well-Known Theoretical Puzzles in Phonetics
Tere are a number of well-known puzzles in the area of pho-
netics whose empirical and theoretical consideration has helped
lead to our current understanding of some fundamental aspects
of the linguistic speech system. As one example, phoneticians
have an abiding interest in understanding how to reconcile a
linguistic view of speech as being composed of concatenated
symbolic units with its physical realization in articulation and
acoustics in which there are no silences, separations, or obvious
criteria for segmentation between these units. We can refer to
this puzzle as lack of segmentability . It is famously acknowledged
in Charles Francis Hocketts ( 1955 ) Easter egg analogy, which
describes the phonetic speech production processes as making a
smeared mess out of neat Easter eggs moving through a wringer.
Gradually, however, the feld has come to understand that rather
than a mess, the speech produced by humans is governed by
Phonetics
602
height [high-mid-low], related to the lowest resonant fre-
quency (the frst formant) of the vowel; and
backness [front-central-back], related to the distance between
the frst and second resonant frequencies (formants) of a
vowel.
Rounding or lip protrusion or compression is also encoded in
the symbol choice itself.
In addition, in order to adequately describe speech units, the
mechanism by which the air moves in the vocal tract must be
place in the vocal tract at which a consonant is articulated or
creates its constriction;
manner of articulation, which refers generally to the type of
constriction: complete closure for stops, narrow closure for
fricatives, constrictions having nasal or lateral airfow; and
voicing , whether the vocal folds are vibrating or not.
For vowels, the variations are captured in a continuous plane
whose dimensions can be identifed with auditory properties
called:
Figure 1. The IPA Chart. Reprinted with permission from the International Phonetic Association. Copyright 2005 by
International Phonetic Association.
Phonetics
603
nonlinguistic variables, such as age, gender, speaking rate and
style, afect, or language background, infuence these detailed
speech properties.
In experimental phonetics, the development of speech syn-
thesis played a critical role in researchers ability to design and
execute speech perception experiments by allowing for stimuli
with well-controlled acoustic properties. Tis ushered in a new
era of experimental speech perception research that examines
how humans utilize all of the myriad informational cues pres-
ent in the acoustic signal . Another particular body of experimen-
tal work called laboratory phonology seeks to inform questions
of linguistic representation and processes in phonology via
experimental phonetic data. Tis work generally takes a cogni-
tive science perspective and has been archived in the multivol-
ume Papers in Laboratory Phonology collection (arising from a
regular Conference on Laboratory Phonology, which has met
every other year since 1987). Browman and Goldstein ( 1991 ) and
Beckman and Edwards ( 1994 ) provide classic examples of this
type of phonetics .
Other Areas of Phonetic Inquiry
Other important areas of inquiry in the feld of phonetics include
investigation of the biomechanics or functional behavior and
coordination of the moving vocal tract (Saltzman and Munhall
1989 ; Guenther 1995 ), the role of audition and auditory processing
in speech communication (e.g., psychophysics of speech ),
and the vocal tract as a sound-producing device, often charac-
terized in terms of source-flter theory (Stevens 1998 ; Fant 1960 ).
Source-flter theory has provided a sophisticated mathematical
understanding of how noise sources at the larynx and along the
vocal tract are shaped by the geometry of the vocal tract and its
particular resonance properties to yield the output speech. Te
nonlinear properties of the articulatory-acoustic mapping have
been argued to be important in understanding constraints on
the sound inventories of languages (Stevens 1989 ). Other pho-
neticians focus on listener-oriented motivations, such as maxi-
mizing auditory distinctions in shaping sound systems, rather
than speaker-generated infuences. Clearly, speech systems are
adaptive to communicative and situational demands (Lindblom
1990 ). Speakerlistener interactions may give rise to change in
word forms over time, that is, diachronically (see phonology,
evolution of; syncrhony and diachrony ), and they may
give rise to synchronic adjustments specifc to the interlocutors
and the situation .
identifed. All languages use pulmonic sounds with air fowing
out from the lungs, but some languages also move air by laryn-
geal ( glottalic ) or tongue ( velaric ) maneuvers. Other important
linguistic properties of speech units can include distinctions
in the tone (i.e., placement in and/or movement through the
speakers pitch range), phonation type (i.e., the mode or quality
of vocal fold vibration and amount of laryngeal airfow), and VOT
( voice onset time: the temporal coordination of an oral constric-
tion with a laryngeal event) .
Much of the most important phonetic work of this sort has
been done in the feld by phoneticians working with native
speakers of languages (Ladefoged 2003 ), often languages that
are poorly documented or possibly endangered. Te most
authoritative description of the consonants and vowels of the
worlds languages can be found in Te Sounds of the Worlds
Languages (Ladefoged and Maddieson 1996 ). In addition to
describing the range of possible variation in the units used to
build human speech, phoneticians also address the question of
universal properties of human speech systems. Finally, descrip-
tive phonetics can also address variation within a language, such
as geographical dialect variation. Tis is one type of socio-
phonetics (other types include investigations of gender, age, or
class, for example.) For English, an impressive example of this
type of phonetic investigation can be found in the Atlas of North
American English (Labov, Ash, and Boberg 2006 ).
Whereas descriptive phonetics was traditionally done by ear,
a wide variety of current instrumental techniques is brought
to bear as well. Instrumental phonetics might utilize acoustic
analysis such as digitized waveforms, and spectrograms (see
Figure 2 ); pitch and formant tracking; articulatory analysis, such
as provided by laryngoscopy, palatography, magnetometry,
ultrasound, and MRI; and perceptual information such as that
provided by discrimination and categorization experiments and
even eye-tracking and neuroimaging .
Work on the other general type of phonetic research, exper-
imental phonetics, also utilizes a wide variety of instrumen-
tal approaches, but in this case, the data characterize human
behavior in the processes of producing and perceiving speech,
or refect quantitative rather than purely qualitative prop-
erties of speech. Experimental phonetics often investigates
how linguistic variables, such as segmental context, sylla-
ble structure, or prosody, infuence the detailed properties
of speech, such as its timing, articulation, spectral character-
istics, or intonation. Alternatively, it might examine how
5000
4000
3000
2000
6000
F
r
e
q
u
e
n
c
y
(
H
z
)
1000
Time (ms)
0 1450
Figure 2. A spectrogram of the sentence There are
no silences here. In a spectrogram, time is displayed
on the x-axis, frequency (in Hz) on the y-axis, and
amplitude in grayscale darkness.
Phonetics
604
understanding dyslexia . It is critical for educational success
that reading teachers are made aware of the importance of
characteristic diferences between speech and reading, of how
speech knowledge can be leveraged in the teaching of reading,
and of how interference from the phonetic properties of native
languages can infuence the acquisition of reading in non-native
languages (Rayner et al. 2002 ) .
A synergistic relationship exists between phonetics and the
feld of biomedical imaging. Advances in imaging of the vocal
tract and larynx have greatly illuminated our understanding of
speech production. In turn, new techniques for upper airway
and laryngeal imaging and image analysis have been developed
by phoneticians. Tese techniques can be incorporated into the
feld of clinical phonetics and speech pathology. Traditional
types of descriptive and instrumental phonetics have also
found utility in the understanding of clinical challenges such
as apraxia, stuttering, phonological disorders, and voice dis-
orders. Indeed, the National Institute on Deafness and Other
Communication Disorders is one of the largest funding sources
for phonetic research. Currently, there is enormous interest in
making cochlear implants as successful as possible for their user
populations. Knowledge of the acoustic properties of speech and
of methods for assessing perception adds to the broad body of
technological, engineering, and audiological knowledge cur-
rently contributing to this efort .
Phonetics is one of the foundational areas of linguistic research
and language science. It focuses on the descriptive, quantitative,
and behavioral aspects of speech production, transmission, and
perception. Phonetic knowledge helps guide our understanding
of the phonological representations and patterning observed in
human language. Phonetics also makes interdisciplinary con-
tact with speech technology, biomedical imaging, forensics, and
pedagogical and clinical felds .
Dani Byrd
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Beckman , Mary E. , and Jan Edwards . 1994 . Articulatory evidence for dif-
ferentiating stress categories. In Phonological Structure and Phonetic
Form: Papers in Laboratory Phonology. Vol 3. Ed. Patricia A. Keating ,
733. Cambridge : Cambridge University Press .
Browman , C. P. , and L. Goldstein . 1991 . Tiers in articulatory phonology,
with some implications for casual speech. In Papers in Laboratory
Phonology. Vol. 1: Between the Grammar and the Physics of Speech .
Ed. J. Kingston and M. E. Beckman , 34176. Cambridge : Cambridge
University Press .
. 1995 . Dynamics and articulatory phonology. In Mind as
Motion: Explorations in the Dynamics of Cognition , ed. Robert F. Port
and Timothy Van Gelder , 17593. Cambridge, MA : MIT Press .
Fant , Gunnar . 1960 . Acoustic Teory of Speech Production . Te
Hague : Mouton .
Guenther , F. H . 1995 . Speech sound acquisition, coarticulation, and rate
efects in a neural network model of speech production . Psychological
Review 102 : 594 621.
Hardcastle , William J. , and John Laver . 1997 , eds. Te Handbook of
Phonetic Sciences . Oxford : Blackwell .
Haskins Laboratories. A speech and reading laboratory in New Haven,
CT, that maintains a Web site at http://www.haskins.yale.edu.
Hockett , Charles Francis . 1955 . A Manual of Phonology .
Baltimore : Waverly.
Connections to Other Fields
Phonetics is an interdisciplinary area of linguistics; for exam-
ple, we have sketched its connection to phonology. It also can
closely tie into other areas of experimental linguistics, particu-
larly psycholinguistic research on spoken language pro-
duction and processing and neurolinguistic research on brain
function. Knowledge of the phonetic properties of languages
and of the characteristics of the speech signal is critical to the
design of experimental linguistic and neurolinguistic research
programs that examine speech production and processing (see
phonetics and phonology, neurobiology of; brain
and language ). Such speech-related work might address lexi-
cal access, speech production planning, neural localization of
various functions related to speaking and speech understanding
(see lexical processing, neurobiology of ), the integra-
tion of visuofacial and auditory information, and the relation of
action and perception (see mirror systems, imitation, and
language ) .
Furthermore, there are many felds outside of linguistics
on which phonetics has a direct bearing. In the area of speech
technology, linguistic phonetic knowledge can contribute
sometimes directly, sometimes indirectly to machine speech
synthesis and recognition (see voice interaction design ).
And conversely, much early work in acoustic phonetics grew
out of the eforts of speech engineers, for example, at Bell
Laboratories, Haskins Laboratories, Massachusetts Institute
of Technology, the Joint Speech Research Unit in England, the
Speech Transmission Laboratory in Sweden, and the Advanced
Telecommunications Research (ATR) Institute International
in Japan. Currently, in the feld of speech engineering, there is
interest in capturing linguistic knowledge in ways that will allow
better system performance with conversational interfaces and
with audiovisual speech .
Phonetic science also has utility in forensics, and forensic
phonetics is a recognized area of applied science (see forensic
linguistics ). Forensic experts bring both instrumental and
expertlistening techniques to the determination of whether a
suspects voice is a likely or unlikely match to forensic evidence
that investigators have in hand. It should be noted, however, that
there is no unique identifer in the voice of an individual that is
analogous to a fngerprint. Phoneticians are frequently called on
in such speaker-identifcation cases to provide expert knowledge
and testimony as to the many subtle properties that may distin-
guish one individuals speech from that of another .
Another feld outside of linguistics is often, in fact, an indi-
viduals frst contact with phonetics second language pedagogy
(see bilingual education ). An accurate understanding of
how a languages speech sounds are articulated proves helpful
in pronunciation instruction. Instrumental techniques for dis-
playing feedback on articulation, speech acoustics, or linguistic
categorization can also help in training production and percep-
tion of non-native linguistic contrasts .
Te paramount area of the infuence of theoretical phonol-
ogy and phonetics on pedagogy is in the teaching of reading
(see teaching reading ) . Linguists from diverse backgrounds
and groups have taken a leadership position in emphasizing
the importance of phonemic awareness (see phonologi-
cal awareness ) for the acquisition of reading skills and for
Phonetics
605
the microlevel are necessary for understanding how neurobio-
logical methods are used to examine phonetics and phonology
(see Kandel, Schwartz, and Jessell 2000 ; Shafer and Garrido-
Nag 2007 , for greater detail).
First, brain function is in terms of electrochemical mes-
sages between neurons. neuroimaging methods index dif-
ferent aspects of these processes and the metabolic processes
that support these. Electrophysiological methods (electroen-
cephalogram [EEG], magnetoencephalogram [MEG]) record
changes in electrical potential at the scalp. Tese changes are the
result of the synchronous fring of large assemblies of neurons.
Functional magnetic resonance imaging (fMRI) and positron
emission tomography (PET) measure changes in the metabo-
lism of oxygen, and PET can also measure changes in the chemi-
cal aspect of the electrochemical signals sent between neurons.
Tese changes in electrochemical and metabolic measures are
used to make inferences about timing and localization of neural
activity related to some stimulus or event .
A second point is that diferent brain regions have distinctive
structure in terms of neurons and connectivity and that these
distinctions are the basis of Korbinian Brodmanns classifcation
system. For example, primary auditory cortex (Brodmanns area
[BA] 41) has a thick layer of neurons specialized to receive infor-
mation from the peripheral auditory system. Tese neurons then
send signals to other cortical regions but not directly back to the
periphery for motor responses. Ultimately, phonological func-
tioning will need to be described in terms of connectivity at this
neural level for a complete understanding of the brainbehavior
relationship.
At the macrolevel, neurobiology of phonetics/phonology is
described in terms of the activated brain regions and the timing
of activation of these regions in perception or production (see
speech perception and speech production ). Tese brain
regions are referred to by Brodmanns areas, by names describing
function (e.g., primary auditory cortex), by the scientist involved
in identifying the regions (e.g., brocas area ), or by some term
describing an attribute of the regions (e.g., Greek hippocampus
for a region that is shaped like a seahorse) .
Te principal brain structures involved in phonetic/phono-
logical perception are found in the perisylvian cortex and
include primary (BA 41) and secondary (BA 42) auditory cortex
for processing the acoustic-phonetic aspects of speech (Scott and
Wise 2004 ) (see Color Plate 10). Sound in general (e.g., noise)
activates bilateral regions of the dorsal plane of the superior
temporal gyrus (STG) and regions of the lateral STG. In contrast
with noise, temporally complex signals, including speech, more
strongly activate the dorsal region of STG, and the lateral STG
activation extends more ventrally. Auditory information identi-
fed as speech compared to non-speech leads to increased acti-
vation of regions of the STG and superior temporal sulcus (STS)
that are more anterior and ventral (inferior). Te left STS appears
to be active in mapping speech onto lexical- semantic repre-
sentation. In contrast, the right STS shows sensitivity to melodic
features. Te left planum temporale (PT, in superior posterior
temporal cortex) is believed to have a special role in phonetic/
phonological processing and appears to support a motor/sen-
sory interface for acoustic information. A left-greater-than-right
asymmetry is generally stronger for speech than non-speech (see
IPA (International Phonetic Association). 1999. Handbook of the
International Phonetic Association: A Guide to the Use of the
International Phonetic Alphabet . Cambridge : Cambridge University
Press . Te association maintains a Web site at http://www.arts.gla.
ac.uk/IPA/ipa.html.
Johnson , Keith . 2003 . Acoustic and Auditory Phonetics . 2d ed.
Oxford : Blackwell .
Labov , W. , Ash , S ., and C. Boberg . 2006 . Atlas of North American
English: Phonetics, Phonology and Sound Change . Berlin : Walter de
Gruyter . Available online at: http://www.langsci.ucl.ac.uk/ipa/.
Ladefoged , P. 2003 . Phonetic Data Analysis: An Introduction to
Instrumental Phonetic Fieldwork . Oxford : Blackwell .
. 2006 . A Course in Phonetics , 5th ed. Boston : Tomson
Wadsworth.
Ladefoged , Peter , and Ian Maddieson . 1996. Te Sounds of the Worlds
Languages . Oxford : Blackwell .
Lindblom , B. 1990 . Explaining phonetic variation: A sketch of the H&H
theory. In Speech Production and Speech Modeling , ed. W. Hardcastle
and A. Marchal , 40339, Dordrecht, the Netherlands : Kluwer .
Miller , J. , ed. 1991. Papers in Speech Communication . A three-volume ser-
ies published by the Acoustical Society of America (New York) through
the American Institute of Physics.
Pisoni , D. , and R. Remez , eds. 2005. Handbook of Speech Perception .
Malden, MA : Blackwell .
Rayner , K. , B. R. Foorman , C. A. Perfetti , D. Pesetsky , and M. Seidenberg.
2002 . How should reading be taught? Scientifc American
286 : 84 91.
Saltzman , E. L. , and K. G. Munhall . 1989 . A dynamical approach to
gestural patterning in speech production. Ecological Psychology
1 : 333 82.
Stevens , K. 1989 . On the quantal nature of speech . Journal of Phonetics
17 : 3 45
Stevens , K. 1998 . Acoustic Phonetics . Cambridge, MA : MIT Press .
PHONETICS AND PHONOLOGY, NEUROBIOLOGY OF
Te study of the neurobiology of phonetics and phonol-
ogy focuses on the brain mechanisms that support percep-
tion and production of linguistic phonological forms. Tis entry
describes the neural structures and processes underlying pho-
netic and phonological processing and briefy discusses four
current theoretical controversies that neurophysiological data
can help address. First, are there invariant relationships between
acoustic properties and phonological categories? Second,
does speech have some special status apart from other acoustic
information? Tird, is there a critical period for language-
specifc learning? Fourth, to what degree does biology constrain
the nature of phonological systems?
Neurobiological Underpinnings
Te physiology (or function) of phonological processing is
described in terms of the structures (anatomy) activated in
processing and the function of these structures. A compre-
hensive understanding of the physiology requires explication
at the micro- and macrolevels of processing. Te microlevel
describes the microstructures and their processing (neuron,
axon, synaptic potential) and are general to brain function,
whereas the macrolevel focuses on larger-scale structures and
processes specifc to a particular motor, sensory, or cognitive
process (e.g., phonetic processing). Several points concerning
Phonetics and Phonology, Neurobiology of
606
appear to index acoustic levels of processing. To date, there is
no clear evidence that language experience at the phonological
level directly afects processing in the time range of the P1 and
N1 components .
ERP components occurring later in time are related to higher-
level cognitive processes. Tose showing modulation by phono-
logical experience include mismatch negativity (MMN), N2b,
P3b, and N400 (Ntnen 2001 ; Kujala et al. 2004 ). Listeners
show more robust MMNs (peaking between 100 and 300 ms and
indicating preattentive, automatic processing) in discriminating
pairs of sounds with which they have had experience (Ntnen
2001 ). Specifcally, the MMN is smaller or later to a contrast in
speech sounds if the speech sounds are assimilated into one pho-
nological category for listeners (e.g., Japanese listeners percep-
tion of English [l] vs. [r]) or if the speech sounds are assimilated
into two categories, but one or both sounds are poor exemplars
of these categories (e.g., English listeners perception of Hindi
retrofex [Da] versus [ba]; Shafer, Schwartz, and Kurtzberg 2004 ).
Te later components, N2b, P3b, and N400, are observed when
a participant is asked to actively discriminate a speech contrast.
No discernible N2b, P3b, or N400 is observed if discrimination
is very dif cult (chance performance). If discrimination is bet-
ter than chance but more dif cult than for native listeners, then
these components are later and larger than those found for the
native group. For example, English speakers showed reasonably
good discrimination of Japanese (JP) vowel duration (taado
versus tado), but a later and larger P3b component compared
to native Japanese listeners (Hisagi 2007 ).
Integrating the knowledge of location obtained from fMRI/
PET and timing obtained from EEG/MEG indicates that acoustic-
phonetic processing occurs in primary and secondary auditory
cortical regions between 10 and 100 ms, followed by phonolog-
ical aspects of processing, presumably in more ventral regions,
between 100 and 400 ms. Tis model is supported by studies
localizing the sources of N1 and the phonologically elicited N400
(Kujala et al. 2004 ) .
Lack of Invariance Problem
A major theoretical debate in speech perception over the past 40
years has been the relationship between acoustic and phonolog-
ical properties. Speech with similar acoustic properties may be
assigned to diferent phonological categories, and, conversely,
speech with diferent acoustic properties is sometimes assigned
to the same phonological category. Much research focused on
discovering invariant properties of speech sound categories that
would allow for precise categorization has failed to do so.
A recent model can be used to illustrate how neurophysi-
ological data can address the lack of invariance issue. In this
model, speech is categorized and identifed by an active pro-
cess of hypothesis testing (e.g., Magnuson and Nusbaum 2007 ).
Diferent types of information are used with regards to the type
and amount of sensory and lexical information available. For
example, clear auditory-speech information and knowledge
of the possible phoneme categories of a language lead to reli-
ance on auditory information in categorization. More ambigu-
ous auditory-speech information can lead to greater reliance on
visual information (e.g., lip closure for [p] but not [t]). In other
words, there are many routes to phonological categorization.
left hemisphere and right hemisphere ). Anterior regions
are also activated in speech perception. Te left prefrontal cor-
tex (BA 46) is activated in phonological processing in access-
ing, sequencing, and monitoring phonemes and processing
transitions from consonants to vowels or vowels to consonants.
Articulation of phonetic information is supported by motor (BA
4) and premotor/Brocas area cortex (BA 6, BA 44/BA 45) .
Recent models have organized these observations into a sim-
ple framework in which the more dorsal regions (i.e., posterior
and superior) are active in auditory-motor integration during
speech perception and the more ventral regions (anterior and
inferior) are more involved in the speech-meaning interface.
Tis indicates that the phonetic aspects of processing, which
are independent of meaning, will be carried out in more dorsal
regions of the auditory and motor cortex, whereas the phonolog-
ical aspects, which are the basis of meaningful distinctions, are
processed in more ventral areas of the auditory cortex. Te exact
roles of STG, STS, and the two hemispheres in phonetic and pho-
nological processing have not been defnitively established yet,
but it is known that these areas are all important in speech pro-
cessing (Poeppel and Hickok 2004 ).
Anterior and posterior brain regions involved in phonetic
and phonological processing communicate directly via bundles
of fbers (axons), such as the arcuate fasciculus, but also via more
indirect routes, including the basal ganglia , thalamus , and
cerebellum . Tese additional structures are involved in gen-
eral functions related to information processing, motor plan-
ning, and coordination and will not be discussed further here.
Te timing of activation of levels of phonetic and phonological
processing has largely been provided by EEG and MEG mea-
sures. Te timing of auditory processes can be roughly related
to levels of processing in the primary and secondary auditory
cortex and to the timing of more basic (e.g., signal detection)
versus higher-level cognitive processes (phonological discrimin-
ation). Te principal method used to investigate these processes
is event-related potentials (ERPs). Te EEG/MEG is time-locked
to a stimulus of interest (e.g., ba), and this stimulus is deliv-
ered multiple times (anywhere from 20 to 10,000, depending on
the ERP component of interest). Te portion of the EEG/MEG
time-locked to the stimulus is averaged to remove noise (i.e.,
activity produced by unrelated processes).
ERPs are described in terms of the latency, polarity, and
topography of peaks that vary with some stimulus property or
cognitive process. Tese identifed peaks are often referred to as
components. Te P and N in a component label refer to positive
and negative polarity, respectively, and the number indicates
the approximate peak latency (e.g., N400) or the position in a
sequence (e.g., N2).
Studies of auditory processing have shown that auditory
information enters primary cortical regions between 10 and 50
ms following contact with the outer ear and that a frontocentral
positivity peaking around 50 ms (P1 component) and negativ-
ity peaking around 100 ms (N1 component) index activity in the
primary and secondary auditory cortex. Neurobiological studies
with animals suggest that P1 indexes input from the periphery
into the superior temporal plane of the auditory cortex and that
N1 refects activity of neurons in the secondary auditory cortex
receiving information from other cortical regions. P1 and N1
Phonetics and Phonology, Neurobiology of
607
Alternatively, listeners may have dif culty refocusing their atten-
tion to the relevant cues needed for rapid processing of the sec-
ond language (Strange and Shafer 2008 ).
Neurophysiological data can address this question by
examining where in the nervous system diferences in process-
ing are found for frst and second language learners. Te cur-
rent research has not shown diferences earlier than the MMN
response. Furthermore, a recent study from our laboratory sug-
gests that attention plays a role in loss of ability to learn novel
categories. Specifcally, listeners learn to automatically attend to
relevant cues in their frst language and can only overcome these
weightings with great attentional efort. Tis result suggests that
the loss of sensitivity in adjusting to novel phonological catego-
ries by second language learners is not directly due to a closure
of a critical period for changing the sensitivity or resolution of the
primary and auditory cortex; rather, it is due, at least in part, to
attentional issues (Hisagi 2007 ).
Tese fndings do not answer all the questions regarding
critical and sensitive periods for setting up phonetic and pho-
nological categories since second language learners acquired
categories for a frst language early in life. Recent research exam-
ining the neurophysiological and behavioral consequences of
deprivation of hearing, which is reversed by cochlear implants,
will have much to contribute for addressing this question. Recent
advances have led to implantation at earlier ages, which is allow-
ing researchers to compare the quality of phonological process-
ing across diferent ages of frst exposure to speech information.
Improvements in these implanted devices will also allow exami-
nation of how the quality of auditory-speech input impacts pho-
netic and phonological systems. Tis emerging area of research
is likely to provide less ambiguous evidence regarding a critical
or sensitive period for speech .
To What Degree Does Biology Constrain the Nature of
Phonological Systems?
A classic debate in linguistics concerns the extent to which lan-
guage is innate . A more useful way to ask this question is what
biological constraints are placed on the nature of phonological
systems and how environmental input contributes to construct-
ing these systems. Across languages, there are common patterns.
For example, all languages contrast /i/ (heep), /u/ (hoop), and
/a/ (hop) (although there can be slight variations in the actual
production of these sounds), and some languages only contrast
these three vowels. However, there is no existing language that
only contrasts i (in bead), I (in bid), and e (in bed) without
also contrasting /i/, /u/, and /a/. It is possible that these univer-
sal patterns are due to biological constraints. On the other hand,
they may be attributed to environmental factors. Examination of
the evidence suggests that the system is constrained by an inter-
action of biological and environmental constraints. For example,
/i/, /u/, and /a/ are perceptually more distinct than i (in bead),
I (in bid), and e (in bed), and this is a property of the auditory
system; the environment (input) leads to less salient distinctions
included in some languages, but many possible distinctions are
never found in languages.
Neurobiological data will aid in further elucidating how devel-
opment of phonological systems is constrained by instructions
from the genetic code and emerges from patterns in the input. In
If this model is viable, then neurophysiological data will show
whether diferent sensory and motor cortex are activated when
speech is more versus less clear and when other information
(e.g., visual) is available. Several recent studies have shown more
involvement of the motor cortex and visual sensory areas for
ambiguous acoustic speech information when facial informa-
tion is available, and less activation of these regions when only
the speech signal is available (e.g., Skipper, Nusbaum, and Small
2006 ).
In summary, this example illustrates the importance of
neurophysiological data for addressing long-standing theoret-
ical controversies .
Does Speech Have Some Special Status Apart from Other
Acoustic Information?
Over the past forty years, there has been a debate regarding
whether speech requires a special type of auditory processing
specifc to humans. Behavioral studies have delivered mixed
answers to this question. For example, studies have shown that
speech (in particular, consonants) is perceived categorically,
rather than continuously, and used this to argue for special sta-
tus. On the other hand, other species (e.g., chinchillas) are shown
to perceive speech categorically, and complex non-speech audi-
tory sounds can be categorically perceived.
Neurophysiological data can help examine this question by
determining whether the same structures and processes sup-
port processing of speech and non-speech. Te current available
data suggest that in one sense, speech and non-speech are simi-
lar. Te same auditory cortical regions are activated in process-
ing speech and non-speech, as described previously (also see
Dehaene-Lambertz and Gigla 2004 ). Furthermore, the sensory-
motor links found for speech are similar to those seen for other
sensory-motor links (e.g., tool manipulation using visual and
motor regions) and seen in other species (see Skipper, Nusbaum,
and Small 2006 ) .
In another sense, the neurophysiological data suggest
that the processing of speech difers from that of non-speech.
Specifcally, as described previously, more ventral areas (lateral
and anterior superior temporal gyrus) become involved in pho-
nological processing of speech sounds because these sounds
are relevant for making meaning contrasts. It is possible that
humans are the only species that fractionate sound symbols into
subcomponents (phonemes) that can be manipulated to create
novel symbols, and, in this way, speech is special .
Is There a Critical Period for Language-Specifc Learning?
Researchers have long noted that learning a second language
late in life typically results in a stronger non-native accent and
poorer speech perception in the second language (see Strange
and Shafer 8 ; see also second language acquisition ). One
explanation for this pattern is that there is a critical or sensi-
tive period in which phonological information must be learned
in order to lead to native-like performance (see phonology,
acquisition of ). Some research suggests a gradual loss of abil-
ity to alter phonological categories up to puberty.
Te reason for this change in ability is unknown. It could be
that the auditory cortex is altered at an early level so that it loses
the ability (or resolution) to respond to non-native contrasts.
Phonetics and Phonology, Neurobiology of
608
Skipper , Jeremy I. , Howard C. Nusbaum , and Steven L. Small . 2006 .
Lending a helping hand to hearing: Another motor theory of speech
perception. In Action to Language via the Mirror Neuron System , ed.
Michael A. Arbib , 25084. Cambridge : Cambridge University Press .
Strange , W. , and V. L. Shafer. 2008 . Speech perception in second language
learners: Te re-education of selective perception. In Phonology and
Second Language Acquisition , ed. M. Zampini and J. Hansen , 15392.
Cambridge : Cambridge University Press .
PHONOLOGICAL AWARENESS
P honological awareness encompasses the broad class of abili-
ties that enable one to attend to, isolate, identify, and manipulate
the speech sounds in spoken words. Te domain of phonologi-
cal awareness abilities can be subdivided into two levels. Te
frst, phonological sensitivity, pertains to conscious aware-
ness of larger, more salient sound structures within words ,
including rhymes and syllable structures (i.e., syllables and
subsyllabic units) (Scarborough and Brady 2002 ). (Rhymes,
defned at the word level, consist of the stressed vowel and
what follows [e.g., b e /w e ; f eather /w eather ]; subsyllabic units
include onsets, i.e., the portion of each syllable preceding the
vowel [e.g., b e; sp ot; m ag n et], and rimes, i.e., the remaining por-
tion [e.g., b e ; sp ot ; m ag n et ]). Te second level of phonological
awareness, phoneme awareness, refers to explicit awareness of
the individual phonemes making up words. Generally, children
acquire at least some degree of phonological sensitivity prior
to phoneme awareness (see phonology, acquisition of ).
However, questions remain as to whether attainment of pho-
nological sensitivity is a necessary prerequisite for the devel-
opment of phoneme awareness (Gillon 2005 ). When children
begin to acquire phoneme awareness, they usually frst are able
to isolate and identify the external phonemes (i.e., the begin-
ning and/or fnal phonemes in words). Ultimately, profciency
in phoneme awareness entails the ability to segment, identify,
and blend all of the individual phonemes, including those
within consonant clusters (e.g., in words such as blast).
Te signifcance of phoneme awareness stems from its role
in reading acquisition (see writing and reading, acqui-
sition of ). Understanding that spoken words are made up of
individual speech sounds provides a conceptual foundation for
understanding the alphabetic principle (i.e., that letters cor-
respond with phonemes). Tis awareness, in turn, facilitates
learning to read and spell . Te relationship between phoneme
awareness and literacy development is reciprocal: With
some emergent awareness of phonemes, the student can start
to acquire lettersound knowledge. In turn, awareness of pho-
nemes is heightened by experience with print.
Since the concept of phoneme awareness was established in
the 1970s (e.g., Liberman 1971 ), evidence for the signifcance of
phoneme awareness for reading achievement has accrued from
correlational, prediction, and training studies. At all ages, includ-
ing adulthood, less-skilled readers demonstrate weaker perfor-
mance on phoneme awareness measures than better-reading
peers, whether the same age or younger reading-age controls.
Prediction studies with kindergarten students document that
phoneme awareness performance is one of the strongest pre-
dictors of their subsequent reading achievement, particularly
for decoding and word recognition skills, but also for reading
particular, examination of the way that genetic variation afects
the development of speech processing and its neurophysiologi-
cal substrate will help us understand the contributions of biology
and the environment. For example, studies of congenitally deaf
populations have revealed that some brain regions that are typi-
cally specialized for audition (e.g., regions of secondary auditory
cortex) are used in visual processing and thus are highly sensi-
tive to input. In contrast, more primary regions specialized for
audition (primary auditory cortex and subcortical areas) do not
reorganize to take on nonauditory functions and are thus less
sensitive to input. An understanding of the relationship among
the genetic code, neural connectivity, and plasticity of auditory
and language-association brain regions will help to create real-
istic models of phonetic/phonological development and pro-
cessing, which in turn will help to answer how biology and the
environment contribute to the development of this system .
Conclusion
Tis entry illustrated the importance of neurobiological data in
addressing signifcant questions concerning phonetic and pho-
nological processing. In particular, an understanding of the neu-
robiology supporting phonetic and phonological processing will
allow researchers to construct better models of processing and to
address questions related to frst and second language learning
and disorders (such as dyslexia and aphasia ) attributable to
defcits in phonological processing .
Valerie Shafer
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Dehaene-Lambertz , G. , and T. Gigla . 2004 . Common neural basis for
phoneme processing in infants and adults . Journal of Cognitive
Neuroscience 16 : 1375 87.
Hisagi , M. 2007 . Perception of Japanese temporally-cued phonetic con-
trasts by Japanese and American English listeners: Behavioral and elec-
trophysiological measures. Ph.D. diss., City University of New York.
Kandel , E. , J. Schwartz , and T. Jessell . 2000 . Principles of Neural Science .
New York : William Heinemann and Harvard University Press .
Kujala , A. , K. Alho , E. Service , R. J. Ilmoniemi , and J. F. Connolly . 2004 .
Activation in the anterior left auditory cortex associated with pho-
nological analysis of speech input: Localization of the phonological
mismatch negativity response with MEG . Cognitive Brain Research
21 : 106 13.
Magnuson , J. S. , and H. C. Nusbaum . 2007 . Acoustic diferences, lis-
tener expectations, and the perceptual accommodation of talker vari-
ability . Journal of Experimental Psychology, Human Perception and
Performance 33 : 391 409.
Ntnen , Risto 2001 . Te perception of speech sounds by the human
brain as refected by the mismatch negativity (MMN) and its magnetic
equivalent (MMNm). Psychophysiology 38 : 1 21.
Poeppel , David , and Gregory Hickok . 2004 . Towards a new functional
anatomy of language. Cognition 92 : 1 12.
Scott , S. , and R. Wise . 2004 . Te functional neuroanatomy of prelexical
processing in speech perception . Cognition 92 : 13 45.
Shafer , V. L ., and K. Garrido-Nag . 2007 . Te neurodevelopmental bases
of language. In Te Handbook of Language Development , ed . M. Shatz
and E. Hof , 2145. Oxford : Blackwel l.
Shafer , V. L. , R. G. Schwartz , and D. Kurtzberg . 2004 , Language-specifc
memory traces of consonants in the brain . Cognitive Brain Research
18 : 242 54.
Phonetics and Phonology, Neurobiology of Phonological Awareness
609
considerable diversity among these views, it is fair to say that
by and large, they focused on the elucidation of the contrastive
properties of elements of surface phonetic form to the exclusion
of other aspects of sound structure.
The Development of Modern Phonology
Poststructuralist theories fall broadly within the tradition of gen-
erative phonology, associated in its origins with Noam Chomsky
and Morris Halle ( 1968 ). Te distinguishing character of this
view was its attention not simply to surface contrasts but also to
patterns of alternation in shape, and its positing of an abstract
underlying representation (where contrasts among elements
are characterized) that is related to surface phonetic form by a
system of rewriting rules. Each of these rules represents a single
generalization about the realization of phonological elements
(e.g., Vowels are long before voiced obstruents). Much of the
theoretical discussion in the 1960s and early 1970s concerned
the role of an explicit formalism for these rules.
Te rules were presumed to apply in a sequence, with each
applying to the result of all previous rules. As a consequence,
some of the generalizations represented by individual rules may
only be valid at an abstract level and not true of all surface forms
to the extent subsequent changes obscure the conditioning fac-
tors of a rule or its efects, leading to the opacity of the rule in
question. For example, in many varieties of American English,
the medial consonants of words like ladder and latter are both
pronounced as the same voiced fap [D]. Te vowels of the ini-
tial syllables of such words continue to difer in length, however,
refecting the abstract diference in voicing between /d/ and /t/,
even though that diference is obscured by the (subsequent)
application of a rule of fapping that renders the vowel-length
rule opaque. Much attention was paid in this period to the theo-
ries of rule ordering necessary for describing such phenomena .
In the years immediately following the publication of Chomsky
and Halle ( 1968 ), a number of scholars reacted strongly to the
perceived abstractness of the underlying phonological repre-
sentations to which it appeared to lead. Various proposals that
intended to restrain this aspect of the theory appeared, some of
them based on the idea that if the rules themselves could be con-
strained so as to permit only highly natural ones, drawn from
some substantively constrained universal set, the underlying
representations would thereby be forced to be closer to surface
forms. Others proposed to constrain the relation between pho-
nological and phonetic representation directly (again, often in
the name of naturalness).
In general, these attempts to limit the power of phonological
systems by fat ran into apparent counterexamples that deprived
them of their appeal. Other developments in phonological the-
orizing shifted scholars attention away from this issue while also
leading (as somewhat unintentional by-products) to a general
reduction in the degree of abstractness of representation. Some
of these elaborations and reorientations of the program of gen-
erative phonology are sketched here.
AUTOSEGMENTAL PHONOLOGY. Te bulk of research during the
classical period of generative phonology was concerned with
segmental phenomena (although the main goal of Chomsky
and Halle 1968 was an account of English stress ). In the early
comprehension (see teaching reading ) . Most compelling,
intervention studies confrm a causal link between instruction in
phoneme awareness and increased success at learning to read,
with greater benefts when discovery of phonemes is linked with
letter knowledge (Ehri et al. 2001 ) .
Susan A. Brady
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ehri , L. C. , S. R. Nunes , D. M. Willows , B. Schuster , Z. Yaghoub-Zadeh ,
and T. Shanahan . 2001 . Phonemic awareness instruction helps
children learn to read: Evidence from the National Reading Panels
meta-analysis . Reading Research Quarterly 36 .3: 250 87.
Gillon , G . 2005 . Phonological Awareness: From Research to Practice .
New York : Guilford .
Liberman , I. Y . 1971 . Basic research in speech and lateralization of
language: Some implications for reading disability . Bulletin of the
Orton Society 21 : 71 87.
Scarborough , H. S ., and S. A. Brady . 2002 . Toward a common terminol-
ogy for talking about speech and reading: A glossary of the phon words
and some related terms . Journal of Literacy Research 34 : 299 334.
PHONOLOGY
As opposed to phonetics , which deals with the properties
of sounds from a language-independent point of view, pho-
nology constitutes the study of the sound structure of units
( morphemes, words , phrases, utterances) within individual
languages. Its goal is to elucidate the system of distinctions in
sound that diferentiate such units within a particular language,
and the range of realizations of a given units sound structure as
a function of the shape of other units in its context. Tese two
goals the study of invariants of sound structure and of the vari-
ation shown by these elements in combination are obviously
closely related, but attention has tended to shift between them
over time .
Late nineteenth- and early twentieth-century study of sound
structure focused on the details of sound production. As these
studies (in both articulatory phonetics and acoustic
phonetics ) became more sophisticated, however, it was
increasingly apparent that the resulting explosion of data about
sound properties was obscuring, rather than enhancing, schol-
ars understanding of the way sound is organized for linguistic
purposes. Much that is measurable in the speech signal is pre-
dictable, internal to the system of a given language, even though
exactly comparable properties may serve to distinguish items
from one another in a diferent language.
Vowels in English, for example, are relatively longer before
certain consonants than before others, but the diference in the
vowels of, for example, cod and cot is entirely predictable from
this principle alone. By contrast, an exactly parallel diference
between the vowels of kaade dip and kade envious in Finnish
serves as the sole diference between these words. A focus on
phonetic features alone fails to reveal the role played by sound
properties within a language.
Te result of this insight was the development within vari-
ous theories of structuralism of attempts (Anderson 1985 )
to defne the phoneme, a presumed minimal unit of contrast
within the sound system of a single language. While there is
Phonology
610
of features) were thereby characterized as similar to one another,
and thus predicted to behave in the same way in rules.
It soon became apparent, however, that feature analysis
by itself does not exhaust this matter. When nasal consonants
assimilate in place of articulation to a following obstruent, for
instance, each individual place is specifed by a distinct feature
(or set of features), and the overall unity of the process as one
applying exactly to all and only nasals, regardless of their place
of articulation, is not expressed. Nothing in the notation, that is,
makes it clear that a rule assimilating labiality, coronality, and
velarity is more coherent in some sense than one assimilating
labiality, voicing, and nasality.
Te response to this problem was a program to treat the
features themselves as organized into a hierarchy, such that all
place-of-articulation features (for example), and no others, are
daughters of a unitary node [Place]. On that approach, place
assimilation could be viewed as a unitary association of the
[Place] node itself, rather than individually to each of its vari-
ous possible values, while no such single unit corresponds to the
hypothetical alternative. Attention focused on such problems of
the internal geometry of the feature system generally led to the
assumption that the way to approach them was to assume that
the theory of rules should be limited to a very simple set of reas-
sociations and deletions within the autosegmental structure of
an utterance, and that a single, universal feature hierarchy could
be specifed on the basis of which all observed natural rules
(and no unnatural ones) could be formulated. Arguments
for and against specifc proposals about such a hierarchy have
drawn considerable attention, though it is perhaps notable that
the theoretical assumptions underlying the program have been
much less discussed .
LEXICAL PHONOLOGY. In classical generative phonology, the
interface between word structure and sound structure is quite
simple. morphological elements are combined into words
in the syntax , these elements are provided with phonological
(underlying) forms, and the resulting syntactically organized
labeled, bracketed structure serves as the input to the phonology.
At least some of the phonological rules were assumed to apply
according to the principle of the cycle , based on this structure,
in a uniform way. To the extent that morphological elements dis-
play diferent phonological properties in their combinations with
others, this was represented as diferences within an inventory of
boundary elements separating them from adjacent items.
Originating from the apparent generalization that elements
with the same phonological behavior (hence, associated with the
same boundary type) tend to appear adjacent to one another,
the theory of lexical phonology proposed a substantial revision to
this architecture. Instead of constructing the entire representa-
tion once and for all and then submitting it to the phonology for
realization, this view proposed that the lexicon of morphological
elements is divided into multiple strata or levels. Basic roots can
combine with elements of the frst stratum; after each such mor-
phological addition, the resulting form is subject to adjustment
by the rules of a corresponding level of the phonology, and the
output is then eligible to serve as the input to further morpho-
logical elaboration. At some point, addition of elements from the
frst stratum is replaced by use of the morphology and phonology
1970s, attempts to describe the phonology of tonal systems led
to important changes in assumptions about representations and
a concurrent shift of attention on the part of phonologists.
Te classical theory had assumed that phonological (and
phonetic) representations were given in the form of a simple
matrix, where each row represented a phonological distinctive
feature and the columns represented successive segments. Such
a representation is based on the assumption that there is a one-
to-one relation between the specifcations for any given feature
and those for all other features, since each column contains
exactly one specifcation for each feature.
Tonal phenomena, however, made it clear that features need
not be synchronized in this way: A given feature specifcation
might take as its scope either more or less than a single segment .
A classic example of this, ofered by W. Leben, is found in Mende,
where each word bears one of a limited set of tonal patterns,
regardless of the number of syllables on which this pattern is
realized. Tus, the tone pattern high-low appears on a single
syllable in mb (and thus the low has scope over only the last half
of the vowel), on two in ngl , and on three in flm (where the
single low of the pattern takes scope over two vowels). Tis led
to the development of autosegmental representations, in which
feature specifcations were linked by lines of association (subject
to specifc constraints), rather than all being aligned into seg-
ments. Te extension of this insight to other phenomena, and its
consolidation, essentially displaced the earlier concerns of rule
notation and ordering in phonologists attention .
METRICAL PHONOLOGY. A similar development took place in
the analysis of stress and the study of the syllable. Te analysis
in Chomsky and Halle ( 1968 ) treated stress as simply one more
phonological feature, with a value assigned to some (but not all)
of the segments in the representation of a word. Tis account was
forced to attribute a number of basic properties to the feature
[Stress], however, that had no obvious correlates in the behavior
of other features.
It became possible to rationalize these properties by viewing
stress not as a segmental feature but as a relational property of
the organization of syllables into larger structures. Tis, in turn,
required the recognition of syllables as signifcant structural
units: a notion that was explicitly rejected in the earlier theory
in favor of an attempt to reformulate all apparent syllable-based
generalizations in terms of segmental structure alone. Te orga-
nization of segments into syllables and these, in turn, into larger
units called feet , which themselves are organized into phonologi-
cal words (and phrases, etc.), allows for the elimination of the
anomalous character of segmentalized stress. Te study within
metrical phonology of these units, their internal organization,
and their relation to one another completed the enrichment of
the notion of phonological representation begun within autoseg-
mental phonology .
FEATURE GEOMETRY. A standard theme of classical generative
phonology was that of natural classes of phonological segments,
groups of segments that function in some parallel fashion in pho-
nological rules to the exclusion of others. It was originally hoped
that the analysis of segments into distinctive features would pro-
vide the solution to this issue: Segments sharing a feature (or set
Phonology
611
example, when languages accommodate loan words to the sur-
face patterns of other words of the language, the adjustments
needed to achieve this may include changes that do not corre-
spond to any rule of the phonology of native forms. Constraints
accomplish this directly and without further stipulation, whereas
a system of rules may have to be arbitrarily extended to account
for loanword adaptation.
On the other hand, some of the same issues that rule-based
phonology dealt with (and at least largely resolved) have resur-
faced as serious challenges to the architecture of grammar gen-
erally assumed in constraint-based theories. Most important
among these is the problem of opaque generalizations. Te
standard model of OT assumes that its constraints apply directly
to surface forms and govern a single-stage mapping between
these and underlying phonological representations, and so
has no place for generalizations that crucially apply to any sort
of intermediate level. Nonetheless, a number of compelling
examples of such phenomena have been demonstrated, and
some sort of accommodation of these facts must be provided by
an adequate phonological theory.
Some responses to this challenge have attempted to maintain
the standard OT model by introducing new sorts of constraints.
Mechanisms such as output - output constraints or sympathy the-
ory , however, have not generally succeeded in dealing with all of
the relevant phenomena and have been shown to produce new
dif culties of their own.
One approach that seems promising is that of stratal OT ,
an architecture that grafts a constraint-based account onto the
standard model of lexical phonology. Te result is a framework
in which the phonological mapping at each stage is a one-step
process governed by a constraint system. Since the model is
built on a cyclic interaction of phonology and morphology, how-
ever, it also provides for multiple successive stages in the overall
derivation, thus accommodating opacity to the extent it can be
related to morphological structure (as in the best-established
examples).
Examples also seem to exist in which the specifc changes
through which a language achieves conformity with a gen-
eral constraint on surface forms do not follow directly from
the content of the constraint (together with other interacting
generalizations). In such a case, something like a rewriting
rule might be necessary, as a supplement to the constraint
system a notion that is clearly antithetical to the basic phi-
losophy of OT.
A quite diferent problem concerns the very nature of the
universals of phonological structure (see phonology, uni-
versals of ). Phonological theorizing has generally accepted
the premise that generalizations that are true of phonological
systems in general result from the cognitive organization of the
human language faculty and, thus, must be incorporated in some
way into the architecture of phonological theory. Recently, how-
ever, it has been argued that at least some such typological regu-
larities result not from the content of a universal grammar
constraining synchronic systems but, rather, from the universals
of language change (see language change, universals of )
governing the diachronic developments resulting in the systems
we observe. To the extent that this is true, it requires investiga-
tors to examine closely the arguments for incorporating any
of the next, and, from then on, no further elements from the ini-
tial stratum can be added. Tis process continues (perhaps vacu-
ously) through all of the strata of the lexicon, yielding a potential
surface word. All of the words in a given syntactic structure are
then subject to adjustment by another set of postlexical phono-
logical processes.
Tere are a number of further points that characterize this
view, including proposed diferences in the properties of lexical
and postlexical rules and the relations between rules on one level
and those on the others. Te central point for a broader theory
of grammar, however, is probably the replacement of a syntax-
based (but purely phonological) notion of cyclic rule application
by a repeated cycle of morphological addition and phonologi-
cal adjustment. Tis results, for example, in the possibility that
a phonologically derived property (on one cycle) can be relevant
to the conditioning of a morphological operation (on a following
cycle), a possibility that has been shown to be quite real .
OPTIMALITY THEORY. In the early 1990s, a much more radical
challenge to the classical model was presented by the develop-
ment of optimality theory (OT), a view of phonology based
on a system of ranked, violable constraints on surface shape,
as opposed to a system of ordered rules deriving the phonetics
from an underlying phonological representation. Tese con-
straints govern (in the standard formulation) a one-step relation
between underlying and surface representations (cf. underly-
ing structure and surface structure ), with no interme-
diate stages of the sort produced in a rule-based description. Te
constraints can be divided into general classes: a) markedness
constraints, which express universally preferred confgurations,
and b) faithfulness constraints, requiring that contrasts present
in the phonological representation be preserved in the surface
form. In general, these are in confict, and the ranking of the con-
straints governs the resolution of those conficts, in conformity
with general principles of grammar.
Initially, OT seemed to ofer its greatest promise in the analy-
sis of stress, syllable structure, and related phenomena, but sub-
sequent development has encompassed a full range of segmental
and other facts. Descriptions in constraint-based terms are at
least superfcially very diferent from those couched in terms of
traditional rules, and theoretical discussion in phonology since
their introduction has been largely dominated by comparisons
of the two frameworks .
Current Approaches to Phonology
Te central issues in phonology in the frst decade of the twen-
ty-frst century concern the comparative merits of OT and
rule-based descriptions. On the one hand, constraint-based
formulations seem much better equipped to describe global
properties of phonological systems. It was noted in work from
the classical period of generative phonology that multiple dis-
tinct processes in an individual language may all have the efect
of ensuring (or avoiding) a single characteristic property of sur-
face form, but no satisfactory account of the unity displayed by
these conspiracies was ever achieved. OT, in contrast, provides
a very direct description of such facts.
In some ways, the surface constraint approach goes beyond
anything available in principle to the rule-based theory. For
Phonology
612
understand the system that relates childrens stored representa-
tions to their productions, and to formalize the developmental
paths that children follow.
Childrens Grammars as Possible Grammars
Te focus of research on phonological acquisition is on the shapes
of early grammars in the segmental and prosodic domains; thus,
it parallels research on end-state (adult) grammars. (Segmental
phonology is concerned with individual speech sounds, prosodic
phonology with larger units including syllables and feet.) Tere is
typically a comparison drawn between the shapes of developing
grammars and some end-state grammar. Order of emergence
of segmental (Dinnsen 1992 ) and prosodic complexity (Fikkert
1994 ; Levelt, Schiller, and Levelt 1999 /2000), as well as error pat-
terns observed in the segmental and prosodic domains, whether
these patterns are expressed through rules (Smith 1973 ; Ingram
1974 ), templates (Macken 1992 ; Fikkert 1994 ), or constraints
(Pater and Barlow 2003 ; Goad and Rose 2004 ), are all considered
in relation to some adult grammar.
One exception is a body of research that views childrens
grammars as self-contained systems subject to their own con-
straints (Stoel-Gammon and Cooper 1984 ; Vihman 1996 ). Tis
research program developed in response to the observation that
childrens grammars are not simply reduced versions of the
target grammar; indeed, variation across learners is rampant
(Ferguson and Farwell 1975 ) .
While childrens grammars may be self-organizing in that
they contain processes not present in the target language, they
can still be viewed as possible grammars (White 1982 ; Pinker
1984 ) if these processes have correlates in other adult languages.
Te notion of possible grammar thus requires that, at each
stage, childrens grammars respect the constraints of adult gram-
mars, even if they bear little resemblance to the target system. In
optimality theory (OT) (Prince and Smolensky [ 1993 ] 2004),
for example, alternate routes observed across learners, as well as
stages in the development of a single learner, are viewed from
the perspective of the typological options that adult languages
display: Both are accounted for by diferent rankings of the same
constraints .
Markedness
Although children take diferent paths to the adult grammar, early
phonologies are also strikingly similar (Jakobson [ 1941 ] 1968).
As Roman Jakobson emphasizes, these similarities refect cross-
linguistically unmarked properties. markedness constrains the
shapes of linguistic systems such that less complex properties are
favored. For example, there is a well-documented preference for
consonant+vowel (CV) syllables among children (Ingram 1978 ;
cf. Grijzenhout and Joppen-Hellwig 2002 ); this is also a syllable
shape that no end-state grammar forbids (Jakobson 1962 ). Since
unmarked patterns are systematically observed across learners,
one might reasonably infer that they refect early grammatical
organization. However, markedness has not always been well
integrated into the theory of grammar (as part of the theory of
representations or formulation of rules/constraints). Tis begs
the question of whether markedness should instead be part of
the theory of acquisition, which interfaces with, but is indepen-
dent of, the theory of grammar.
particular regularity into phonological theory per se, as opposed
to seeking its basis elsewhere .
Conclusion
While there have, of course, been other trends not covered here,
it seems fair to say that the bulk of the theoretical discussion in
phonology from the 1960s to the present has been devoted to
the elaboration and refnement of the generative program of
Chomsky and Halle ( 1968 ). Te most recent developments in
that tradition, involving the wholesale replacement of rules by
constraints as the mechanism for expressing regularities of a lan-
guages sound pattern, have shown great promise but cannot yet
be considered wholly consolidated. Apparently, some appropri-
ate synthesis of the classical and OT models remains to be found,
and it is that search that dominates discussion today .
Stephen R. Anderson
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Stephen R. 1985 . Phonology in the Twentieth Century: Teories
of Rules and Teories of Representations . Chicago : University of Chicago
Press . Describes the development of phonological theory, from its ori-
gins through the classical period of generative phonology.
Chomsky , Noam , and Morris Halle . 1968 . Te Sound Pattern of English .
New York : Harper and Row.
Gussenhovern , Carlos , and Haike Jacobs , 2005. Understanding Phonology .
2d ed. New York : Oxford University Press . Lucid elementary introduc-
tion to current phonology.
Kager , Ren 1999 . Optimality Teory: A Textbook . Cambridge : Cambridge
University Press . Introduces the main ideas of optimality theory in
phonology and their implementation.
Kenstowicz , Michael 1994 . Phonology in Generative Grammar .
Cambridge, MA : Blackwell . Provides a comprehensive description of
the principal themes in phonology up to the introduction of optimality
theory.
PHONOLOGY, ACQUISITION OF
A diversity of issues informs work in the feld of phonological
acquisition, as it encompasses both frst (L1) and second (L2)
language acquisition examined by researchers in linguistics,
psychology, speechlanguage pathology, and language edu-
cation. In L1, there are questions such as how the acquisition
of phonology interfaces with perceptual and motor develop-
ment (Locke 1993 ), and how an examination of disordered
development can illuminate the normally developing gram-
mar (Bernhardt and Stemberger 1998 ; Dinnsen 1999 ). In L2,
there are questions as to whether the acquisition process is
fundamentally like L1 acquisition (Flege 1995 ), or whether L2
grammars are in some sense impaired due to, for example, L1
constraints that impede native-like attainment (see Brown 1998
on perception).
Due to space constraints, this entry focuses on L1, although
many of the same issues arise for L2. Te acquisition of phonol-
ogy is examined from the perspective of generative grammar ;
thus, a principal theme is to examine how acquisition research
has used linguistic theory to inform development. Tis theme
considers the starting hypothesis to be that childrens produc-
tions are largely system driven: Acquisition research strives to
Phonology, Acquisition of
613
As shown in Table 1 , at Stage 1, both perception-specifc
faithfulness (F-perc) and general faithfulness (F) are outranked
by markedness (M). Te result is unmarked forms stored in per-
ception and uttered in production. In the example provided, the
ambient form [wei] away undergoes truncation of the pre-
tonic syllable (an unstressed syllable immediately preceding a
stressed syllable) in both components of the childs grammar;
accordingly, words of this shape are perceived and produced
without this syllable. At Stage 2, the childs perceptual abilities
become more target-like (i.e., he/she learns to correctly iden-
tify information in the ambient language); this indicates that
the relevant markedness constraints have been demoted below
perception-specifc faithfulness. General faithfulness is still out-
ranked, yielding a mismatch between what the child perceives
and what he/she produces. At Stage 3, markedness is demoted
below general faithfulness, and the form is correctly produced.
Te perception-production time lag results because forms that
are correctly perceived at Stage 2 are not correctly produced until
Stage 3 .
Phonological Theory and Phonological Acquisition
As the preceding discussion reveals, research in phonological
acquisition has been directly impacted by thinking in generative
phonology. Modern generative phonology began with Chomsky
and Halles ( 1968 ) Sound Pattern of English (SPE) . Although more
recent work has situated the shapes of developing grammars
within the typological range manifested by adult systems, this
was less the case in the SPE-based literature. Much of this work
used SPE as a tool only, in part because, with the formal appa-
ratus employed by the theory, it was dif cult to constrain what
a possible grammar is: developing or end state. And although
the theory contained an evaluation metric to guide learners in
selecting the most highly valued among descriptively adequate
grammars, rules for unattested processes were as easy to formal-
ize as rules for commonly attested processes. Finally, SPE con-
tained no workable theory of markedness and, thus, childrens
grammars could not be considered relative to some notion of
optimal.
To facilitate a comparison between SPE and later theories,
we draw on truncation, further exemplifed in (1) from Amahl,
age 2.60 (Smith 1973 ; [b,g] are voiceless unaspirated lenis stops).
(Te discussion focuses on the stage when perception is target-
like and truncation is restricted to production.)
(1) [gep] escape
[an] banana
In SPE, every deviation from adult forms required one or more
rules, and so there was little in common between the rule sets
An advantage of OT is that the formal devices for expressing
phonological generalizations include a set of markedness con-
straints. Most researchers have proposed that learners begin
acquisition with a ranking wherein markedness constraints dom-
inate faithfulness (which favor identity between inputs [stored
representations] and outputs) (e.g., Demuth 1995 ; Gnanadesikan
[ 1995 ] 2004; Smolensky 1996 ; Pater 1997 ; Ota 2003 ; cf. Hale and
Reiss 1998 ). Troughout development, constraints are reranked
to yield more marked outputs. However, many paths can be fol-
lowed, as there are many options for what to rerank. Tus, the
idea that grammars are initially unmarked is not inconsistent
with their being self-organizing .
Perception Versus Production
Most work in phonological acquisition has focused on produc-
tion; indeed, researchers typically assume that children accu-
rately perceive the ambient input. Tis is due, in part, to the
observation that prelinguistic infants can perceptually discrimi-
nate perhaps all contrasts exploited by the worlds languages
(Eimas et al. 1971 ; Werker et al. 1981 ). Tis ability largely declines
by age one (cf. Best, McRoberts, and Sithole 1988 ), coinciding
with a reorganization of perceptual categories according to what
is contrastive in the target language (Werker and Tees 1984 ). As
children start to speak around age one, it would appear that per-
ception is complete by the onset of production.
Research on phonemic perception, which requires the abil-
ity to form soundmeaning pairings, has challenged this view
(Shvachkin [ 1948 ] 1973; Edwards 1974 ; Brown and Matthews
1997). Although experiments examining minimal contrasts
between native-language sounds have revealed that perceptual
development is mostly complete by age two, some contrasts
develop as late as three. Even age three is probably conservative
because, for consonant perception, this research has focused
almost exclusively on word-initial position. Since contrasts in
other positions are harder to discriminate, many non-target
patterns that childrens productions display could refect per-
ceptual miscoding, rather than production constraints (Macken
1980 ).
If perception and production both refect aspects of childrens
competence , both must be included in the grammar (cf. Hale
and Reiss 1998 ). However, the time lag observed (production
trails perception) has suggested to some researchers that they
form independent (interacting) grammatical modules (see Menn
and Matthei 1992 ). Tis approach, though, cannot predict that
perception and production abilities develop in a similar order.
Te latter favors the postulation of a single grammar if the time
lag can be built in. In Pater (2004), this is accomplished by intro-
ducing perception-specifc faithfulness constraints into OT .
Table 1.
Ambient
form:
Stage Grammar:
Stored form: Grammar: Produced
form:
[wei] 1 M >> F-perc, F /wei/ M >> F-perc, F
[wei]
2 F-perc >> M >> F
/wei/
F-perc >> M >> F
3 F-perc, F >> M F-perc, F >> M [wei]
Phonology, Acquisition of
614
To illustrate, concerning truncation in Table 2 , the con-
straint ParseSyllable (syllables are parsed into feet), along
with other markedness constraints, must be satisfied at the
expense of the lower-ranked faithfulness constraint Max-IO
(every segment in the input has a correspondent in the out-
put). Fully faithful (a) is thus eliminated because the initial syl-
lable is unfooted. Concerning onset selection, Max [labial] -IO
(every [labial] in the input has a correspondent in the out-
put) must be ranked over I-Contiguity (the portion of the
input standing in correspondence forms a contiguous string).
Preservation of [labial] in banana will thus be favored, (c),
even though the result violates I-Contig through morpheme-
internal segment deletion.
OT has had a major impact on acquisition research.
Phonological processes are now generally expressed through
constraints, rather than rules, as this provides a better concep-
tualization of the observation that markedness shapes early
grammars. As discussed, childrens productions become more
target-like when markedness constraints are demoted below
faithfulness. A similar idea, that development is best viewed as
the gradual relaxing of constraints, had been proposed earlier
(Stampe 1969 ; Menn 1980 ), but it was dif cult to formally imple-
ment it in the rule-based frameworks of the time.
OT seems to provide an appealing view of the initial state and
of development; researchers can address important questions,
such as how the theory may restrict what a possible develop-
ing grammar is, and how, in turn, data from development may
inform the theory. However, this is not to say that OT has solved
all problems in phonological acquisition. One understudied
problem is rogue behavior. We have been assuming that chil-
drens grammars are possible grammars, thereby ignoring the
fact that some commonly attested processes, notably conso-
nant harmony (CH), have no adult analogs (Drachman 1978 ). In
CH, consonants share place over vowels of any quality (Vihman
1978 ), as seen in (3) for Amahl, age 2.60 (Smith 1973 ):
(3) [aig] tiger
[ok] stroke
Some recent accounts of CH (Goad 1997 ; Rose 2000 ) incor-
rectly predict that the process should be attested in adult gram-
mars; others (Pater 1997 ) appeal to child-specifc constraints,
thereby challenging the notion that childrens grammars are pos-
sible grammars. Neither of these approaches questions whether
CH is truly grammar-driven nor addresses, more generally, the
criteria that should factor into the determination concerning
what is grammar-driven and what is not. I leave these questions
to future work .
Heather Goad
for developing and target systems. To capture truncation, Neil
Smith ( 1973 ) provides the rules below, neither of which operates
in the adult grammar:
(2) R14: V / # (C) ______ C V
[-stress] [+stress]
R16: [+sonorant] / [+consonantal] ______
R14 deletes initial vowels in words like escape. For consonant-
initial forms like banana , the result is [bnan], which then under-
goes R16, yielding [ban].
Since SPE employed linear representations, the theory did
not ofer any insight into why pretonic rather than posttonic syl-
lables delete (escape [gep], but tiger [aig], *[ai()]).
Te development of nonlinear phonology (see Goldsmith 1985
for an overview), notably the move to highly articulated prosodic
representations, led to signifcant breakthroughs in understand-
ing this asymmetry. In trochaic languages, where the foot (the
rhythmic unit in which stress is assigned) is left-headed (stress-
initial), escape cannot form a single foot, [s(kip)
Ft
]
Wd
, whereas
tiger can, [(tig)
Ft
]
Wd
.
Much work in nonlinear phonology has explored the idea
that prosodically defned templates constrain output shape
(McCarthy and Prince 1995 ). Paula Fikkert ( 1994 ) proposes that
templates, which at early developmental stages refect what
is unmarked, are responsible for truncation. If the childs pro-
ductions are limited to one foot, circumscribed from the adult
output, this template will determine which material is preserved
from the adult form and which is deleted:
In contrast to SPE, nonlinear phonology reveals the rela-
tionship between target and truncated forms, and the role that
markedness plays in shaping outputs. Te material inside the
foot survives, as syllables organized by feet ([keip]) are less
marked than those linking directly to the word ([s]). One prob-
lem with the templatic approach, however, is that it is too rigid: If
the segments predicted to survive are precisely those delimited
by the constituent that serves to organize them in the adult form,
it becomes dif cult to capture the observation that material from
the truncated syllable can also survive. For example, in Amahls
pronunciation of banana in (1), onset selection favors [b], replac-
ing [n] from the stressed syllable; that is, his production is [an]
not *[nan] as expected from adult [b(nan)
Ft
]
Wd
(see Kehoe
and Stoel-Gammon 1997 for other problems with the templatic
approach) .
Tis problem is rectifed in OT. First, there are no templates;
templatic efects arise from the interaction of markedness con-
straints. Second, segmental content (e.g., labial preservation)
is the responsibility of faithfulness constraints. Finally, all con-
straints are interranked; thus, the co-occurrence of truncation
and onset selection is not unexpected (see Pater 1997 ).
Adult output:
Wd
Ft
s
keip
Ft
Child output: Wd
e
Table 2.
ParseSyll Max [lab] -IO Max-IO I-Contig
a. [b(nan)
Ft
]
Wd
*!
b. [(nan)
Ft
]
Wd
*! **
c. [(ban)
Ft
]
Wd
** *
Phonology, Acquisition of
615
Ingram , David . 1974 . Phonological rules in young children . Journal of
Child Language 1 : 49 64.
. 1978 . Te role of the syllable in phonological development.
In Syllables and Segments , ed. Alan Bell and Joan B. Hooper , 14355.
Amsterdam : North-Holland.
Jakobson , Roman . [1941] 1968. Child Language, Aphasia and Phonological
Universals . Trans. Allan Keiler. Te Hague : Mouton.
. 1962 . Selected Writings. Vol. 1. Phonological Studies . Te
Hague : Mouton .
Kehoe , Margaret , and Carol Stoel-Gammon . 1997 . Te acquisition of
prosodic structure: An investigation of current accounts of childrens
prosodic development . Language 73 : 113 44.
Levelt , Clara , Niels Schiller , and Willem Levelt . 1999 /2000. Te acquisi-
tion of syllable types . Language Acquisition 8 : 237 64.
Locke , John . 1993 . Te Childs Path to Spoken Language .
Cambridge : Harvard University Press .
Macken , Marlys . 1980 . Te childs lexical representation: Te puzzle-
puddle-pickle evidence . Journal of Linguistics 16 : 1 17.
. 1992 . Wheres phonology? In Phonological Development: Models,
Research, Implications, ed. Charles Ferguson , Lise Menn , and Carol
Stoel-Gammon , 24969. Timonium, MD : York .
McCarthy , John , and Alan Prince . 1995 . Prosodic morphology. In
Goldsmith 1995, 31866.
Menn , Lise . 1980 . Child phonology and phonological theory. In Child
Phonology. Vol. 1: Production . Ed. Grace Yeni-Komshian , James
Kavanaugh , and Charles Ferguson , 2342. New York : Academic Press .
Menn , Lise , and Edward Matthei . 1992 . Te two-lexicon account to
child phonology. In Phonological Development: Models, Research,
Implications , ed. Charles Ferguson , Lise Menn , and Carol Stoel-
Gammon , 21147. Timonium MD : York .
Ota , Mits . 2003 . Te Development of Prosodic Structure in Early Words .
Amsterdam: Benjamins .
Pater , Joe . 1997 . Minimal violation and phonological development .
Language Acquisition 6 : 201 53.
. 2004 . Bridging the gap between receptive and productive
development with minimally violable constraints. In Constraints
in Phonological Development , ed. Ren Kager , Joe Pater , and Wim
Zonneveld , 21944. Cambridge : Cambridge University Press .
Pater , Joe , and Jessica Barlow . 2003 . Constraint confict in cluster reduc-
tion . Journal of Child Language 30 : 487 526.
Pinker , Steven . 1984 . Language Learnability and Language Development .
Cambridge : Harvard University Press .
Prince , Alan , and Paul Smolensky . [1993] 2004. Optimality
Teory: Constraint Interaction in Generative Grammar.
Oxford : Blackwell .
Rose , Yvan . 2000 . Headedness and Prosodic Licensing in the L1 Acquisition
of Phonology . Ph.D. diss., McGill University .
Shvachkin , N. Kh . [1948] 1973. Te development of phonemic speech
perception in early childhood. In Studies of Child Language
Development , ed. Charles Ferguson and Dan Slobin , 91127. New
York: Holt, Rinehart and Winston .
Smith , Neil . 1973 . Te Acquisition of Phonology: A Case Study .
Cambridge : Cambridge University Press .
Smolensky , Paul . 1996 . On the comprehension/production dilemma in
child language . Linguistic Inquiry 27 : 720 31.
Stampe , David . 1969 . Te acquisition of phonetic representation .
Chicago Linguistic Society 5 : 433 44.
Stoel-Gammon , Carol , and Judith Cooper . 1984 . Patterns of early lex-
ical and phonological development . Journal of Child Language
11 : 247 71.
Vihman , Marilyn . 1978 . Consonant harmony: Its scope and function in
child language. In Universals of Human Language. Vol. 2: Phonology.
Ed. Charles Ferguson , 281334. Stanford, CA : Stanford University Press .
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bernhardt , Barbara , and Joe Stemberger . 1998 . Handbook of Phonological
Development from the Perspective of Constraint-Based Nonlinear
Phonology. San Diego, CA : Academic Press .
Best , Catherine , Gerald McRoberts , and Nomathemba Sithole . 1988 .
Examination of perceptual reorganization for nonnative speech
contrasts: Zulu click discrimination by English-speaking adults and
infants . Journal of Experimental Psychology: Human Perception and
Performance 14 : 345 60.
Brown , Cindy . 1998 . Te role of the L1 grammar in the L2 acquisition of
segmental structure . Second Language Research 14 : 136 93.
Brown , Cindy , and John Matthews . 1997 . Te role of feature geometry
in the development of phonetic contrasts. In Focus on Phonological
Acquisition , ed. S. J. Hannahs and Martha Young-Scholten , 67112.
Amsterdam : Benjamins .
Chomsky , Noam , and Morris Halle . 1968 . Te Sound Pattern of English .
New York : Harper & Row .
Demuth , Katherine . 1995 . Markedness and the development of prosodic
structure . Proceedings of NELS 25 : 13 25.
Dinnsen , Daniel . 1992 . Variation in developing and fully developed pho-
netic inventories. In Phonological Development: Models, Research,
Implications , ed. Charles Ferguson , Lise Menn , and Carol Stoel-
Gammon , 191210. Timonium, MD : York .
. 1999 . Some empirical and theoretical issues in disordered child
phonology. In Handbook of Child Language Acquisition , ed. William
Ritchie and Tej Bhatia , 647704. San Diego, CA : Academic Press .
Drachman , Gaberell . 1978 . Child language and language change: A con-
jecture and some refutations. In Recent Developments in Historical
Phonology , ed. Jacek Fisiak , 12344. Te Hague: Mouton .
Edwards , Mary Louise . 1974 . Perception and production in child pho-
nology: Te testing of four hypotheses . Journal of Child Language
1 : 205 19.
Eimas , Peter , Einar Siqueland , Peter Jusczyk , and James Vigorito . 1971 .
Speech perception in infants . Science 171 : 3036.
Ferguson , Charles , and Carol Farwell . 1975 . Words and sounds in early
language acquisition . Language 51 : 419 39.
Fikkert , Paula . 1994 . On the Acquisition of Prosodic Structure . Te
Hague : Holland Academic Graphics .
Flege , James . 1995 . Second language speech learning: Teory,
fndings, and problems. In Speech Perception and Linguistic
Experience: Teoretical and Methodological Issues , ed. Winifred
Strange , 23377. Timonium, MD : York .
Gnanadesikan , Amalia . [1995] 2004. Markedness and faithfulness
constraints in child phonology. In Constraints in Phonological
Development , ed. Ren Kager , Joe Pater , and Wim Zonneveld , 73108.
Cambridge : Cambridge University Press .
Goad , Heather . 1997 . Consonant harmony in child language: An
optimality-theoretic account. In Focus on Phonological
Acquisition , ed. S. J. Hannahs and Martha Young-Scholten , 11342.
Amsterdam : Benjamins .
Goad , Heather , and Yvan Rose . 2004 . Input elaboration, head faith-
fulness and evidence for representation in the acquisition of left-
edge clusters in West Germanic. In Constraints in Phonological
Development , ed. Ren Kager , Joe Pater , and Wim Zonneveld , 10957.
Cambridge : Cambridge University Press .
Goldsmith , John , ed. 1995. Te Handbook of Phonological Teory .
Oxford : Blackwell .
Grijzenhout , Janet , and Sandra Joppen-Hellwig. 2002 . Te lack of onsets
in German child phonology. In Te Process of Language Acquisition ,
ed. Ingeborg Lasser , 31939. Frankfurt am Main : Peter Lang .
Hale , Mark , and Charles Reiss . 1998 . Formal and empirical arguments
concerning phonological acquisition . Linguistic Inquiry 29 : 656 83.
Phonology, Acquisition of
616
Communication during this period was based on gestures .
Spoken language emerged in the second transition. In Donalds
scenario, it takes until the end of the period associated with
archaic Homo sapiens (45,000 years ago) for spoken language to
appear.
If mimesis was a basically gestural mode of communication,
would it not imply a proto-language that was signed, rather than
spoken (cf. Arbib 2005 )? Donald assumes that as mimetic mes-
sages grew more elaborate, they eventually reached a complex-
ity that favored faster and more precise ways of communicating.
Te vocal/auditory modality ofered an independent, omnidir-
ectional channel useful at a distance and in the dark. It did not
impede locomotion, gestures, or manual work. Te vocal system
came to be exploited more and more and further adaptations
occurred: frst lexical invention and high-speed phonological
speech , syntax later .
Specializations of the Vocal/Auditory Modality
A number of comparative studies have been undertaken in
attempts to evaluate the adaptive signifcance of novel fea-
tures of human anatomy (see speech anatomy , evolution
of ): for example, disappearance of air sacs (Hewitt, MacLarnon,
and Jones 2002 ), bigger hypoglossal and vertebral canals, smaller
masticatory muscle mass, genetic changes, and uniqueness of
craniofacial sensorimotor system (Fitch 2000 ; Kent 2004 ).
Perhaps the most conclusive example of a speech-related
adaptation is the descent of the larynx , which makes swallowing
more hazardous but expands the space of possible sound quali-
ties (Lieberman 1991 ; Carr, Lindblom, and MacNeilage 1995 ).
A human ability, central to language but curiously absent
in primates, is vocal imitation . A beginning of a neural account
of imitation was suggested by the discovery of mirror neurons .
First identifed in the macaques premotor cortex, these neurons
discharge when the monkey manipulates objects and when it
observes other monkeys or humans do the same. Neurons that
respond to sound and to communicative or ingestive actions
have also been identifed. Although there is no direct evidence
for a human mirror system, brain stimulation and imaging stud-
ies indicate increased activity in speech muscles when subjects
listen to speech (Hurley and Chater 2005 ; see mirror systems,
imitation, and language ) .
Signals for Speaker, Listener, and Learner
In technical jargon, phonology has been characterized as provid-
ing an impedance match between semantics and phonetics in
the sense that it succeeds in coding a large number of meanings
despite its use of only a small set of phonetic dimensions (Bellugi
and Studdert-Kennedy 1980 ). How was this match achieved?
GESTURES AS BASIC UNITS. One answer is that the building
blocks of speech are phonetic gestures , units corresponding to
the discrete articulators. Te argument is that, evolutionarily,
as holistic utterances were processed by the mirror system, they
came to be parsed into the basic articulators of the vocal tract
and their preferred, natural motions. Data from early speech
have been used to argue that these units, when properly timed
and modulated in amplitude, produce the vowels and conso-
nants of the ambient adult input (Studdert-Kennedy 2005 ).
. 1996 . Phonological Development: Te Origins of Language in the
Child . Oxford : Blackwell .
Werker , Janet , John Gilbert , Keith Humphrey , and Richard Tees . 1981 .
Developmental aspects of cross-language speech perception . Child
Development 52 : 349 53.
Werker , Janet , and Richard Tees . 1984 . Cross-language speech percep-
tion: Evidence for perceptual reorganization during the frst year of
life . Infant Behavior and Development 7 : 49 63.
White , Lydia . 1982 . Grammatical Teory and Language Acquisition .
Dordrecht, the Netherlands : Foris .
PHONOLOGY, EVOLUTION OF
True Phonology: How Did It Evolve?
phonology is the study of how languages use segmental and
prosodic categories to build spoken words and signal their dif-
ferences in meaning. Many animals that communicate vocally
use distinct sound patterns to signal diferent meanings, but
their repertoires are typically small and closed sets. By contrast,
humans have large vocabularies and learn new words all of
their lives. From infancy to adolescence, children acquire lexi-
cal entries at a remarkably fast rate. Te vocabulary size of high
school students has been conservatively estimated at 60,000 root
forms a number that implies an average acquisition rate of
more than 10 words per day (Miller 1991 ).
Tis diference is linked to the uniquely human method of
coding information: the combinatorial use of discrete entities.
Combinatorial structure, the hallmark of true language, creates
the conditions for open-ended lexical and syntactic systems
that provide the foundation for the singular expressive power of
human languages. How did it evolve?
We focus on two areas of empirical research. One is the study
of human cognition . Te other is the investigation of the phonetic
signal space from which all phonological patterns are drawn.
Te frst theme highlights mans rich semantic abilities. Te
second looks for phenomena that presage combinatorial sound
structure.
Cognitive Growth
Te virtually infnite set of meanings encodable by language
raises the question of how mans cognitive capacity evolved
from skills not unlike those of present-day apes. How do we pic-
ture the transition from a nonhuman to a human primate mind?
What was the role of language?
According to Merlin Donalds synthesis of neurobiologi-
cal, psychological, archeological, and anthropological evidence
(1991), our ancestors broke away from the stimulus-driven
behavior of apes in two steps. First, during the period of Homo
erectus (from 1.5 million years ago), an adaptation called mimesis
occurred, a communicative culture allowing individuals to share
mental states and begin to represent reality in new and expand-
ing ways.
Mimetic behavior is an ability to voluntarily access and
retrieve motor memories and to rehearse and model them for
communication with others. Te whole body is used as a rep-
resentational device, as in imitating vocal, manual, and postural
movements for a communicative purpose. Mimesis involved
major changes of motor and memory mechanisms based on
existing capacities.
Phonology, Evolution of
617
why phonologies do not recruit more of a humans total sound-
making capabilities (e.g., mouth sounds and other non-speech
vocalizations; Catford 1982 ) but prefer practically the same small
set of phonetic properties. However, the study of these constraints
only partially illuminates the roots of combinatorial coding. Tis
topic has been explicitly addressed in computer modeling exper-
iments. One such study shows how discrete phonetic targets and
reuse can emerge from a dynamic systems network of agents
(speaker/listener models) whose vocalizations initially ran-
domly distributed in phonetic space tend to converge (driven
by a magnet-like dominance of the patterns heard most often) on
a few targets (Oudeyer 2006 ) .
TARGETS AND MOTOR EQUIVALENCE. Traditionally, the basic units
of speech have been assumed to be targets , the intertarget transi-
tions being primarily determined by the response characteristics
of the production system . Speech, like other movements, exhib-
its motor equivalence : the ability of motor systems to compensate
and reach a given goal irrespective of initial conditions. Tis view
implies that the end state of phonetic learning is a set of context-
independent targets and a system capable of motor equivalence.
It moreover suggests that once a target has been learned in one
context, it can immediately be reused in other contexts, since the
motor equivalence capability handles the new trajectory. Also, it
means that, developmentally, discrete segments derive from the
emergent targets and recombination from motor equivalence .
A further relevant observation on the target hypothesis is that
linguistic systems with phonemically coded vocabularies would
be learned faster, more easily, and in an open-ended manner
than repertoires based on holistic forms (Lindblom 2007 ).
Conclusion
Where does combinatorial structure come from? From prespeci-
fcations in our genetic endowment? Or from a modality-inde-
pendent principle shared by sign and speech and perhaps also
operating in genetics and chemistry (cf. the particulate principle
[Abler 1989 ])? Or from a mutually reinforcing interplay between
cognitive growth and a suite of conditions entailed by communi-
cating by vocal sounds?
In view of the materials reviewed here, a positive treatment of
the last possibility appears within reach. More lexical inventions
imply an increasing number of soundmeaning pairs. Te link-
ing of phonetic shapes with distinct meanings would be subject
to numerous user-based constraints and processes shaping the
instrinsic content of lexical entries, fractionating them into dis-
crete units and facilitating unit recombination. Sound structure
could, thus, plausibly have evolved in response to the expressive
needs associated with growing semantic abilities and as a pro-
cess of phonetically biased scaling, self-organizing without
any formal a priori or modality-independent blueprint .
Bjrn Lindblom
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Abler , William L . 1989 . On the particulate principle of self-diversifying
systems. Journal of Social and Biological Structures 12 : 1 13.
Arbib , Michael . 2005 . Te mirror system hypothesis: How did proto-
language evolve? In Language Origins: Perspectives on Evolution , ed.
Maggie Tallerman , 2147. New York : Oxford University Press .
PHYLOGENY OF THE SYLLABLE. Te frame / content theory
(MacNeilage 2008) ofers an evolutionary account of the syllable.
Syllables are universally associated with openclose alternations
of the mandible, vowels being open and consonants closed artic-
ulations. Tis movement has a parallel in childrens babbles,
which resemble consonant-vowel sequences such as [bababa],
but are in no way organized in terms of discrete segments. Rather,
their syllabic and segmental character arises fortuitously from
adding phonation to the openclose jaw motion. Tis rhythmic-
ally repeated up-and-down movement is also found in so-called
lipsmacks, a facio-visual behavior in primates often combined
with phonation during grooming .
Accordingly, the evolutionary path to the syllable began in
deep prehistoric time when mammal biomechanics evolved for
feeding. A second stage was the use of this machinery in primate
communication. In a third step, this primate mechanism was co-
opted for speech by scafolding early phonology on its pseudo-
syllables and pseudo-segments .
QUANTAL THEORY. Te acoustic consequences of a continuous
articulatory movement are often noncontinuous, as illus-
trated by the pseudo-segmental character of babbling. In the
babble example, the jaw moves continuously but the acoustics
shows an abrupt change from a vowel-like to a stoplike pattern.
Tis quantal jump illustrates a general fact about the phonetic
space. Te mapping of articulation onto acoustic parameters
creates a set of acoustic patterns that forms a number of disjoint
subspaces, rather than a single continuous, coherent space.
Within each such subregion, sound quality is homogeneous.
Voiced and voiceless sounds, as well as diferent manners of
articulation (e.g., stops, nasals, fricatives, trills), exemplify such
distinct subspaces (Stevens 1989 ) .
USER-BASED CONSTRAINTS: ON-LINE SPEECH. Te human voice
is an expressive instrument that undergoes moment-to-moment
retuning by many nonlinguistic factors. Consequently, the pho-
netic patterns conveying linguistically the same utterance exhibit
great variability. However, the need for messages to be both intel-
ligible and pronounceable imposes a systematic distribution on
phonetic variations, placing them between clear hyper forms and
reduced hypo forms. Tis view portrays speakerlistener interac-
tions as a tug-of-war between the listeners need for comprehen-
sion and the speakers tendency to simplify. Tere is a great deal of
experimental evidence for this view of speech (Lindblom 1990 ) .
USER-BASED CONSTRAINTS: PHONOLOGY. Tese user-based con-
straints also leave their mark on phonology, as is evident from
typological data on strengthening and weakening processes in
phonological rules and sound changes (Kiparsky 1988 ) and from
attempts to simulate segment inventories. Tese studies indicate
that systemic selections have been favored that simultaneously
optimize distinctiveness and articulatory ease . An example of
the efect of these conditions is the size principle : Te larger the
system, the greater the proportion of articulatorily complex seg-
ments (Lindblom and Maddieson 1988 )
SELF-ORGANIZATION. Tese user-based constraints in conjunc-
tion with the quantal nature of the signal space help explain
Phonology, Evolution of
618
supposed human innate capacity for language. Tere is a sizable
literature on phonological universals (e.g., Greenberg, Ferguson,
and Moravcsik 1978 ; Maddieson 1984 ), and it will not be possible
in this limited space to discuss and exemplify more than a few
of those that have been discovered. What is more important and
what will be emphasized here is a consideration of the explana-
tion for phonological universals. Te best evidence presented so
far points to their phonetic origin .
A caveat: Phonological universals as with any other phono-
logical generalization are inevitably stated in terms of a tradi-
tional pretheoretic taxonomy. One should always be alert to the
possibility that the taxonomic terms devised for purely practical
and descriptive purposes may not conform to the true essence
of speech, just as, for example, a pretheoretic category for living
animals of those that fy would result in a heterogeneous class
that included birds, bats, fying fsh, and winged insects to the
exclusion of penguins, ostriches, emus, and kiwis.
Universals Deriving from Speech Aerodynamics
All languages have consonants and vowels. Among conso-
nants, all languages employ stops. Among stops, voiceless stops
are the default; that is, if a language employs voiced stops it
will also have voiceless stops, but not the reverse. Tis can be
explained by the aerodynamic voicing constraint (AVC) (Ohala
1983 ): Voicing requires air fow through the approximated vocal
cords, and this requires a positive pressure diferential between
the subglottal and the oral air pressures. During obstruents, the
fowing air is blocked by the consonantal closure so that air accu-
mulates in the oral cavity, thus increasing the oral air pressure
above the glottis such that eventually the required pressure dif-
ferential diminishes, thereby reducing transglottal airfow below
the level needed for vocal cord vibration. Another universal pat-
tern explained in part by the AVC is that among languages that
do have voiced stops, it is often the case that the back-articulated
stop is missing, for example, as in Dutch and Tai. Tis is because
insofar as the AVC can be ameliorated, it is due to the compli-
ance of the surfaces of the vocal tract to the impinging oral air
pressure. Te magnitude of this compliance is greatest for labial
obstruents (due to expandability of the cheeks), less for apicals,
and least for velars, which have the least surface area exposed
to the oral pressure. Tese factors also help to explain the kind
of sound change that occurred in Nubian, now manifested as
a morphophonemic alternation, whereby geminated voiced
stops become voiceless at all places of articulation except labial
( Table 1 ; data from Bell 1971 ).
Among languages that have both voiced and voiceless stops,
there are many that have only voiceless fricatives (e.g., Tai,
Galician, Taba). Again, the AVC is part of the explanation: As
Bellugi , U ., And M. Studdert-Kennedy . 1980 . Signed and Spoken
Language: Constraints on Linguistic Form . ( Dahlem Konferenzen.)
Weinheim, Germany : Verlag Chemie GmbH .
Carr , Ren, B . Lindblom , and P. MacNeilage . 1995 . Rle de lacoustique
dans lvolution du conduit vocal humain. Comptes Rendus de
lAcadmie des Sciences (Paris) t 30, srie Iib: 471 76.
Catford , John C . 1982 . Fundamental Problems in Phonetics .
Bloomington: Indiana University Press .
Donald , Merlin . 1991 . Origins of the Modern Mind . Cambridge : Harvard
University Press
Fitch , W . Tecumseh . 2000 . Te evolution of speech: A comparative
review. Trends in Cognitive Science 4 .3: 25867.
Hewitt , G. , A. MacLarnon , and K. E. Jones . 2002 . Te functions of laryn-
geal air sacs in primates: A new hypothesis. Folia Primatol 73 : 70 94.
Hurley , Susan , and Nick Chater . 2005 . Perspectives on Imitation: From
Neuroscience to Social Science . Vols. 1, 2. Cambridge, MA : MIT Press .
Kent , Ray D . 2004 . Development, pathology and remediation of
speech. In From Sound to Sense: 50+ Years of Discoveries in Speech
Communication , ed J. Slifka et al. Cambridge, MA : Research
Laboratories of Electronics, MIT .
Kiparsky , Paul . 1988 . Phonological change. In Linguistics: Te Cambridge
Survey . Vol. 1. Ed. F. J. Newmeyer , 363415. Cambridge : Cambridge
University Press .
Lieberman , Philip . 1991 . Uniquely Human . Cambridge : Harvard
University Press.
Lindblom , Bjrn . 1990 . Explaining phonetic variation: A sketch of the H&H
theory. In Speech Production and Speech Modeling, ed. W. Hardcastle
and A. Marchal, 40339. Dordrecht, the Netherlands : Kluwer .
. 2007 . Te target hypothesis, dynamic specifcation and seg-
mental independence. In Syllable Development: Te Frame/Content
Teory and Beyond , ed. B. Davis and K. Zajd . Hillsdale, NJ : Lawrence
Erlbaum .
Lindblom , B ., and I. Maddieson . 1988 . Phonetic universals in conso-
nant systems. Language, Speech and Mind , ed. Larry M. Hyman and
C. N. Li , 6278. London and New York : Routledge .
MacNeilage , Peter F . 2008 . Te Origin of Speech. New York : Oxford
University Press .
Miller , George A . 1991 . Te Science of Words . New York : Freeman .
Oudeyer , Pierre-Yves. 2006 . Self-Organization in the Evolution of Speech .
New York : Oxford University Press .
Stevens , Kenneth N . 1989 . On the quantal nature of speech. J Phonetics
17 : 3 46.
Studdert-Kennedy , Michael. 2005. How did language go discrete? In
Language Origins: Perspectives on Evolution , ed. Maggie Tallerman ,
4867. New York : Oxford University Press .
PHONOLOGY, UNIVERSALS OF
Phonological universals are those aspects of languages sound
system that are found either in every or most human lan-
guages or in diverse languages where their presence cannot be
accounted for by inheritance from a common parent language,
geographical proximity, or borrowing. Tey are often referred
to as unmarked or default conditions in languages phonologies
when these terms imply reference to common cross-language
patterns. Tere are universal patterns in languages 1) sound
inventories, including their prosodies, 2) sequential constraints
(how sounds are sequenced), and 3) sound changes and the
phonological alternations they create within a given language.
Tey are of interest because they give insight into the physical
factors that shape human speech, help to elucidate mechanisms
of sound change, and, perhaps, suggest something about the
Table 1
Noun stem Stem + and English gloss
/fab/ /fab:n/ Father
/sgd/ /sgt:n/ Scorpion
/ka/ /ka: n/ Donkey
/mg/ /mk:n/ Dog
Phonology, Universals of
619
glides following, unless blocked by a buccal obstruent (that is,
one made in the oral cavity from the uvular-velar region to the
lips). Nonbuccal obstruents such as the glottal fricative [h] or
the glottal stop [] do not block it. Tis follows from a straightfor-
ward physiological constraint: Buccal obstruents, insofar as they
require the buildup of oral pressure, cannot tolerate venting of
this pressure via an open velic port. Te nonbuccal obstruents
require a pressure buildup in a cavity that does not access the
velic port, and so whether the velic port is open or closed is irrel-
evant to their production .
Among fricatives, the most common are the apical s -like fric-
atives (Maddieson 1984 ). Tis stems from a combination of aero-
dynamic and anatomical factors. Apical fricatives have relatively
long and intense noise in the high frequencies (3 to 8 kHz) and
are thus easily detected and are distinct from all other speech
sounds. Tis is due to the fact that the approximation of the
tongue apex at or near the alveolar ridge enables the generation
of a relatively focused high-velocity air jet, which itself generates
noise, but the air jet is also directed at the incisors, which act as a
baf e and cause the generation of more high-frequency noise as
the air hits the teeth surface (this is why s sounds are impaired
in the speech of juveniles when they lose their primary teeth and
before the growth of their permanent teeth). Additionally the
small space between the tongue apex and the lips constitutes a
resonator that reinforces high frequencies .
Te existence and properties of a resonator downstream of
the point where turbulent noise is generated underlies another
marked asymmetry in the incidence of stop types. We saw previ-
ously that in languages that have both voiced and voiceless stops,
the voiced velar stop [g] is often missing. Among voiceless stops,
the bilabial [p] is often missing (Sherman 1975 ), for example, in
Arabic and in Aleut (except for loanwords) and in Proto-Celtic.
Noise generated by the air turbulence at the lips have no down-
stream resonator to amplify the noise.
mentioned, optimal conditions for voicing require oral pressure
as low as possible (with respect to subglottal pressure), but opti-
mal conditions for generating frication (turbulence) at an oral
constriction requires the oral pressure as high as possible (with
respect to atmospheric pressure). Tese conditions are con-
tradictory. Tus, voiced fricatives are less common than voice-
less ones. Phonetically, in languages that have both voiced and
voiceless fricatives (e.g., English, French, Italian, etc.), the frica-
tion noise of voiced fricatives is always less than that for voiceless
fricatives .
Although all languages have voiced vowels, some languages
feature voiceless vowels as well, though often these are con-
textually determined, for example, word fnally or in the envi-
ronment of voiceless consonants. In any case, it seems to be
always the case that a voiceless vowel has a voiced counterpart.
J. H. Greenberg ( 1969 ) provided a survey of the incidence of
voiceless vowels in several languages. He found a virtually uni-
form pattern: Voiceless vowels appear as the counterparts to
vowels higher in the vowel space. (See Table 2 .) Te explanation
for this also requires reference to the AVC. Among vowels, high,
close vowels like [i] and [u] are almost obstruents . If articulated
suf ciently close, they impede the exiting airfow almost as much
as fricatives. Tis, in combination with other factors that could
create a slightly open glottis via coarticulation, such as appear-
ing in word- (and thus utterance-) fnal position or near voice-
less obstruents, can lead to the vowel being voiceless. Te same
factors apply to glides (approximants) that are high, close like
[j], [w], and [] and account for the frequent devoicing and frica-
tivization that gives rise through sound change to cases like the
dialectal alternations in English as Tuesday [t
h
juzdi] ~ [t
h
uzdi],
and lieutenant [lwtnnt] ~ [lftnnt] and truck [t
h
k]~[t
h
k]
(and similar patterns in many other languages). Te same factors
frequently lead to the africation of stops before high close vow-
els or glides as in Japanese, for example, or the sound change that
converted Benjamin Franklins natural [ntjul] to the modern
pronunciation [n t
l].
Aerodynamic factors also explain patterns of nasal prosody
in languages as diverse as Sundanese (spoken in the Indonesian
archipelago) and Tereno (spoken in the Mato Grosso, Brazil).
As shown in Table 3 , in these (and other) languages, the pres-
ence of a nasal consonant induces nasalization on all vowels and
Table 3.
a. Sundanese:
n to wet
b h r to be rich
h kn to inform
msih to love
b. Tereno:
1st person 3rd person
piho
m
biho I/he went
ahjaao a
n
aao I/he desire(s)
iso
n
zo I/he hoed
owoku w
gu my/his house
ajo j o my/his brother
emou m my/his word
iha
n
za my/his name
Sources : For Sudanese: Robins 1957; for Tereno: Bendor-Samuel 1960, 1966.
Table 2.
Language Voiceless and voiced Voiced only
Awadi i, u, e a, o
Campa i o, e, a
Chatino i, u o, e, a
Dagur i, u, e o, a
Huichol i,
, e u, a
Serbo-Croatian i, u e, o, a
Tadjik i, u, a e, o, u
Tunica u i, e, , a, , o
Uzbek i, u e, , o, a
Phonology, Universals of
620
continuant) + glide + vowel and at syllable ofset, the reverse.
Te English words swamp and tryst , the French words plume
[plym] and soir [swa], and the Czech Psov name of a city
[psf] would thus adhere to this generalization. But there are
reasons to be skeptical of the sonority hierarchy. First, there is
no empirical content to the term sonority ; it has never been ade-
quately defned. Second, it ignores such very common clusters
as /sp/, /st/, and so on in syllable initial position and /ps/, /ts/,
and so on in syllable fnal position. Tird, it ignores cross-lan-
guage prohibitions of onset sequences like /tl/, /dl/, /ji/, /wu/,
/t
w
u/, and /b
j
i/, that is, sequences that have similar elements.
John Ohala and H. Kawasaki-Fukumori ( 1997 ) suggest replac-
ing a) the one-dimensional concept of sonority with a multidi-
mensional measure, where similarity of sounds is a function of
acoustic amplitude, formant frequencies and spectral shape in
general, degree of periodicity (whether from fricatives or stop
bursts), and even fundamental frequency; and b) the notion of
the fxed hierarchy with a measure of the degree of similarity of
sounds according to (a). Te more two sounds are similar, the
less common would such sequences be found; the greater the
diference in sounds, the more common. By this criterion, initial
sequences like /sp/and fnal sequences like /-kst/ (in English
text ), and so forth are normal, and initials /tl/, /ji/, /wu/, and so
forth are less preferred .
Sound Changes
Of the thousands of regular sound changes that have been
identifed using the comparative method in historical pho-
nology, certain ones are recognized as showing independent
cross-language incidence. One such is velar palatalization , k >
t, t
, s, /__i (j), (and similar changes involving the voiced velar
/g/), for example, English cheese [t
h
iz]] from Latin caseus (cf.
Dutch kaas ); Ikalanga [t
i-ledu] chin < Proto-Bantu *ki-dedu .
Traditionally, the causes of sound change were attributed to two
opposite tendencies: speakers striving for ease of articulation,
which would lead to assimilations and reductions, and speak-
ers striving to speak more clearly, which would lead to exag-
geration of articulation and augmentation of pronunciation.
Tere is no doubt that speakers do alter their pronunciation
in these manners, but it may be seriously questioned whether
Virtually all languages employ nasal consonants (Ferguson
1963 ). However, there are never more place distinctions among
nasals than there are obstruents, and there are often fewer. Te
acoustics of nasals probably account for this. All nasal conso-
nants have in common the pharynx-plus-nasal air space. What
diferentiates one nasal consonant from another is the efect of
the oral cavity, which branches of the nasal-pharyngeal cavity
(see Figure 1). Tus, although nasals are highly distinct as a class
from non-nasals, they are auditorily very similar to one another.
Tis also partly accounts for the frequent pattern whereby nasals
assimilate in place to a following stop, for example, English
incredible [ik
h
bl ]< in (neg. prefx) + credible ; Latin qunctus
> quntus (where original n = [] > [n] / __t).
An interesting cross-language pattern is the character of nasal
assimilation to labial velar consonants such as [k
p ], [ b ] and
[w], that is, segments that have equal constrictions in the labial
and velar region. Te nasal that appears before such segments
is invariably a velar [] not the labial [m], for example, Kpelle
and Melanesian ( Table 4 ). Te explanation for this pattern can
be seen in Figure 1. What matters for the place of articulation of
a nasal consonant is the frst buccal constriction encountered
from the nasal pharyngeal cavity. In a labial velar, this is the velar
constriction; the labial constriction, being beyond that, is acous-
tically largely irrelevant (Ohala and Ohala 1993 ) .
Phonotactics
Te conventional view of common cross-language sound
sequencing or phonotactics is couched in terms of whats
called the sonority hierarchy (attributed to E. Sievers and O.
Jespersen), whereby the favored pattern at syllable onset shows
sounds sequenced in the following order (where omissions
are possible): stop + fricative + nasal + liquid (i.e., non-nasal
Table 4.
Kpelle: [w] patterns with velars in nasal assimilation:
Indefnte Defnite
`m i wax
lu `nui fog, mist
l `il dog
we `wei white clay
Notes: Melanesian: m > / __w:
Common Melanesian /limwa/ hand ~ Fijian /linga/ (= phonetic [li
w
a])
/mala/ ~ /mwala/ ~ /wala/ (name of the Mala Island in different dialects of
the island)
Sources: For Kpelle: Welmers 1962; for Melanesian: Ivens 1931.
Figure 1. A schematic representation of the resonating cavities during the
production of different nasal consonants. The solid line demarcates the
main pharyngeal-nasal cavity, which is the same for all such nasals. What
differentiates one nasal from another is the effect produced by the oral
resonator, which branches off this main cavity. Even though a labial velar
consonant has two main constrictions, it is only the rearmost, the velar
constriction, that matters and thus sounds similar to the velar nasal [].
Phonology, Universals of
621
S. Eliasson and E. H. Jahr , 34365. Trends in Linguistics: Studies and
Monographs, Vol. 100. Berlin : Mouton de Gruyter .
Ohala , J. J ., and J. Lorentz . 1977 . Te story of [w]: An exercise in the
phonetic explanation for sound patterns . Berkeley Linguistics Society,
Proceedings, Annual Meeting 3 : 577 99.
Ohala , J. J ., and M. Ohala . 1993 . Te phonetics of nasal phonol-
ogy: Teorems and data. In Nasals, Nasalization, and the Velum , ed.
M. K. Hufman and R. A. Krakow , 22549. San Diego, CA : Academic
Press .
Pertz , D. L ., and T. G. Bever . 1975 . Sensitivity to phonological universals
in children and adolescents . Language 39 : 347 70.
Robins , R. H . 1957 . Vowel nasality in Sundanese. In Studies in Linguistic
Analysis , 87103. Oxford : Blackwell .
Sherman , D . 1975 . Stop and fricative systems: A discussion of paradig-
matic gaps and the question of language sampling . Stanford Working
Papers in Language Universals 17 : 1 31.
Welmers , W. E . 1962 . Te phonology of Kpelle . Journal of African
Languages 1 : 69 93.
Winitz , H. , M. E. Scheib , and J. A. Reeds . 1972 . Identifcation of stops
and vowels for the burst portion of /p,t,k/ isolated from conversation
speech . Journal of the Acoustical Society of America 51.4 : 1309 17.
PHRASE STRUCTURE
It is an ancient observation that natural language syntax is hier-
archically organized. As can be seen from a variety of diagnostics,
the words comprising a sentence do not behave as beads on
a string but group into successively larger units, or constituents.
Phrase structure (PS) is a formal representation of this constit-
uent structure . PS is typically depicted as a tree-structured
graph ( Figure 1), which encodes three sorts of structural informa-
tion: i) dominance , specifying the words and constituents that a
constituent contains within it (e.g., as shown by vertical placement
in the fgure, prepositional phrase (PP) dominates on and televi-
sion ); ii) precedence , specifying the temporal orderings among the
words and constituents (e.g., as shown by horizontal position, the
constituent most fans precedes the constituent watched the game
on television ); and iii) labeling , specifying the grammatical cate-
gory of each word and constituent (e.g., the constituent the game
is a noun phrase (NP)). In PS-based approaches, this structural
information plays an important role in defning the conditions
under which grammatical dependencies may obtain (see agree-
ment , anaphora, binding , and case ), and PS is often taken to
be the input to transformational operations (see movement and
transformational grammar ). Further, PS representations
serve as the interface between syntax and semantics , as they pro-
vide the structural information necessary for interpretation (see
compositionality, thematic roles , and logical form ) .
A fundamental question concerns how the range of possible
PS is specifed in a grammar. Te earliest answer comes from
Noam Chomsky ( 1957 ), who suggests that PS is generated by a
set of phrase structure rules, like the following:
1. S NP VP
2. NP N
3. NP Det N
4. VP V NP
5. VP VP PP
6. PP P NP
these changed forms replace previous norms of pronunciation.
Tere is no evidence for this. Tere is, however, an alternative
scenario of sound change that does have empirical support: lis-
teners errors. Tere have been numerous speech-perception
experiments, some involving natural speech, which revealed
errors that mirrored sound change; for example, in a study in
Winitz Scheib, and Reeds ( 1972 ) where listeners heard a frag-
ment of consonant/vowel (CV) syllables, [k
h
i] was misidentifed
as [t
h
i] 47 percent of the time, paralleling the change in place
found in velar palatalization. Ohala ( 1981 ) has elaborated a
theory of sound change based on listeners misperception or
misparsing of the speech signal. Such common sound change
includes VN > V , for example, Sanskrit dant tooth > Hindi /
dt/, Latin bon - good > French /b/, and the assimilation of
place in C
1
C
2
consonant clusters, Latin scriptu > Italian scritto ,
English congress [k
h
agrs] < (ultimately) Latin com - together
+ gradi to walk.
Phonological Universals and Universal Grammar
It has also been proposed that phonological universals arise from
humans genetic endowment in the form of whats called univer-
sal grammar (Pertz and Bever 1975 ). Such claims have been dis-
puted by those who fnd phonological universals rooted in the
physical and physiological attributes of all human speakers and
hearers. Te dust has not settled on this issue as yet .
John Ohala
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bell , H . 1971 . Te phonology of Nobiin Nubian . African Language
Review 9 : 115 59.
Bendor-Samuel , J. T . 1960 . Some problems of segmentation in the pho-
nological analysis of Tereno . Word 16 : 348 55.
. 1966 . Some prosodic features in Terena. In In Memory of J. R.
Firth , ed. C. E. Bazell, J. C. Catford , M. A. K. Halliday , and R. H. Robins ,
309. London : Longmans .
Ferguson. C. A . 1963 . Some assumptions about nasals. In Universals of
Language , ed. J. H. Greenberg , 427. Cambridge, MA : MIT Press .
Greenberg , J. H . 1969 . Some methods of dynamic comparison in linguis-
tics. In Substance and Structure of Language , ed. J. Puhvel , 147203.
Berkeley and Los Angeles : University of California Press .
. 1970 . Some generalizations concerning glottalic consonants,
especially implosives . International Journal of American Linguistics
36 : 123 45.
Greenberg , J. H , C. A. Ferguson , and E. A. Moravcsik , eds. 1978. Universals
of Human Language. Vol. 2. Phonology . Stanford, CA : Stanford
University Press .
Ivens , W. G . 1931 . A grammar of the language of KwaraAe, North Mala,
Solomon Islands . Bulletin of the School of Oriental Studies 6 : 679 700.
Maddieson , I . 1984 . Patterns of Sounds . Cambridge : Cambridge University
Press .
Ohala , J. J . 1981 . Te listener as a source of sound change. In Papers
from the Parasession on Language and Behavior , ed. C. S. Masek ,
R. A. Hendrick , and M. F. Miller , 178203. Chicago : Chicago Linguistics
Society .
. 1983 . Te origin of sound patterns in vocal tract constraints.
In Te production of Speech , ed. P. F. MacNeilage , 189216. New
York : Springer-Verlag .
Ohala , J. J ., and H. Kawasaki-Fukumori . 1997 . Alternatives to the sonor-
ity hierarchy for explaining segmental sequential constraints. In
Language And Its Ecology: Essays In Memory Of Einar Haugen , ed.
Phonology, Universals of Phrase Structure
622
(Baker and Muhlhausler 1990 ). Pidgins used widely for trading
purposes (but not limited to such functions) include Russenorsk,
Chinese Pidgin English, and Chinook Jargon, once widely used in
the American Northwest. Other pidgins arose where large num-
bers of slaves and/or indentured laborers had to work together
on colonial plantations. Such pidgins were usually short-lived,
evolving into creole languages; contrary to some claims, careful
examination reveals manifold signs of their pidgin ancestry, and
the earliest attestations of some (Baker and Corne 1982 ; van den
Berg 2000 ) show pidgin-like structures. Although the reality of this
pidgin-to-creole cycle has been denied (see entry on creoles ),
massive empirical evidence exists in Hawaii, as described by S. J.
Roberts ( 1995 , 1998 ); there is also evidence of prior pidginization
in other creoles, such as fossilized sequence markers and marked
vocabulary mixture (see the following examples).
As compared with natural languages (including creoles), all
pidgins are severely impoverished, with sharply reduced vocab-
ularies, few structural consistencies, and few if any infectional
af xes; complex sentences very seldom occur. Function words
are rare, if not completely absent; categories normally expressed
via auxiliary verbs of tense, mood and aspect are indicated, if at
all, by two adverbial forms meaning roughly soon or fnish
that are attached, not adjacent to the verb as in natural lan-
guages generally, but clause-fnally or clause-initially. We fnd,
for example, baimbai (English by-and-by) and pau (Hawaian
fnished), baimbai and pinis in Tokpisin; also a number of sim-
ilar pidgin fossils in creoles (sometimes inside, sometimes still
outside the verb phrase), such as fn(i), fnish in French-related
creoles, done in English-related creoles, or ( ka)ba (Portuguese
acabar fnish) in Portuguese-related creoles .
If a pidgin persists in a relatively stable population (one not
subject to the rapid expansion and turnover that typically char-
acterize creole societies) and is widely used over a long enough
period, it may acquire a more stable (although still limited)
structure. However, pidgins still sufer from widespread misun-
derstanding of the linguistic mechanisms through which they
arise. According to many writers (e.g., Bakker 1995 ; Manessy
1995 ) they are reduced or simplifed versions of preexisting
languages, or failed attempts by speakers with inadequate access
to acquire the locally dominant language a view reinforced by
standard usage of expressions such as Pidgin English, Pidgin
French, and so on.
Pidgins do not derive from processes applied to any preexist-
ing natural language, however, but (as is clear both from histori-
cal data in Hawaii and reminiscences of older residents; see, e.g.,
Bickerton 1981 , 11) arise naturally from strategies employed by
individuals of any ethnic background in a multilingual situation
where no single existing language is both viable and accessible.
Speakers seek to communicate by any means possible, using iso-
lated words from their own language, from their interlocutors
language (if they know any), and from any third or fourth lan-
guage that they may happen to have picked up .
Tese words are seldom assembled in the way words are
assembled in modern human languages, that is, hierarchically.
Except for occasional rote-learned phrases, words are attached
sequentially, like beads on a string. Consequently, no true gram-
matical relations exist, limiting utterances to brief strings of a few
words without embedding.
In these rules, a symbol appearing to the left of the arrow can be
rewritten as the sequence of symbols to the right of the arrow. Te
process of PS generation begins with a distinguished start sym-
bol S and successively rewrites the symbols in the string using
the rules of the grammar until no rewritable symbols remain.
An example of this process follows, with the number above each
arrow indicating the rewriting rule used:
S
1
NP VP
3
Det N VP
5
Det N VP PP
4
Det N V NP PP
3
Det N V Det N PP Det N V Det N P NP Det N V Det N P N
6 2
Te PS in Figure 1 can be understood as a history of this
derivation: Te children of a node correspond to the sequence
of symbols into which that node is rewritten. Some recent
approaches have maintained rewriting as part of the grammar
but have questioned the nature of the rewrite rules employed
in this system, generalizing and modifying them in a variety
of respects (see x-bar theory and minimalism ). Other
approaches have abandoned rewriting, taking well-formed
PS representations to be those that best satisfy a set of gram-
matical constraints (see head-driven phrase structure
grammar, lexical-functional grammar , and opti-
mality theory ) .
Robert Frank
WORKS CITED AND SUGGESTIONS FOR FURTHER READING.
Baltin , Mark , and Anthony Kroch , eds. 1989. Alternative Conceptions of
Phrase Structure . Chicago : University of Chicago Press .
Chomsky , Noam . 1957 . Syntactic Structures . Te Hague : Mouton .
PIDGINS
Pidgins are the worlds only non-native languages. Tey are
typically acquired by adults, after the critical period for language
acquisition has passed. Tey normally arise wherever suf cient
speakers of mutually incomprehensible languages must inter-
act with one another. Some pidgins arose through and for trade;
the most plausible derivation ofered for the origin of the name
pidgin attributes it to the Chinese pronunciation of business
Figure 1. Phrase structure representation for Most fans watched the game
on television.
Pidgins
623
Bakker , P . 1995 . Pidgins. In Pidgins and Creoles: An Introduction , ed. J.
Arends, P. Muysken, and N. Smith, 2539. Amsterdam : Benjamins .
Bickerton , D . 1981 . Roots of Language . Ann Arbor, MI : Karoma .
Broch , I ., and E. H. Jahr . 1984 . Russenorsk: A new look at the
Russo-Norwegian pidgin in northern Norway. In Scandinavian
Language Contacts , ed. P. S. Ureland and I. Clarkson, 2164.
Cambridge : Cambridge University Press .
Gibbs , G . 1863 . A Dictionary of the Chinook Trade Jargon or Trade
Language of Oregon . New York : Gramoisy.
Manessy , G . 1995 . Croles, Pidgins, Varits Vehiculaires . Paris : CNRS
Editions .
Price , R . 1976 . Te Guiana Maroons: A Historical and Bibliographical
Introduction . Baltimore : Johns Hopkins University Press.
Reinecke , J. E . 1969 . Language and Dialect in Hawaii . Honolulu: University
of Hawaii Press .
Roberts , S. J . 1995 . Pidgin Hawaiian: A sociohistorical study . Journal of
Pidgin and Creole Languages 10 : 156.
. 1998 . Te role of difusion in the genesis of Hawaiian Creole .
Language 74: 139.
. 2005 . Te Emergence of Hawaii Creole English in the Early 20th
Century : Te Sociohistorical Context of Creole Genesis . Ph.D. diss.,
Stanford University .
Smith , N. , I. Robertson , and K. Williamson . 1987 . Te Ijaw element in
Berbice Dutch . Language in Society 16 : 4990.
van den Berg , Margot . 2000 , Mi no sal tron tongo: Early Sranan in Court
Records, 16671767. Unpublished Masters thesis, Radboud University
Nijmegen.
Wiley , J . 1995 . Te origin of Lesser Antillean French Creole: Some lit-
erary and lexical evidence . Journal of Pidgin and Creole Languages
10 : 71126.
PITCH
When an object vibrates, its movement produces changes in
air pressure that radiate like waves from the source. If the fre-
quencies of the vibrations are roughly between 20 and 20,000
cycles per second, or Hertz (Hz), ideally they can be heard by a
young, healthy human listener. Te range of hearing frequen-
cies declines with age. Te physical characteristic of the vibrat-
ing body, frequency , produces a psychological experience called
pitch. In general, a low frequency produces the sensation of a low
pitch (for example, the 60 Hz hum produced by electrical power
in a poorly grounded radio), with the pitch increasing as the fre-
quency increases (a male voice at 100 Hz, a female voice at 200
Hz, a childs voice at 300 Hz). Because there is a close corres-
pondence between frequency and pitch, people frequently use
the terms interchangeably. However, in addition to frequency,
the sensation of pitch is infuenced by an interaction between
the amplitude of the vibration and the range of the frequency.
Pitch is also infuenced by the complexity of the vibration and its
corresponding wave form.
A vibrating body oscillates as a single entity, producing a fre-
quency referred to as the fundamental frequency . So, when the
key for A above middle C is played on a piano, a string vibrates
at 440 Hz. Vibrating bodies are not perfectly rigid, though, and
the string also vibrates in parts as if it is two strings (producing
a frequency of 880 Hz), and three strings (1320 Hz), etc. Tus, a
vibrating body produces a series of frequencies beginning with
the fundamental frequency (f0) and including its harmonics,
which are multiples of the f0. Te distribution of acoustic energy
Te degree to which pidgins (and subsequent creoles) show
lexical mixture has been underestimated in the literature.
Noteworthy are Russenorsk (with roughly equal quantities of
Norwegian and Russian words, but also 14% of its vocabulary
drawn from other languages; Broch and Jahr 1984 ) and Chinook
Jargon (only 41% from Chinook, with at least 11 other languages,
European and non-European, contributing to the remainder;
Gibbs 1863 ). Te baragouin that preceded the formation of the
Lesser Antillean French Creoles showed a similar mixture (Wylie
1995 ) .
Evidence from creoles suggests that the pidgins they evolved
from had equally mixed vocabularies. Berbice Dutch draws 27
percent of its vocabulary from one African language, Ijaw (Smith,
Robertson, and Williamson 1987 ). Saramaccan may have as
many as 50 percent African words (Price 1976 ). Comparison of
Saramaccan and Sranan vocabularies shows that these creoles,
both derived from the same pidgin, difer in perhaps as many as
75 percent of their vocabulary items; contra most sources, rela-
tively few of these diferences involve a Portuguese/English con-
trast, strongly suggesting an antecedent macaronic pidgin that
drew on English, Dutch, and Portuguese, as well as a variety of
African and Amerindian languages, and from which Sranan and
Saramaccan each made a diferent selection .
Why pidgins have so often been regarded as simplifca-
tions of particular (almost invariably, European) languages is
revealed by the massive database of contemporary citations
gathered by Roberts (summarized in Roberts 1995 , 1998 . 2005 ,
but not yet published in its entirety). From this data, it is clear
that pidgin descriptions have been shaped by observer bias.
Most citations from English-language sources contain a pre-
ponderance of English words, showing why J. E. Reinecke ( 1969 )
and others characterized the lingua franca of early Hawaii as a
predominantly English pidgin with a sprinkling of Hawaiian
words. However, the abundant Hawaiian-language sources
reverse this picture, presenting a predominantly Hawaiian
vocabulary with a sprinkling of English words, while the much
sparser Japanese-, Chinese-, and Portuguese-language sources
each contain a higher admixture of their own languages (brief
sentences containing words from three diferent languages are
by no means uncommon). Clearly, in a pidgin situation, observ-
ers record what they best understand and downplay or ignore
the rest.
Te well-attested existence of a pidgin phase in the life cycle
of creoles also helps to explain the strong structural similari-
ties that hold between creoles of widely diferent provenance.
For such similarities to arise, input had frst to be reduced to an
abnormally low level of structure, forcing children to draw on
their innate language faculty for the systematic structures that a
pidgin can manage without, but that are essential for any natural
language .
Derek Bickerton
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baker , P ., and C. Corne . 1982 . Isle de France Creole . Ann Arbor,
MI : Karoma .
Baker, P., and P. Muhlhausler. 1990. From business to pidgin . Journal
of Asian Pacifc Communication 1 : 87115.
Pidgins Pitch
624
In sum, the perception of pitch can play linguistic and para-
linguistic roles at the suprasegmental and segmental levels of
utterances. Pitch is closely related to the physical stimulus fre-
quency, but as a psychological event, it is infuenced by the com-
plexity, frequency range, and loudness of the tone. Pitch can be
processed in a low-resolution mode at many levels in the ner-
vous system or at high-resolution mode in specialized areas of
the cerebral cortex of the brain in the right temporal lobe. Pitch
can also be processed in linguistic and nonlinguistic modes by
the left and right temporal lobes of the brain, respectively. Te
variation of pitch during fuent speech can be considered a truly
integrative process that conveys both linguistic and paralinguis-
tic information .
John Sidtis
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Crystal , David . 1987 . Te Cambridge Encyclopedia of Language . New
York : Cambridge University Press . Presents multiple brief discussions
of pitch in diferent linguistic roles.
Sidtis , John J . 1980 . On the nature of the cortical function underlying right
hemisphere auditory perception . Neuropsychologia 18 . 3 : 32130.
Sidtis , John J. , and D. Van Lancker Sidtis . 2003 . A neurobehav-
ioral approach to dysprosody . Seminars in Speech and Language
24 . 2 : 93105. Describes how diferent aspects of prosody rely on dif-
ferent brain structures.
Van Lancker , Diana , and V. A. Fromkin . 1973 . Hemispheric specializa-
tion for pitch and tone : Evidence from Tai. Journal of Phonetics
1 : 101 9.
POETIC FORM, UNIVERSALS OF
A poetic universal is manifested by a feature that is found very
widely (for example, rhyme; see rhyme and assonance ) or by
a relation between features that is found very widely (for exam-
ple, rhyme is generally found in verse, not in prose). Like linguis-
tic universals, poetic universals might be studied by comparative
work (often depending on fairly salient features) or by focused
work on the abstract forms (hypothesized abstract universals)
underlying the surface poetic forms in a particular language.
A theory of universals can be formulated in terms of universal
parameters (sets of related formal options) from which a specifc
poetic tradition makes specifc choices. Unlike a language, where
only one choice can be made, a literary tradition can divide into
subtraditions, each making a diferent choice (thus, for example,
classical Sanskrit literature includes quite diferent kinds of met-
rical verse). Tere is no presupposition that diferent modalities
will throw up signifcantly diferent universals; thus, the gen-
eral assumption is that written, oral, and signed literatures will
have similar characteristics (see oral composition and sign
languages) . Te term literature is here used interchangeably
with verbal art and should not be used to imply a special status
for written literature.
With the exception of folklore studies (which, however, tend
to have a narrow areal range), no discipline or subdiscipline
takes as its responsibility the investigation of poetic universals.
Some researchers are actively hostile to universals in favor of
an alternative emphasis on the special characteristics of each
tradition, and some feldworkers ignore verbal arts when they
across the harmonic series contributes to the quality or timbre
of the sound. In addition to sound quality, the harmonics con-
tribute signifcantly to the perception of pitch. Te fundamental
frequency of a harmonic series can be artifcially removed with-
out changing the pitch, a demonstration referred to as the miss-
ing fundamental . In speech, the harmonic series is a function of
the complex wave produced by the glottal source. Formants are
bands of resonance that concentrate the acoustic energy pro-
duced by the glottal source as a function of the vocal tract confg-
uration and have center frequencies that refect the vocal tract,
rather than the harmonic series .
Pitch can be experienced from pure tones (f0 alone), which
do not occur naturally, as well as from complex tones (f0 + har-
monic series), but the pitch of complex tones is a stronger per-
cept, allowing fner discriminations of f0 frequency diferences.
Te processing of pitch from a determination of the f0 versus
the pattern recognition of a harmonic series relies on difer-
ent neurological systems. Simple frequency determination can
occur at multiple levels of the nervous system, but complex pitch
processing occurs in auditory areas in the right cerebral hemi-
sphere that complement speech and language areas in the left
cerebral hemisphere (Sidtis 1980; see also right hemisphere
language processing and left hemisphere language
processing ) .
For practical purposes, pitch in language can be viewed as a
direct function of f0. Unlike the pitch distinctions made in music,
linguistic pitch distinctions are comparatively coarse. Whereas
a musical octave can be divided into 12 semitones, and vibrato
in experienced singers can be consistently less than a semitone,
most linguistic communicative situations only require distinc-
tions of three semitones or more. Further, the pitch distinctions
in language are relative, allowing men, women, and children to
make the same linguistic and paralinguistic distinctions despite
diferent vocal f0s (see paralanguage ), whereas pitch distinc-
tions in music reference specifc frequencies (e.g., a musical
scale tuned to 440 Hz).
A number of linguistic and paralinguistic phenomena are
provided by pitch. At the suprasegmental level, pitch produces
the melodic line of an utterance to convey linguistic intona-
tion (e.g., declination efect: falling pitch anticipating the end
of a statement, rising pitch indicating a question), sociolin-
guistic information (e.g., uptalk , rising pitch at the end of a
statement, falling pitch as a cue for turn-taking), and paralin-
guistic information (e.g., emotion, attitude). Pitch can also be
used with loudness to provide syllable accent at the segmen-
tal level .
Pitch also has a lexical role in tone and pitch accent lan-
guages. Tone languages may have many tone patterns (esti-
mates vary, but the numbers are fewer than the 12 notes in the
musical octave), and they tend to fall in relative categories like
high, medium, and low, further distinguished by rising and fall-
ing patterns. Because such distinctions are relative, the listener is
required to perform a tone normalization to identify a speakers
lexical tones. Just as simple and complex pitch perception rely
on diferent brain mechanisms, the processing of pitch for lin-
guistic and nonlinguistic purposes engage diferent neurological
systems, principally the temporal lobes in the left and right
cerebral hemispheres (Van Lancker and Fromkin 1973 ) .
Pitch Poetic Form, Universals of
625
Tis generalization also draws attention to the fact that literary
practices can have features in common because of imitation
of an admired writer or foreign tradition. Finally, generaliza-
tion 5 is not true of all ways of writing verse, but we might ask
whether the wide acceptance of this practice tells us some-
thing about the cognitive status of lines (e.g., that we cognize
each line as a separate, isolated unit). Tese are the kinds of
questions we might ask in exploring the possibility of univer-
sals of poetic form.
Relative to the line, six categories of poetic form might be iden-
tifed, which follow; there may be others (such as the tendency
to use specifc words at line edges), and the grouping in this list
depends on theoretical assumptions and is not simply given to
us by the data. For each of the categories, we might explore its
status as a universal. None of these kinds of form is required in a
verse tradition. Most verse traditions are either metrical (i.e., they
involve the counting of syllables) or parallelistic, which is itself
an interesting universal. Either metrical or parallelistic verse can
also have rhyme and alliteration , though rhythm and word
boundary rules are usually found only in metrical verse. Tere
are some verse traditions, such as modern free verse, that do not
consistently manifest any of these categories of poetic form (but,
as noted, they may manifest other categories of poetic form, such
as the tendency to use particular words or particular syntactic
structures at line edges).
1. Te counting of syllables: In all metrical verse, the
line has a specifc number of syllables (Irish deibhidhe has 7,
Icelandic drttkvaett has 6) or a defned range of possible num-
bers of syllables (English iambic pentameter is normatively 10
but permits 911, French alexandrin 1213, Homeric dactylic
hexameter 1317, Japanese haiku 35 or 47 in diferent lines,
etc.). Some theories of meter suggest that units other than syl-
lables can also be counted, such as morae (subsyllabic units)
or larger groupings of syllables. Across literary traditions, we
fnd that not all syllables in the line are counted for metrical
purposes; in particular, when a vowel-fnal syllable precedes a
vowel-initial syllable, many traditions permit or require these
to count as a single metrical syllable. Various other generaliza-
tions can be made about the counting of syllables, which may
be the source of universals (this is the basic claim of Fabb and
Halle 2008 ).
2. Te patterning of syllables, requiring a division of sylla-
bles into two classes for metrical purposes: Accentual rhythms
manifest this type of patterning, where syllables are distin-
guished into two classes as stressed versus unstressed and pat-
terned on this basis, for example, into triplets where every third
syllable is stressed. More generally, most kinds of metrical verse
divide syllables into two classes, on the basis of stress, syllable
weight, lexical tone, or whether they alliterate and on possibly
other characteristics yet undiscovered); the class membership
of a syllable then admits it to specifc positions within the verse
line. In some cases, the distribution of the two types of syllable is
periodic (e.g., a regular recurrence as in an iambic rhythm) and
in other cases partially periodic or apparently nonperiodic (as
in the superfcially aperiodic sequences of heavy and light syl-
lables required in Classical Sanskrit verse). An interesting rhyth-
mic universal is that syllables are divided into just two classes
for metrical purposes, even when there would be a basis in the
gather information about a language; some missionary linguists
have even been known to displace indigenous verbal arts with
hymns or Bible stories, with the odd result that it is these, rather
than indigenous texts, that are gathered in grammars and other
reports. Tus, there has been relatively little work, either descrip-
tive or theoretical, on poetic universals.
Te universal of poetic form that is most widely manifested,
and may indeed be found everywhere, is the possibility of verse,
as a way of organizing language. A text that is in verse is a text
cut into a sequence of lines (= verse lines ). A line is a section of
text that supports two or more generalizations, and the investiga-
tion of universals of poetic form is largely an investigation of the
generalizations formulated in terms of the line.
For example, Miltons poem Paradise Lost is in lines, and
so is verse. (Tis example is chosen [see Fabb 2002 ] because
in the eighteenth century some critics claimed that it was not
verse.) Here are fve generalizations that are supported by the
line in this poem: 1) Tere are 10 syllables in each line; 2) the
end of the line coincides with the end of a word; 3) while there
is a tendency for stressed syllables to be in even-numbered
positions, the line-initial position is often also occupied by a
stressed syllable; 4) the word of is found with greater than
expected frequency as the frst word in a line, as seen, for exam-
ple, in the frst two lines of the poem; and 5) the printed form of
the text arrays lines in a vertical sequence. Te study of poetic
universals seeks to establish the distribution of each of these
kinds of form (i.e., each of these generalizations) and then to
understand whether these distributions imply anything about
poetic universals.
What would the study of poetic universals make of these
generalizations? Generalization 1 can be understood more
abstractly as the line contains a specifc number of syllables,
and this is defnitely a universal, in the sense that it is true of
many verse traditions; however, we might also ask whether the
fact that there are specifcally 10 syllables in the line also consti-
tutes a (more narrowly manifested) universal. Generalization
2 is also very widespread; while lineation does not necessarily
respect phrase or sentence boundaries, word boundaries
are usually respected (and this connects with the fact dis-
cussed later that metrical rules control for word boundaries
but not for phrase or sentence boundaries); hence, there is an
interesting potentially universal relation between the line that
is a nonlinguistic section of text and the word that is a linguis-
tic constituent. Generalization 3 in its specifc formulation is
generally true of English, but more abstractly the possibility of
relaxing a rule at the beginning of a line is found widely; for
example, Greek verse lines often begin with a syllable whose
weight is uncontrolled ( anceps ). Generalization 4 holds true
also of eighteenth- and nineteenth-century verse after Milton;
is it telling us something signifcant about of or about the natu-
ralness of beginning a line with a preposition phrase? On the
one hand, we might say that post-Miltonic verse is just imitat-
ing a kind of form that Milton may have invented, but we might
also note that much twentieth-century free verse also favors
preposition-initial lines. It is worth noting that the Greek early
elegiac poets tended to begin and end lines mainly with words
used previously by Homer, suggesting again that choice of par-
ticular words at line edges has the potential to be a universal.
Poetic Form, Universals of
626
intervening lines) between related elements in diferent lines.
(Bruce Hayes and Margaret MacEachern 1998 discuss universals
in stanza structure.)
Are there any kinds of poetic form that are unrelated to the
line? In part, this is a matter of defnition (i.e., of whether we
intend poetic to mean verse). Clearly, various types of fgure
and trope are widely found in the worlds literatures, and not only
in verse (though these may be better understood as linguistic or
pragmatic universals, rather than poetic universals; see prag-
matics, universals in ). And there are possible universals of
narrative form (see narrative universals ) that might also be
thought of as poetic, some of which may in fact be related to
universals of verse form. It is possible that there are universals
that relate verbal art to counting (perhaps via the aesthetic,
and perhaps extending beyond verbal art). Metricality is based
on counting, as are the kinds of form closely related to metrical-
ity, such as rhythm and word-boundary placement. Parallelism
may be based on counting of a diferent kind (a tally or one-to-
one alignment). Narratives seem to involve counting at various
levels, including Dell Hymess ( 1992 ) suggestion that narratives
are structured around pattern numbers, with narrative units
organized in two and four or in three and fve in a particular
tradition.
What is the relation between poetic form and linguistic
form? One widely held view (associated, for example, with
Jakobson) is that the forms of poetry in a particular language
are dependent on the linguistic form of that language; the fxing
of a choice from a poetic parameter is thus dependent on the
fxing of a choice from a linguistic parameter. Tus, for exam-
ple, the claim might be that some languages are better suited to
quantitative meters (where the distinction between heavy and
light syllables is criterial) and others better suited to accentual
meters (where the distinction is instead between stressed and
unstressed syllables); English, for one, has successful accentual
verse, but neither nonaccentual syllable counting nor quanti-
tative meters have taken hold in the poetic tradition, despite
attempts to introduce them. Kristin Hanson and Paul Kiparsky
( 1996 ) propose a theory of poetic universals that has a param-
eter ofering a range of diferently sized phonological units that
can match metrical positions; in a specifc tradition, a specifc
size of phonological unit matches the metrical position. Tus,
for example, Chinese and Japanese verse both have fve- and
seven-unit lines, whose positions are flled by syllables in the
former and subsyllabic morae in the latter. An alternative posi-
tion is taken by Nigel Fabb and Morris Halle ( 2008 ), who argue
that poetic form and linguistic form have systemic subcom-
ponents in common, including some parameters (and indeed
share subcomponents also with music); however, there is no
necessary relation between the poetic form and the linguistic
form of a particular language .
Is verbal art itself a universal? Tat is, is there any single way
in which it can be characterized, and distinguished from gen-
eral verbal behavior? Te most commonly given answer to this
question is yes: that verbal art is distinguished from verbal
behavior because it draws attention to its own form. Tis is the
basis of Jakobsons ( 1960 ) projection principle, of Nelsons
Goodmans ( 1978 ) notion of style as exemplifed by a text, or
of Richard Baumans (1984) notion of verbal art as a text that is
language for more than two classes. For example, Vietnamese
has six types of lexical tone, which are grouped into just two
tonal classes for the purposes of metrical regulation. It has been
claimed that another rhythmic universal is based on the ( opti-
mality-theoretic ) phonological notion of the moraic
trochee as a basic rhythmic unit (see, for example, Golston and
Riad 2005 ). Robbins Burling ( 1966 ) claimed that a certain com-
bination of meter and rhythm is found universally in childrens
verse .
3. Word-boundary rules: In the metrical line, two adjacent
syllables can be required to be in separate words that is, a
word boundary must intervene (by a caesura rule); or they can
be required to be in the same word that is, a word boundary
must not intervene (by a bridge rule). Thus, for example, the
sixth syllable in a French 12-syllable alexandrin must be word-
final. Word-boundary rules are widespread, and this suggests
an underlying universal. In particular, the word seems to
have a special status in meter: Metrical rules do not control
for phrase or sentence boundaries, and this also points to a
universal .
4. Rhyme understood as the repetition of the end of the
syllable (usually including its nucleus): Rhyme is very wide-
spread, including in nonmetrical verse: Parallelistic verse can
have rhyme, and we may even fnd rhyme of a kind in prose.
Rhyme would seem to manifest a universal. Furthermore, it is
cross-linguistically true that sound sequences can be counted
as rhyme that are phonetically dissimilar but share underlying
similarities; this possibility, and perhaps the way in which dis-
similar phonetic sequences are admitted as equivalent, may
manifest universals .
5. Alliteration: U nderstood as the repetition of the beginning
of the syllable (sometimes including its nucleus), alliteration is
much rarer than rhyme, which may itself tell us something about
poetic universals. Te fact that words beginning with dissimilar
vowels are considered to alliterate in separate traditions (e.g.,
Old English and Somali) may suggest a universal. Alliteration
also appears to be subject to locality constraints that do not hold
for rhyme; thus, alliteration tends to be line-internal or between
adjacent lines and does not interlace as rhyme does in ABAB
structures (Fabb 1999 ).
6. Parallelism: Tis formal property is very widespread in the
literatures of the world, and Roman Jakobson ( 1960 ) thought of
it as a defning formal characteristic of poetry because it draws
attention to form by repeating it (he included meter, rhythm,
rhyme, and alliteration as types of parallelism). Tere are dif-
ferent kinds of parallelism, all quite widely distributed, includ-
ing parallelism of sound sequences, parallelism of words, and
parallelism of syntactic structures. Universals have yet to be
established.
Lines may be organized into larger units, such as stanzas. Te
possibility of organizing lines into stanzas is suf ciently wide-
spread as to count as a universal. Stanzas have characteristics,
such as having a specifc number of lines, lines of the same or
varying lengths, or rhyme. While there is clearly much variety,
universals may be discovered, perhaps involving the way lines
are counted in a stanza, or the possible ways in which rhyme
patterns can be structured. For example, there may be locality
efects, such as limits on the possible distance (e.g., number of
Poetic Form, Universals of
627
In Te Origins of Consciousness in the Breakdown of the
Bicameral Mind ([1978] 1990), psychologist Julian Jaynes
claimed that Homers Iliad and the oldest books of the Hebrew
Old Testament portrayed human beings in a twilight state of
awareness. According to Jaynes, the characters in the worlds
most ancient poetry take action not as the result of personal
thought and conscious decision but because they hear the voice
of a god ordering them to do so. He hypothesized that the voices
of the gods were actually auditory hallucinations produced in the
brains right temporal lobe . Transmitted to the left temporal
lobe, seat of left hemisphere language processing , they
were perceived as coming from outside the self. Pointing to the
metered verse spoken by Greek oracles, the language of Hebrew
prophets, and the god-dictated Vedas of India as evidence of
the link between poetry and god-speech, Jaynes asserted that
the god-voices spoke in verse. Beginning around 1000 b.c.e., he
believed, the discovery and spread of writing brought about a
breakdown in the functioning of the bicameral mind, although
the auditory hallucinations of modern-day schizophrenics fur-
nish evidence that contemporary consciousness can revert to its
earlier state. While the book was a fnalist for the National Book
Award, his theory has generated controversy.
Homers dactylic hexameter verse line is among those
surveyed by literary scholar Frederick Turner and psychophysi-
cist Ernst Pppel in their essay, Te neural lyre: Poetic meter,
the brain, and time (1989). Comparing the metrical verse line
lengths of various language cultures, Turner and Pppel found
that almost all of the lines took two to four seconds to recite,
with distribution peaking in the range of 2.5 to 3.5 seconds. Te
authors suggested that their fndings might refect a constant in
human neural processing: a human present moment or infor-
mation bufer averaging about three seconds in length, subject
to variation due to cultural factors. Literary critics have targeted
the essays biological reductionism and its underlying politics,
as the authors view free verse as an historical anomaly compat-
ible with bureaucratic or even totalitarian modes of cognition. To
date, their thesis has not been subjected to empirical scientifc
testing.
Initial data on the neurobiology of poetic language came from
studies of subjects who had sustained brain damage or under-
gone commissurotomy , surgical severing of the corpus cal-
losum . Tose fndings suggested that comprehension of many
poetic devices involved right hemisphere language pro-
cessing , even though the left hemisphere was known to control
language in most persons: Verbal intelligence tests of the isolated
left hemispheres of commissurotomy subjects fell in the normal
range, while subjects experienced aphasia after left (but rarely
right) hemisphere damage. However, over time, tests of right-
hemisphere-damaged (RHD) subjects revealed subtle linguistic
defcits in comprehending poetic devices such as metaphor or
connotation, while other studies showed that the isolated right
hemisphere recognized certain concrete nouns (i.e., images),
vowel sounds (i.e., assonance), and emotional prosody in spo-
ken or written language all important for understanding poetry
(Kane 2004 ) .
For example, Ellen Winner and Howard Gardner ( 1977 ) had
left-hemisphere-damaged (LHD), RHD, and control subjects
match a spoken expression such as He has a heavy heart to one
fully performed. All of these answers assume that the question
is not about the work itself, which cannot categorically be said
to be either verbal art (literature) or not verbal art (not litera-
ture); instead, works are more or less verbal art to the extent that
they carry the distinguishing characteristics of verbal art (or as
Goodman would say, works may carry symptoms of verbal art).
Being verbal art is thus a matter of degree. All of these answers
also imply that verbal art should be universal and that all users of
language should be able to have a literature because literature is
just a particular and always-possible way of using language .
Nigel Fabb
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bauman , Richard . 1984 . Verbal Art as Performance . Prospect Heights,
IL : Waveland .
Burling , Robbins . 1966 . Te metrics of childrens verse: A cross-linguis-
tic study . American Anthropologist 68 : 1418 41.
Edmonson , Murray S . 1971 . Lore: An Introduction to the Science of Folklore
and Literature . New York : Holt, Rinehart, and Winston . A survey of the
worlds literatures, with an interest in universals.
Fabb , Nigel . 1997 . Linguistics and Literature: Language in the Verbal Arts
of the World . Oxford : Blackwell . A survey of linguistic work on litera-
ture, drawing out potential universals.
. 1999 . Verse constituency and the locality of alliteration . Lingua
108 : 223 45.
. 2002 . Language and Literary Structure: Te Linguistic Analysis of
Form in Verse and Narrative . Cambridge : Cambridge University Press .
Fabb , Nigel , and Morris Halle . 2008 . Meter in Poetry: A New Teory .
Cambridge: Cambridge University Press. Te frst comprehensive the-
ory of the worlds meters.
Golston , Chris , and Tomas Riad . 2005 . Te phonology of Greek lyric
meter . Journal of Linguistics 41 : 77 115.
Goodman , Nelson . 1978 . Ways of Worldmaking . Indianapolis : Hackett .
Hanson , Kristin , and Paul Kiparsky . 1996 . A parametric theory of poetic
meter . Language 72 : 287 335.
Hayes , Bruce , and Margaret MacEachern . 1998 . Quatrain form in English
folk verse . Language 64 : 473 507.
Hymes , Dell . 1992 Use all there is to use. In On the Translation of
Native American Literatures , ed. B. Swann, 83124. Washington,
DC : Smithsonian Institution Press .
Jakobson , Roman . 1960 . Linguistics and poetics. In Style in Language ,
ed. T. Sebeok, 35077. Cambridge, MA : MIT Press .
Preminger , A ., and T. V. F. Brogan , eds. 1993. Te New Princeton
Encyclopedia of Poetry and Poetics . Princeton, NJ : Princeton University
Press . Te best source of information about the worlds poetic
traditions.
POETIC LANGUAGE, NEUROBIOLOGY OF
Not unlike the elephant approached by a delegation of blind
men, each of whom investigated a body part seemingly unre-
lated to the others, the neurobiology of poetic language has
been approached from such widely varying perspectives that
the results hardly seem to share tusks and a tail. One can peer
into ancient poetry in search of evidence that consciousness has
changed over time, or explore cross-cultural poetics for clues
to common neural processing mechanisms. One can map the
regions of the brain involved in the processing of poetic devices,
or pursue the question of how poets and nonpoets may difer in
their neural functioning.
Poetic Language, Neurobiology of
628
to normal subjects, and from behavioral tests to technology-as-
sisted observations. In the 1970s and 1980s, RHD subjects per-
formed poorly on connotative word meaning tests, while LHD
subjects experienced problems with denotation (Gardner and
Denes 1973 ; Brownell, Potter, and Michelow 1984 ; Drews 1987 ).
One might have assumed that the left hemisphere processed
denotation and the right, connotation, but over time, a more
complex picture emerged. Christine Chiarello and others estab-
lished, using visual-feld testing, that primary and subordinate
word meanings are initially activated in both hemispheres, but
that subordinate meanings are quickly suppressed in the left
hemisphere, resulting in a more ef cient processing time for the
dominant meaning not unlike Gioras graded salience model,
where the most salient meaning of an expression, metaphoric or
not, gets processed frst and faster than a less commonly occur-
ring meaning (Chiarello and Maxfeld 1995 ) .
Finally, the neurobiology of poets may play a signifcant role
in the neurobiology of poetic language. Poets are known to sufer
from afective disorders in particular, hypomania and bipolar
illness at rates far exceeding those of the general population
or other categories of writers (Andreasen 1987 ; Jamison 1989 ;
Ludwig 1994 ; Post 1996 ). Feeling negative emotion strongly,
being introspective, and spending time alone are traits associ-
ated with expressive writing as well as mental dysfunction, and
mentally ill persons may feel drawn to express their anguish in
writing; James Kaufman and John Baer ( 2002 ) propose these
and other behavioral explanations for the poetry/afective dis-
order connection. Taking a neurobiological approach, Felix
Post ( 1996 ) suggests that the intensive intellectual and emo-
tional efort involved in writing poetry may trigger overactivation
of neural networks and, thus, cause mental illness. Julie Kane
( 2004 ) suggests the opposite, that overactivation may precede
poetic output: Pointing to substantial evidence that handed-
ness and dominance for language can shift temporarily from the
left to right hemisphere during manic episodes, she proposes
that abnormal mood elevation may activate right-brain regions
involved in processing poetic language . Recently, too, Dawn
Blasko and Victoria Kazmerski ( 2006 ) have shown that poets and
nonpoets difer in the brain regions that they activate while read-
ing poems .
Tere is a vast amount of territory yet to be covered in explor-
ing the elephant of poetic language, complicated by the fact
that new research fndings often seem to challenge the old. But
as neuroimaging techniques become more precise and less inva-
sive, illuminating features that could only be guessed at before,
one thing becomes increasingly clear: Te neurobiology of poetic
language is not the same animal as the neurobiology of ordinary
language .
Julie Kane
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Andreasen , Nancy . 1987 . Creativity and mental illness: Prevalence
rates in writers and their frst-degree relatives . American Journal of
Psychiatry 144 : 1288 92.
Blasko , Dawn , and Victoria Kazmerski . 2006 . ERP correlates of individual
diferences in the comprehension of nonliteral language . Metaphor
and Symbol 21.4 : 267 84.
of four pictures, with the correct response being metaphoric. To
their surprise, RHD patients performed poorly, often selecting
the literal match for example, an illustration of someone carry-
ing a giant heart. Similar results were obtained from metaphoric
word-matching studies. Ten G. Bottini and colleagues ( 1994 )
used PET (positron emission tomography; see neuroimag-
ing ) to scan normal brains processing literal and metaphoric
sentences; blood fow (signaling brain activation) increased in
six regions of the RH when metaphoric but not literal sentences
were being processed. Te right hemispheres role in controlling
metaphor seemed obvious or was it?
As advances in technology have made fMRI ( functional mag-
netic resonance imaging ) studies of normal linguistic processing
possible, the results have raised as well as answered questions.
It is now known that conventional or frozen metaphors are
processed much like ordinary denotative language, primarily
in the left hemisphere, whereas novel metaphors as well as
ironies and the literal meanings of idioms light up additional
regions of the right hemisphere (Giora et al. 2000 ; Mashal, Faust,
and Hendler 2005 ; Sotillo et al. 2005 ; Eviatar and Just 2006 , Faust
and Mashal 2007 ). Tanks to fMRI, the precise brain regions
involved in novel metaphoric processing can be pinpointed: the
right homologue of wernickes area , right and left premotor
areas, right and left insula, and brocas area (Mashal, Faust,
and Hendler 2005 ). Of course, novel and not conventional met-
aphors are the stuf of poetic language, unless ones defnition
of poetry extends to greeting card verse, and so the role of the
right hemisphere remains signifcant. It was at frst assumed that
the right hemispheres increased involvement in novel meta-
phoric processing corresponded to visuospatial processing of
evoked imagery, whereas conventional metaphors were unlikely
to evoke pictures in the mind. However, Rachel Gioras graded
salience hypothesis (1997; Giora et al. 2000), which assumes that
the most common or salient meaning of an expression is pro-
cessed frst, regardless of whether it is literal or metaphoric, and
that right hemisphere language processing regions get recruited
only when secondary meanings must be accessed, provides an
alternate explanation .
Concrete nouns are the building blocks of poetic images, and
preliminary studies of commissurotomy patients led by Michael
Gazzaniga showed that the isolated right hemisphere was capa-
ble of recognizing simple nouns. Subsequent tests of normal sub-
jects, isolating either the right or left visual feld, suggested that
the left hemisphere excelled at processing abstract nouns and
low-imagery nouns, adjectives, and verbs, while the right per-
formed as well as the left in processing high-imagery nouns and
adjectives. Once again, neural-imaging studies have revealed a
more nuanced model than the simple association of left with
words and right with pictures (Kiehl et al. 1999 ). Marcel Just
and his colleagues ( 1996 ) and Jean Franois Demonet, Guillaume
Tierry, and Dominique Cardebat ( 2005 ) suggest that as cognitive
processing increases in complexity, right-hemispheric regions
get recruited to handle the additional demand. Tat hypothesis
does not necessarily confict with behavioral data showing the
right hemisphere to be poor at processing abstractions but good
at processing concrete nouns on its own.
Studies of connotation, another essential element of poetic
language, have followed a similar trajectory from brain-damaged
Poetic Language, Neurobiology of
629
Post , Felix . 1996 . Verbal creativity, depression and alcoholism: An inves-
tigation of one hundred American and British writers . British Journal
of Psychiatry 168 : 545 55.
Sotillo , Maria , Luis Carreti , Jos Hinojosa , Manuel Tapia , Francisco
Mercado , Sara Lpez-Mrtin , and Jacobo Albert . 2005 . Neural activ-
ity associated with metaphor comprehension: Spatial analysis .
Neuroscience Letters 373 : 5 9.
Turner , Frederick , and Ernst Pppel . 1989 . Te neural lyre: Poetic meter,
the brain, and time. In Expansive Poetry: Essays on the New Narrative
and the New Formalism , ed. Frederick Feinstein, 20954. Santa Cruz,
CA: Story Line.
Winner , Ellen , and Howard Gardner . 1977 . Te comprehension of met-
aphor in brain-damaged patients . Brain 100 : 717 29.
POETIC METAPHOR
Since Aristotles frst articulation of a comparative theory of
metaphor , metaphor studies in literary and ordinary language
have proceeded without interruption in philosophy, rhetoric ,
linguistics, and literary criticism. Two traditions have emerged
in metaphor theory: conceptual and linguistic traditions. Te
conceptual view emphasizes metaphors fundamental role in
everyday thought and language; the linguistic tradition lim-
its the range of metaphor to local pragmatic and aesthetic
functions (Ortony 1993 ). Te range of accounts within both
traditions is variegated and well beyond the scope of this entry.
Nonetheless, for the language sciences, it seems that the con-
ceptualist tradition has dominated in recent years. Te present
discussion assumes a conceptual view of metaphor as under-
stood through the frameworks of conceptual metaphor
theory (CMT) and conceptual blending theory (CBT),
where poetic metaphor is regarded as a special case of these
underlying conceptual operations. At present, poetic or literary
metaphor cannot be easily extracted from the central questions
of metaphor theory in general, namely: What is metaphor, and
what is metaphor for? Te present discussion merely touches on
the frst question, in favor of a more elaborate treatment of the
second question.
Te question of what metaphor is and how poetic metaphor
can help language sciences understand the everyday mind and
language is addressed in the frst section, where I compare and
contrast the two models of metaphor. Te question of what met-
aphor is for is addressed in the second section.
What Is Metaphor?
CMT purports to unearth the systematic correlations of experi-
ence and meaning. Meaning arises from everyday experience.
Abstract notions such as time, causation, states, change, and
purposes depend on a rich system of metaphors. Metaphor is
the name given to the process of conceptual mappings from
source to target domains (see source and target ). Te latest
incarnation of CMT (Lakof and Johnson 1999 ) builds on Joseph
Gradys ( 1997 ) theory of primary metaphor , in which the onto-
genetically basic process of domain correlation constitutes the
experiential basis of conceptual metaphors. A primary metaphor
is a correlation of subjective experience with a more abstract
concept. For instance, MORE IS UP is a primary metaphor, based
on the tight ontological correlation between the accumulation of
the same entities and vertical height .
Bottini , G. , R. Corcoran , R. Sterzi , E. Paulesu , P. Schenone , P. Scarpa , R.
Frackowiak , and C. Freith . 1994 . Te role of the right hemisphere in
the interpretation of fgurative aspects of language: A positron emis-
sion tomography study . Brain 117 : 1241 53.
Brownell , Hiram , Heather Potter , and Diane Michelow . 1984 . Sensitivity
to lexical denotation and connotation in brain-damaged patients: A
double dissociation . Brain and Language 22 : 253 65.
Chiarello , Christine , and Lisa Maxfeld . 1995 . Initial right hemispheric
activation of subordinate word meanings is not due to homotopic cal-
losal inhibition . Psychonomic Bulletin and Review 2 : 375 80.
Demonet , Jean-Franois , Guillaume Tierry , and Dominique Cardebat .
2005 . Renewal of the neurophysiology of language: Functional neu-
roimaging . Physiological Reviews 85 : 49 95.
Drews , Etta . 1987 . Qualitatively diferent organizational structures of
lexical knowledge in the left and right hemispheres . Neuropsychologia
25 : 419 27.
Eviatar , Zohar , and Marcel Just . 2006 . Brain correlates of discourse pro-
cessing: An fMRI investigation of irony and conventional metaphor
comprehension . Neuropsychologia 44.12 : 2348 59.
Faust , Miriam , and Nira Mashal . 2007 . The role of the right cere-
bral hemisphere in processing novel metaphoric expressions
taken from poetry: A divided visual field study. Neuropsychologia
45.4 : 860 70.
Gardner , Howard , and Gianfranco Denes . 1973 . Connotative judge-
ments by aphasic patients on a pictorial adaptation of the semantic
diferential . Cortex 9 : 183 96.
Giora , Rachel . 1997 . Understanding fgurative and literal language: Te
graded salience hypothesis . Cognitive Linguistics 7.1 : 183 206.
Giora , Rachel , ed. 2007. Is metaphor unique? Neural correlates of nonlit-
eral language. Brain and Language 100 .2 (Special Issue).
Giora , Rachel , Ofer Fein , Ann Kronrod , Idit Elnatar , Noa Shuval , and
Adi Zur . 2004 . Weapons of mass distraction: Optimal innovation and
pleasure ratings. Metaphor and Symbol 19 : 115 41.
Giora , Rachel , Eran Zaidel , Nachum Soroker , Gila Batori , and Asa
Kasher . 2000 . Diferential efects of right- and left-hemisphere dam-
age on understanding sarcasm and metaphor . Metaphor and Symbol
15 : 63 83.
Jamison , Kay Redfeld . 1989 . Mood disorders and patterns of creativity
in British writers and artists. Psychiatry 52 : 125 34.
Jaynes , Julian . [1978] 1990. Te Origins of Consciousness in the Breakdown
of the Bicameral Mind . 2d ed. Boston : Houghton .
Just , Marcel , Patricia Carpenter , Timothy Keller , William Eddy , and Keith
Tulborn . 1996 . Brain activation modulated by sentence comprehen-
sion . Science 274 .5284: 114 16.
Kane , Julie . 2004 . Poetry as right-hemispheric language . Journal of
Consciousness Studies 11.5/ 6 : 21 59.
Katz , Albert , ed. 2006. Metaphor and Symbol 21.4 . Special issue on neural
processing of nonliteral language .
Kaufman , James , and John Baer . 2002 . I bask in dreams of sui-
cide: Mental illness, poetry, and women . Review of General Psychology
6.3 : 271 86.
Kiehl , Kent , Peter Liddle , Andra Smith , Adrianna Mendrek , Bruce Forster ,
and Robert Hare . 1999 . Neural pathways involved in the processing of
concrete and abstract words . Human Brain Mapping 7 : 225 33.
Ludwig , Arnold . 1994 . Mental illness and creative activity in women
writers . American Journal of Psychiatry 151 : 1650 6.
Mashal , Nira , Miriam Faust , and Talma Hendler . 2005 . Te role of
the right hemisphere in processing nonsalient metaphorical mean-
ings: Application of principal components analysis to fMRI data .
Neuropsychologia 43.14 : 2084 100.
Mashal , Nira , Miriam Faust , Talma Hendler , and Mark Jung-Beeman .
Processing salient and less-salient meanings of idioms: An fMRI
investigation. Cortex. In press.
Poetic Language, Neurobiology of Poetic Metaphor
630
(8) Who made each mast, and sail and rope;
(9) What anvils rang, what hammers beat;
(10) In what a forge and what a heat
(11) Were shaped the anchors of thy hope.
(12) Fear not each sudden sound and shock,
(13) Tis of the wave, and not the rock;
(14) Tis but the fapping of a sail,
(15) And not a rent made by the gale.
(16) In spite of rock and tempests roar,
(17) In spite of false lights on the shore,
(18) Sail on, nor fear to breast the sea,
(19) Our hearts, our hopes are all with thee;
(20) Our hearts, our hopes, our prayers, our tears,
(21) Our faith triumphant oer our fears
(22) Are all with thee, are all with thee.
A CMT analysis begins by positing cross-domain mappings
between the source domain of ships and the target domain of
states or nation-states. Te conventional mappings between
source and target domains include the following correspon-
dences ofered in Grady, Oakley, and Coulson ( 1999 , 109):
All these metaphoric mappings derive from the basic pri-
mary metaphoric couplings of ACTION-AS-SELF-MOTION,
COURSES-OF-ACTION-AS-PATHS, SOCIAL-RELATIONSHIPS- AS-
DEGREES-OF-PHYSICAL-PROXIMITY, and CIRCUMSTANCES-AS-
WEATHER. Tese experiential correlations (and perhaps others)
interact in a way that motivates the framing of a nation and its his-
tory as a ship gliding through water .
As George Lakof and Mark Turner ( 1989 , 6772) argue, the
power of poetic metaphor, in particular, issues from the exten-
sion of these mappings for local expressive purposes. A conven-
tionalized metaphor never gives you all you need, and poetic
thought is marked by its ability to stretch or extend conventional
metaphors. Notice that with lines 67, DeMond extends the typi-
cal range of mappings to include ship building and the role
shipwright.
Poems also employ expressions in which the schemas and
domains underlying the metaphor can be elaborated in unusual
or novel ways. Lines 6 and 7, when understood against the con-
text of the whole speech, take on unusual signifcance. Te impli-
cation of line 6 is that the master shipwright is God, while the
workman is identifed with the Negro, echoing a consistent
theme of the speech the hard labor of the Negro race in build-
ing America.
CBT, while not a theory of metaphor, accounts for metaphor
as a species of conceptual blending that often involves the inte-
gration of concepts that do not normally go together. CBT takes
a decidedly usage-based perspective to metaphor and other
phenomena, in which systematic correlations arise from con-
ceptual blending itself, the process of constructing new scenes
and scenarios with specifc emergent properties from multiple
mental models . Te aim is to see how metaphors arise on the
fy as we think and talk. CMT has as its basic unit of cognitive
structure the conceptual domain. CBT has as its basic unit of
organization a mental space, or scenes and scenarios set up as
we think, talk, and otherwise interact. CBT models the dynamic
unfolding of a language users representations. In this respect,
CBT has developed analytic routines and modeling techniques
that capture constitutive principles and governing constraints of
blending. (See Fauconnier and Turner 2002 , 30952.)
What Is Metaphor For?
Tis question has no straightforward answer, but Samuel Levin
provides an initial approximation by suggesting that these onto-
logically bizarre notions are constructed for the purpose of
conceiving what a world would have to be like were it in fact to
comprise such states of afairs (1993, 121). Levin suggests that
we construct worlds in which the metaphor is literally true, but
only in order to tease out inferences that guide reasoning about
the real world. Consider the opening line of John Miltons poem,
On Time, in which the poet commands:
Fly envious Time, till thou run out thy race,
Tis conceit depends on the conventional metaphoric mapping
TIME IS A MOVER, creating a world in which time is literally an
intentional being running a race, the purpose of which is to focus
attention on the theological implications of speeding up the pace
at which the known world ends. In the poets world, the notion
of time as running a race can be considered preternatural, but
the theological implication of the end of days is the great truth
to be disclosed. In a similar vein, consider now the conventional
metaphor STATES ARE SHIPS .
Te text in question is the sermon, Te Negro Element in
American Life: An Oration. Delivered by Reverend A. L. DeMond
on January 1, 1900, this oration illustrates the degree to which
a conventional metaphor can be extended and elaborated. Te
reverend ends with a poem that makes elaborate use of the Ship-
of-State metaphor, a potentially disastrous rhetorical maneuver,
given the history of the forced importation of Africans. Te ser-
mon ends thus:
As the old ship of State sails out into the ocean of the 20th century,
the Negro is on board, and he can say:
(1) Sail on, O ship of State,
(2) Sail on, O Union, strong and great;
(3) Humanity, with all its fears,
(4) With all the hope of future years
(5) Is hanging breathless on thy fate.
(6) We know what master laid thy keel,
(7) What workman wrought thy ribs of steel;
Nation-State = Ship
Leader = Ships captain
National policies/actions = Ships course
National success/improvement = Forward motion of the
ship
National failures/problems = Sailing mishaps
Circumstances affecting the
nation
= Sea conditions
Poetic Metaphor
631
been conventionalized for just that purpose because the image
potential associated with building, operating, and navigating is
of rich social activities. In the blended space, however, the choice
to sail on is framed as an all-or-nothing proposition. If the ship of
state does not sail, it ceases to exist. In the sailing space, however,
the ship, once built, exists whether or not the crew sails; in the
sailing space, a captain and crew can choose when and when not
to sail, and the crew can still be referred to as sailors whether on
land or on sea. In the blend, a refusal to board and sail is tanta-
mount to renouncing ones citizenship. By exploiting elements
of the shipwright (a collective activity) and by attributing that
activity to a divine creator, DeMonds version of the ship of state
takes on the voice of a divine decree.
As suggested, DeMond takes considerable risk in quoting a
poem that makes extensive use of this metaphor, for members of
the congregation may generate a metaphoric mapping in which
the cross-domain counterpart of American Negro is not passen-
ger but cargo, destroying the political legitimacy of the image.
DeMond, however, assiduously avoids focusing any attention
on the circumstances that brought them to America. Instead, he
picks up the story at their arrival and tells of the Negro race as
those who built the nation .
Te present analysis presents CMT and CBT as complemen-
tary analytic frameworks, wherein the frst focuses solely on con-
ventionalized mappings, while the latter is much more interested
in how these mappings operate in local rhetorical contexts, and
thus can point scholars in the direction of a usage-based theory
of poetic metaphor .
Todd Oakley
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
DeMond , A. L. 1900. Te Negro element in American life: An oration.
Available online at: http://lcweb2.loc.gov/cgi-bin/query/r?ammem/
murray:@feld(DOCID+@lit(lcrbmrpt0e10div2).
Fauconnier , Gilles , and Mark Turner . 2002 . Te Way We Tink . New
York : Basic Books .
Grady , Joseph . 1997 . Foundations of meaning: Primary metaphors and
primary scenes. Ph.D. diss., University of California, Berkeley.
Grady , Joseph , Todd Oakley , and Seana Coulson . 1999 . Blending and
metaphor. In Metaphor in Cognitive Linguistics , ed. F. Gibbs and G.
Steen, 10124. Amsterdam : Benjamins .
Lakof , George , and Mark Johnson . 1999 . Philosophy in the Flesh . New
York : Basic Books .
Lakof , George , and Mark Turner . 1989 . More Tan Cool Reason: A Field
Guide to Poetic Metaphor . Chicago : University of Chicago Press .
Levin , Samuel . 1993 . Language, concepts, and worlds. In Metaphor
and Tought . 2d ed. Ed. A. Ortony, 11223. Cambridge : Cambridge
University Press .
Ortony , Andrew . 1993 . Metaphor, language, and thought. In Metaphor
and Tought . 2d ed. Ed. A. Ortony, 116. Cambridge : Cambridge
University Press .
POETICS
In ancient Greece, as Aristotle pointed out, there was no com-
mon name for all the diferent poetic genres ( Poetics , 47b),
including epic, tragic drama, dialogue, elegy, and poems writ-
ten in various meters. Poetry in the sense of making or creation
became the general name for literary expressions in diverse
A blending analysis helps account for ways in which the
NATION-AS-SHIP metaphor is not a simple and obvious map-
ping between two conceptual domains. While conceptual
domains name large depositories of knowledge about the physi-
cal and social world, mental spaces comprise on-line scenes and
scenarios; they are specifc and sensitive to pressures from local
context .
Levins ( 1993 ) account of poetic metaphor is more com-
pletely captured in CBT, a theoretical framework in which pre-
ternatural scenes are constructed to reveal how to reason and
draw inferences about something else. Conceptual blends are
often richly counterfactual, but rarely do they exist for their
own ends. In this case, the blended scenario extends and elabo-
rates the conventional metaphor for local rhetorical purposes. A
basic blending analysis of DeMonds introductory sentence and
the frst fve lines of the poem proper would include a discourse
ground specifying the participants, the situation, and setting, a
mental space for Seafaring, a mental space for Nation, and the
initial blended space for Nation-as-Ship, each of which is set up
in the very frst line of the poem.
Let us assume the analysis from the perspective of a worshiper
sitting in the Dexter Avenue Baptist Church in 1900. Under these
conditions, the ground includes the identities of the churchgo-
ers, the speaker, and the setting. Let us further assume that the
discourse participants are African Americans and that the poetic
persona represents them. Initially, the Seafaring and Nation
spaces project conceptual structure into the blend under the
infuence of the cross-space mappings as specifed here. In the
blend, America is a ship, Negro citizens are among its passen-
gers, the ocean is time, and the twentieth century is an unspeci-
fed landmark on open water. Te blend allows the audience
to imagine temporally, causally, and spatially difuse political
events as attaining, for the moment, the look and feel of primary
experience.
Once composed, the blend and the network of mental spaces
permit the addition of new information and relations. A notewor-
thy contribution of the blending framework here is that it ofers
precise ways of accounting for the elements of the Nation-as-
Ship image that have no specifc counterparts in the target space
of nations and politics. Once the network is up and running,
readers can combine concepts fuidly. For instance, line 2 com-
mands, Sail on, O Union, strong and great, wherein the poet
fuses elements from diferent mental spaces into tight syntactic
units. Tus, in the blend it is perfectly natural and logical for a
union to sail. What is more, it is perfectly natural for the ship to
plot a straight course. Once the image is created, many other ele-
ments of ships become mentally accessible. For instance, ships
must be made of particular materials in order to be seaworthy.
Te phrase, ribs of steel, in line 6 satisfes local formal and con-
ceptual imperatives in 1) providing completion for the couplet
with line 5, and 2) suggesting that the nation is made of sturdy
material and (opportunistically) made from the very material
that the Negro worker has been responsible for manufacturing.
Importantly, the mapping between Shipwright and Creator is
responsible for all aspects of the nation.
Te goal of using this conventional metaphor is to construct
a view of social reality for the Negro race, focusing on communal
activities and on achieving collective goals. Te ship of state has
Poetic Metaphor Poetics
632
Literary Mind and the Carving of Dragons is deservedly famous
as the most systematic study of the literary art in the Chinese
critical tradition. Tis substantial work of Chinese poetics relates
literature to the cosmic tao and the exemplary classics of ancient
sages, thereby elevating literature to a position of high social and
moral values. Its focus, however, is on the art of literature. Te
Literary Mind frst formulates some basic principles of the idea of
wen or literature, gives a survey of all the literary genres in clas-
sical Chinese literature, commenting on their origin and devel-
opment, and then presents a highly developed theory of literary
creation, making contributions to the important issues of the
relationship between poetry and reality, the style and character-
istics of a literary work, the efect of imagery and poetic imagina-
tion, and the regulations of metric composition. Since the eighth
century in Tang China, and particularly the eleventh century in
the Song Dynasty, there have been numerous works in a critical
genre known as remarks on poetry ( shihua ), which often contain
valuable insights into the nature of poetry, the techniques of the
literary art, and the principles of aesthetic appreciation. Like the
aforementioned Indian example, the Chinese critical tradition
also ofers an alternative form of poetics outside the Aristotelian
and European tradition .
In a broad sense, then, poetics can be understood as a crit-
ical, theoretical, and more or less systematic treatment of poetry
or literature in general. In such an expanded usage, what the
term signals is a theoretical discourse on a subject in arts or lit-
erature, covering a considerable range of oeuvre, and ofering
some philosophical insights into the nature of the subject under
discussion. Poetics, therefore, becomes a general term for a sus-
tained argument or a long essay in literary and art criticism .
Zhang Longxi
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aristotle . 1987 . Poetics with the Tractatus Coislinianus, Reconstruction
of Poetics II, and the Fragments of the On Poets . Trans. Richard Janko .
Indianapolis : Hackett .
Averros . 1977 . Tree Short Commentaries on Aristotles Topics,
Rhetoric, and Poetics . Ed. and trans. Charles E. Butterworth .
Albany : State University of New York Press .
De , S. K . 1988 . History of Sanskrit Poetics . Calcutta : Firma KLM Pvt. Ltd .
Liu , James J. Y . 1975 . Chinese Teories of Literature . Chicago : University
of Chicago Press .
. 1988 . Language-Paradox-Poetics: A Chinese Perspective . Ed.
Richard John Lynn . Princeton, NJ : Princeton University Press .
Miner , Earl . 1990 . Comparative Poetics: An Intercultural Essay on Teories
of Literature . Princeton, NJ : Princeton University Press .
Pathak , R. S . 1998 . Comparative Poetics . New Delhi : Creative Books .
Rajendran , C . 2001 . Studies in Comparative Poetics . Delhi : New Bharatiya
Book Co .
POINT OF VIEW
In narrative studies (see narratology ), this term, also per-
spective or focalization , refers to textual strategies that provide
the reader with the illusion of seeing things through the eyes
of a character. Tese strategies are mostly linguistic in nature,
ranging from deictic positioning in the characters mental here
and now (see deixis ) to lexical choices linking up with the
forms, and Poetics , the term used by Aristotle for his treatise on
tragedy and epic, thus represented the kind of critical and ana-
lytical treatment of poetry that would be called in later times lit-
erary criticism or literary theory.
Aristotles Poetics ofers an important model in Western liter-
ary criticism, but it was not widely known in Europe in antiquity
or in medieval times, and it did not become a classic until the
latter half of the sixteenth century. During the time that it was lost
in medieval Europe, however, the Poetics , along with some other
works by Aristotle, was being studied by Arabic scholars, notably
Ibn Rushd , known in the West as Averros. But once it was redis-
covered and commented on by such infuential Renaissance crit-
ics as Lodovico Castelvetro (150571) and Francesco Robertello
(151667), the Poetics quickly became one of the most infuential
works in Western literary criticism. Epic and tragedy discussed
therein became the two major classical genres before the rise
of the modern novel and, after Dante, poets of every European
nation tried to create an epic in the vernacular to mark the matu-
rity of a modern language and the establishment of a national
literary tradition. Aristotles philosophical treatment of plot, lan-
guage, and rhetoric of the tragic drama provides a model of criti-
cal analysis, and many basic concepts used in the Poetics , such
as imitation, recognition, the reversal of fortune, tragic hubris,
and the catharsis of pity and fear, have all had a tremendous
infuence on later criticism. In our own time, Aristotles Poetics
remains a major classic and continues to be discussed and com-
mented on by important critics and theoreticians from various
perspectives.
As the aforementioned Arabic commentaries suggest, the
systematic study of the literary art is by no means confned to
the European tradition. Tere are, for example, well-established
traditions of sophisticated literary criticism or poetics in South
and East Asia. Te earliest treatise on dance and dramatic art in
ancient India, Bharatamunis Ntyastra (ca. second cent. b.c. ),
ofers a comprehensive discussion of Sanskrit drama in terms of
taste and emotions ( rasa ) and of language and bodily gestures
that give expression to various emotions. In the seventh century,
Sanskrit poetics was fully established by such important theorists
as Bhmaha and Dandin. In the ninth century, nandavardhana
made signifcant contributions to its further development with
discussions of the theoretical notions of rasa and dhvani , while
Abhinavagupta and Kuntaka in the tenth century explored new
areas by debating on the issue of indirect and suggestive expres-
sions ( vakrokti ) in poetic language. Indeed, as an Indian scholar
remarks, A study of Sanskrit poetics from Bharata (5th century
b.c. ) to Panditarja Jaganntha (17th century a.d. ) will bear wit-
ness to the existence of a highly developed poetics in ancient
India, with a rigorous scientifc method for description and anal-
ysis of literature (Pathak 1998 , 345) .
In China, the Great Preface to the Mao edition of the Book
of Poetry (second century b.c. ) articulated the Confucian ideas
about poetry and its functions, and laid the foundation of a poet-
ics that both acknowledges the release of emotions as the origin
of poetry and the ef cacy of moral teaching as its ultimate justif-
cation. Lu Ji (261303), with his Rhyme-Prose on Literature added
to the critical tradition a more focused attention on the impor-
tance of emotions ( qing ), and he argued for the necessity to learn
both from nature and from the ancients. Liu Xies (465?520?)
Poetics Point of View
633
Friedman in looking at types of narrative. Boris Uspensky ([1973]
1983 ), too, extends the meaning of point of view under his term
perspective to include a) vision (spatio-temporal perspective); b)
language (phraseological perspective); c) knowledge and feelings
(psychological perspective); and d) ideology . Although these
four types of perspective are all determinable from the language
of the text (the spatio-temporal perspective through deictics; the
ideological through tell-tale phrases like tovarish [comrade]
for fellow man; the psychological through the syntax and
lexis of emotion ) , it is the phraseological level of perspective that
is most linguistic in its deployment of register and style to
signal narrators or characters perspectives, for instance, in the
citing of dialect words, hints at pronunciation typical of certain
social groups, or the contrast between high and low register in
heteroglossic texts (Bakhtin 1981 ; see dialogism and het-
eroglossia ). For instance, Uspensky cites Tolstoys sentence
Anna Pavlova had been coughing for the last few days: she had
an attack of la grippe , as she said (1983, 33) as an example of
phraseological point of view, where la grippe registers Anna
Pavlovas class and social snobbery. Psychological perspec-
tive can be exemplifed by a sentence from Toni Morrison : He
examined the bushes, the branches, the ground for a berry, a nut,
anything (1977, 255; emphasis added). Te sentence traces the
order of Milkmans order of perception and the urgency of his
quest for food. Uspensky devotes a whole chapter to the inter-
relation of the four types of perspective in texts .
Grard Genettes reconceptualization of point of view as focal-
ization (zero; external; internal) abides by the visual metaphor,
with focalization opposed to voice (who sees? vs. who speaks?
[1972] 1980, 186). Genettes typology of focalization is one of
limited perspective either no limitation of point of view (zero
focalization) or limitation to a view on characters from outside
(external focalization) or a subjective view from inside (internal
focalization). Te narrator and narrative voice are excluded from
the discussion, in contrast to Friedmans or Stanzels analyses.
More recent models of focalization are discussed by Manfred
Jahn ( 2005 ), who has himself proposed the distinction among
strict, ambient, weak, and zero focalization based on an optical
analogy .
Linguistic Signals of Point of View
Te textual inscription of point of view depends on the insertion
of signals of subjectivity and individual knowledge, opinion, or
worldview in the text such that they can be aligned with a char-
acter. Te same signals can also be employed to relate the sub-
jectivity or individual stance of the speaker/narrator of a text/
utterance, and this alignment is usually discussed under the
heading of voice and not point of view. Voice and point of view
can get into confict or overlap as in free indirect discourse , a tech-
nique for rendering speech or thought in which the language of
the reported speaker/thinker (his/her point of view) is to some
extent preserved in the report: She had never, ever told fbs, not
for worlds . (Here, the syntax and vocabulary of the reported
speaker are integrated into the report.)
Free indirect discourse (thought representation) is one of the
most common signals of point of view in literary texts since it
introduces a characters perspective (feelings, intentions, world-
view) to the reader. Moreover, the narrative can be studded with
characters worldview and ways of thinking and perceiving the
world. Point of view, from a linguistic perspective, is therefore an
important aspect of linguistic pragmatics .
Origins and History
Te strategies of point of view narration are of fairly recent date.
Tey came into existence as part of the shift toward increasingly
subjective literary narratives near the end of the nineteenth cen-
tury, and document authors attempts to portray characters
individuality not merely in the rendering of idiosyncratic dia-
logue (for instance, in dialect ) but also in the extensive depic-
tion of characters minds or consciousness. Already in the 1790s,
the Gothic novel fgured the female protagonists anxious medi-
tations, and Ann Radclife and Charles Maturin also portrayed
the impact that the contemplation of sublime scenery had on
their heroines. Gothic novels are, therefore, important anticipa-
tions of the point of view technique which, in English literature,
came into its own in the work of George Eliot , Elizabeth Gaskell ,
Tomas Hardy , Joseph Conrad , and Henry James and in the
stream of consciousness novels of literary modernism (James
Joyce, Virginia Woolf, Katherine Mansfeld, D. H. Lawrence,
Aldous Huxley, E. M. Forster). Since then, the point of view tech-
nique has been standard in fctional narrative, especially so in
the short story, even though postmodernist texts of the radical
experimental sort do not employ it as often.
Te term point of view (interchangeably with center of vision )
was frst used by Percy Lubbock in Te Craft of Fiction (1921),
although Henry James in his prefaces had already analyzed the
phenomenon under the heading of center of consciousness
and the image of the house of fction having many windows
(James [18801/1908] 1975 , 7). Point of view in James refers to
the presentation of the story from the perspective, hence point of
view, of a character, for instance, Strether in Te Ambassadors : It
afected Strether: horrors were so little superfcially at least in
this robust and reasoning image. But he was none the less there
to be veracious (James [1903] 1994 , 99). However, even in James,
the point of view technique, in the meaning of limited perspec-
tive (seeing the world through the naive, obsessed, or puzzled
perspective of a character), is extended from the new narrative
form of the (third person) stream of consciousness novel (follow-
ing the protagonists associations in the depiction of their con-
sciousness) to experiments with unreliable or otherwise limited
frst person narrators, as in Jamess Te Real Ting or Daisy
Miller. In these texts, the narrator is very naive, and has a clearly
reduced intellectual capacity. For Lubbock ( 1921 ), Jean Pouillon
( 1946 ), and Norman Friedman ( 1955 ), the term, by contrast,
comprises not one technique of focusing the narrative through a
central characters mind but a variety of three (Pouillon) to eight
(Friedman) alternative points of view that authors can choose.
Not only was point of view a vague term because it included
so many diferent aspects of narrative; it, moreover, was very
limiting since it focused on the visual metaphor. As a conse-
quence, the term point of view , though still used as a general
label, became displaced in narratology by more inclusive or more
specifc terms: narrative situation , perspective , and focalization .
Franz Karl Stanzels ([1979] 1984 ) three narrative situations
([1] authorial roughly: omniscient; [2] frst person; and [3]
fgural the presentation through a characters mind) follows
Point of View
634
models of speech that are employed to create an illusion of
authenticity. Moreover, the attribution of expressivity markers
to the primary frame speaker (narrator) or reported speaker
(character) is frequently problematic. Te mere presence of
expressivity markers does not convey a clear point of view;
point of view needs to be constructed interpretatively by the
listener or reader in the overall context of the utterance or text.
Tus, though point of view can be fruitfully analyzed by linguis-
tic means, it cannot be exhaustively described within a purely
formal framework. Point of view, therefore, is a pragmatic phe-
nomenon located on the threshold between narrative pragmat-
ics and literary narratology (see also literary character
and character types ) .
Monika Fludernik
WORKS CITED AND SUGGESTED FURTHER READING
Bakhtin , Michael M . 1981 . Te Dialogic Imagination: Four Essays . Ed.
Michael Holquist . Austin : University of Texas Press.
Bhler , Karl . 1934 . Sprachtheorie. Die Darstellungsfunktionen der Sprache .
Jena : Gustav Fischer .
Dickens , Charles . [18523] 1962. Bleak House . London : Oxford University
Press .
. [18645] 1952. Our Mutual Friend . London : Oxford University
Press .
Fillmore , Charles . 1983 . How to know whether youre coming or going.
In Essays on Deixis , ed. Gisa Rauh , 21927. Tbingen : Narr .
. 1997 . Lectures on Deixis . Stanford, CA : Center for the Study of
Language and Information .
Fludernik , Monika . 1993 . Te Fictions of Language and the Languages
of Fiction: Te Linguistic Representation of Speech and Consciousness .
London : Routledge .
Friedman , Norman . 1955 . Point of view in fction: Te Development of a
Critical Concept . PMLA 70 : 1160 84.
Genette , Grard . [1972] 1980. Narrative Discourse: An Essay in Method .
Trans. Jane E. Lewin . Ithaca, NY : Cornell University Press .
Halliday , M. A. K . 1971 . Linguistic function and literary style: An inquiry
into William Goldings Te Inheritors . In Literary Style: A Symposium ,
ed. Seymour Chatman , 33065. London : Oxford University Press .
Harweg , Roland . 1968 . Pronomina und Textkonstitution . Munich : Wilhelm
Fink .
Jahn , Manfred . 1999 . More aspects of focalization: Refnements and
applications. In Recent Trends in Narratological Research , ed. John
Pier , 21, 85110. Tours : Groupes de Recherches Anglo-Amricaines de
Tours , University of Tours .
. 2005 . Focalization. In Routledge Encyclopedia of Narrative
Teory , ed. David Herman , Manfred Jahn , and Marie-Laure Ryan ,
1737. London : Routledge .
James , Henry . [18801/1908] 1975. Te Portrait of a Lady . Ed. Robert D.
Bamberg . New York : W. W. Norton .
. [1903] 1994. Te Ambassadors . Ed. S. P. Rosenbaum . New
York : Norton .
. [1934] 1953. Te Art of the Novel. Intro. R. P. Blackmur . New
York : Scribne r.
Lubbock , Percy . 1921 . Te Craft of Fiction . London : Jonathan Cape .
Morrison , Toni . 1977 . Song of Solomon . New York : Signet .
Pouillon , Jean . 1946 . Temps et roman . Paris : Gallimard .
Shen , Dan . 2005 . Mind-style. In Routledge Encyclopedia of Narrative
Teory , ed. David Herman , Manfred Jahn , and Marie-Laure Ryan ,
31112. London : Routledge .
Stanzel , Franz Karl . [1979] 1984. A Teory of Narrative .
Cambridge : Cambridge University Press .
stylistic and lexical markers relating to the characters social
position, age, gender, and so on. For instance, when in Charles
Dickenss Our Mutual Friend Mrs. Veneering remarks that
these social mysteries make one afraid of leaving Baby ([18645]
1952, 414; emphasis added), the word Baby relates to the moth-
erchild relationship of the reported speaker and represents her
point of view. At the same time, the phrase these mysteries and
the pronoun one underline Mrs. Veneerings upper-class status.
Addressee-oriented expressions like forms of address ( Maam ,
Sir , Your Excellency , etc.) also invoke social position by linguistic
means (cf. Fillmore 1983 , 1997 ) .
Most basically, deictics serve the function of positioning
speakers and, hence, creating point of view. For example, in
Bleak House Mr. Bucket (still grave) inquires if to-morrow morn-
ing, now, would suit ([18523] 1962, 720; emphasis added),
in which the futurity of to-morrow relates to Mr. Buckets
moment of utterance . Among linguists, Charles Fillmores work
on deixis (1983, 1997) needs to be credited with incisive insights
into the generation of point of view by means of deixis. From a
linguistic perspective, these signals are expressivity markers,
implying a speaking or thinking consciousness, a deictic center
(Bhlers [ 1934 ] origo ) from which the world is being viewed. In
the widest possible sense, such expressivity markers are indica-
tive of ideation and emotion, the latter capable of being textually
suggested by syntactic means, such as intensifying repetitions
besides merely lexical intensifers and emphatic vocabulary.
Evaluative point of view can be illustrated in sentences like Do
talk to the poor dear . Incomplete sentences (indicating hesi-
tation or derangement), sentence modifers ( in any case , sure
enough ), clause-initial adjuncts ( oh , well ), interjections ( good
grief ), negative inversion ( Never will he forget ) or left and right
dislocation are among the most common strategies used (cp.
Fludernik 1993 , 22779). In oral discourse, moreover, expressiv-
ity shows up in intonation and the echoing of idiosyncratic
pronunciation (imitated in writing: she sho was happy ).
In medieval literature, point of view is often signaled by inter-
jections like alas or by means of repetition. Such signals of point
of view occur intermittently in medieval literature and early
modern English prose but do not constitute a continuous repre-
sentation of a characters perspective as in the Gothic novel and
the later stream of consciousness novel.
Like the study of narrative discourse markers, the focus on
expressivity signals can help to emphasize the specifcally narra-
tive uses of point of view for the linguist. Point of view markers not
only establish free indirect discourse; they are, moreover, crucial
to text beginnings, where they help distinguish between narra-
tives with a prominent speaker (= narrator) function and others
in which the reader is eased into the story by means of a protago-
nists perspective. (Roland Harweg [ 1968 ] has contrasted these
as emic and etic text beginnings, respectively.) Peculiarities of
thought and worldview are also constitutive of M. A. K. Hallidays
mind-style (1971) as the distinctive linguistic representation of
individual self (Shen 2005 , 312). Ultimately, an analysis of point
of view as expressivity links up with the linguistic enquiry into
individual style.
It should be noted that all of these signals of expressivity are
clichs and cannot directly claim mimetic relevance (Fludernik
1993 , 43464). On the contrary, they depend on typical recurrent
Point of View
635
Politeness as Conversational Maxims
A diferent approach understands politeness as a set of social
conventions coordinate with Paul Grices ( 1975 ) coopera-
tive principle for maximally ef cient information transmis-
sion (Make your contribution such as required by the purposes
of the conversation at the moment), with its four maxims of
quality , quantity , relevance , and manner (see conversational
implicature ) . Robin Lakof ( 1973 ) argued that three rules
of rapport underlie choices of linguistic expression, rules that
can account for how speakers deviate from directly expressing
meanings . Choice among the three pragmatic rules gives rise
to three distinct communicative styles: Rule 1, Dont impose,
produces a distant style; Rule 2, Give options, gives rise to a
deferent style; and Rule 3, Be friendly, results in a style of cama-
raderie. Geofrey Leechs ( 1983 ) proposal is in the same vein.
Complementary to Grices cooperative principle, Leech postu-
lated a politeness principle , Minimize the expression of impolite
beliefs, with six maxims of tact , generosity , approbation , mod-
esty , agreement , and sympathy . As with Grices maxims, devia-
tions from what is expected give rise to inferences. Cross-cultural
diferences derive from the diferent importance attached to par-
ticular maxims.
Te conversational maxim view shares with the social norm
view the emphasis on codifed social rules for minimizing friction
between interactors and the idea that deviations from expected
levels or forms of politeness carry a message .
Politeness as Face Management
A more sociological perspective places face work at the core of
politeness. Erving Gofman ( 1967 ) considered politeness as
an aspect of interpersonal rituals, central to public order. He
defned face as an individuals publicly manifest self-esteem and
proposed that social members have two kinds of face require-
ments: positive face, or the want of approval from others, and
negative face, or the want not to ofend others. Attention to these
face requirements is a matter of orientation to Gofmans dip-
lomatic fction of the virtual ofense, or worst possible reading
(1971, 138 f), the working assumption that face is always poten-
tially at risk, so that any interactional act with a social-relational
dimension is inherently face threatening and needs to be modi-
fed by appropriate forms of politeness. Deference (attention
owed to the others face) can be distinguished from demeanor
(attention owed to oneself).
Building on Gricean and Gofmanian approaches, Penelope
Brown and Stephen C. Levinson ([1978] 1987 ) introduced a
comparative perspective by drawing attention to the detailed
parallels in the construction of polite utterances across widely
difering languages and cultures, arguing that universal prin-
ciples underlie the construction of polite utterances. Te par-
allels they noted are of two sorts: how the polite expression of
utterances is modifed in relation to social characteristics of
the interloculors and the situation, and how polite utterances
are linguistically formulated. At least three social factors are
involved in deciding how to be polite: 1) One tends to be more
polite to social superiors; 2) one tends to be more polite to people
one doesnt know. In the frst case, politeness tends to be asym-
metrical (the superior is less polite to an inferior); in the second,
politeness tends to be symmetrically exchanged. In addition,
Uspensky , Boris . [1973] 1983. A Poetics of Composition: Te Structure of the
Artistic Text and Typology of a Compositional Form . Trans. Valentina
Zavarin and Susan Wittig. Berkeley : University of California Press .
POLITENESS
Politeness is essentially a matter of taking into account the feel-
ings of others as to how they should be interactionally treated,
including behaving in a way that demonstrates appropriate
concern for interactors social status and their social relation-
ship. In this broad sense of speech oriented to an interactors
social persona or face , politeness is ubiquitous in language use.
Since taking account of peoples feelings generally involves
saying things in a less straightforward or more elaborate man-
ner than when one is not considering such feelings, ways of
being polite provide a major source of indirectness, reasons for
not saying exactly what one means, in how people frame their
utterances.
Tere are many folk notions for these kinds of attention to
feelings, captured in terms like courtesy , tact , deference , sensibil-
ity , poise , rapport , and urbanity , as well as terms for the contrast-
ing behaviors rudeness , gaucheness , social gafes and their
consequences, embarrassment or humiliation. Such terms attest
both to the pervasiveness of notions of politeness and to their
cultural framing.
Peoples face is invested in their social status and in their
relationships with one another, and so indexing this relationship
appropriately is necessary for maintaining face expectations. In
addition, one often has interactional goals that potentially con-
travene face, and the expression of such communicative inten-
tions (e.g., requests, ofers, disagreements, complaints) tends to
be mitigated by attention to face.
Politeness is crucial to the construction and maintenance
of social relationships; indeed, it is probably a precondition for
human cooperation in general. Politeness phenomena have,
therefore, attracted interest in a wide range of social sciences,
particularly linguistics, anthropology, psychology, sociology,
and communication . Work in these disparate felds can be char-
acterized in terms of three main classes of theoretical approach.
Politeness as Social Rules
To the layperson, politeness is a concept designating proper
social conduct, rules for speech and behavior stemming generally
from high-status individuals or groups (cf. standardization ).
Tese notions range from polite formulae like please and thank
you , codifed forms of greetings and farewells, honorifc address
forms, and so on, to more elaborate routines, for example, for
table manners or the protocol for formal events. Politeness in
this view is conventionally attached to certain linguistic forms
and formulaic expressions, which may be very diferent in difer-
ent languages and cultures.
Some analytical approaches to politeness are formulated in
terms of the same sorts of culture-specifc rules for doing what is
socially acceptable, for example, the work by Sachiko Ide ( 1989 )
and others on Japanese politeness as social indexing or discern-
ment. In these approaches, politeness inheres in particular lin-
guistic forms when used appropriately as markers of pregiven
social categories .
Politeness
636
Goody , Esther . 1995 . Social Intelligence and Interaction .
Cambridge : Cambridge University Press .
Grice, H. Paul . 1975 . Logic and conversation. In Syntax and
Semantics. Vol. 3: Speech Acts. Ed . P. Cole and J. Morgan , 4158. New
York : Academic Press .
Hickey , Leo , and Miranda Stewart , eds. 2005. Politeness in Europe .
Clevedon, UK : Multilingual Matters .
Ide , Sachiko . 1989 . Formal forms and discernment: Two neglected
aspects of linguistic politeness . Multilingua 8.2/3 : 223 38.
Lakof , Robin . 1973 . Te logic of politeness or minding your ps and qs.
In Papers from the Ninth Regional Meeting of the Chicago Linguistic
Society , 292305.
Leech , Geofrey . 1983 . Principles of Pragmatics . London : Longman .
Placencia , Mara E. , ed. 2006. Research on Politeness in the Spanish-
Speaking World . Mahwah, NJ : Erlbaum .
Watts , Richard J . 2003 . Politeness . Cambridge : Cambridge University
Press .
Watts , Richard J. , Sachiko Ide , and Konrad Ehlich , eds. 1992. Politeness in
Language . Berlin : Mouton de Gruyter .
POLITICS OF LANGUAGE
Politics of language is not a domain or subdiscipline. It is an idea
that puts the study of language in a perspective: the idea that
language is a politically invested object and that people, con-
sequently, act politically in, through, and on language. In that
sense, the term covers an enormous range of phenomena and
cuts across numerous disciplinarily organized practices. Te
issue is one of function, and the politics of language suggests
that political meanings and efects are among the functions of
language. In fact, some would emphasize that there are no non-
political meanings.
Such political functions are metapragmatic: Tey operate
through meta-discourses on language, on things people say
about language in language. Tus, the politics of language is a
language-ideological phenomenon. Clear instances of it are
widespread utterances such as English is the language of busi-
ness or Xhosa is a language for community interaction. In
both instances, a particular language is defned as a language
that operates with a specifc load, a specifc set of social, cul-
tural, economic political attributes, all of them implicitly
articulated: Whenever I use English, my language use will be
framed as business, and I will speak like a businessman. Te
politics of language has to do with the way in which we associ-
ate particular varieties of language (forms) with particular nor-
mative complexes, genres and topical domains, and identities
(functions). Te relationship between forms and functions, thus
defned and seen as relatively stable (stable enough to generate
shared meanings), is usually defned as ideology , and authors
explicitly addressing the politics of language often focus on ide-
ology, hegemony, and ideological naturalization.
In what follows, I frst give a brief overview of some key notions
and authors, then engage in a brief survey of some recent work
and focus on language ideologies as a frame for understand-
ing such political functions. I conclude with an appraisal of this
work.
Key Notions and Authors
Language has been defned as politically invested since Aristotle
and the Sophists; it is therefore futile to attempt a historical survey.
3) in any culture there are norms and values afecting the degree
of imposition or unwelcomeness of an utterance, and one tends
to be more polite for more serious impositions. Te linguistic
structures for conveying particular kinds of politeness are also
underlyingly similar across languages, with the politeness of
solidarity (positive politeness) characterized by expressions of
interest in the addressee, exaggerated expressions of approval,
use of in-group identity markers and address forms, seeking
of agreement, and avoidance of disagreement, whereas avoid-
ance-based politeness (negative politeness) is characterized
by self-efacement, formality, restraint, deference, hedges, and
impersonalizing mechanisms like nominalization or passive
constructions.
To explain these kinds of detailed parallels across languages
and cultures in the minutiae of linguistic expression in socially
analogous contexts, Brown and Levinson proposed an abstract
model of politeness as strategic attention to face, deriving strat-
egies for constructing polite utterances in diferent contexts on
the basis of assessments of three social factors: the relative power
(P) of speaker and addressee, their social distance (D), and the
intrinsic ranking (R) of the face-threateningness of an imposition.
In contrast with rule-based approaches, Brown and Levinson
argued that politeness inheres not in words or sentences per se;
politeness is an implicature that may be conveyed by utterances
spoken in context, by virtue of successful communication of a
polite attitude or intention .
Politeness continues to be a major focus for research in many
disciplines concerned with social interaction, and the topic
now has its own professional journal, the Journal of Politeness
Research . Over the past 30 years, empirical descriptions of par-
ticular politeness phenomena from many diferent parts of the
world have accumulated, with the research emphasis largely
on cross-cultural diferences. Tere has been much theoretical
controversy over whether, indeed, there are any universal prin-
ciples of politeness and if so, what form they take. Te recent
trend seems to be toward emphasizing emic rather than etic
approaches (cf. Watts 2003 ; Eelen 2005 ). But the importance of
politeness goes far beyond the ps and qs of appropriate behav-
ior and speech in a particular cultural setting. Its wider signif-
cance is in the interactional, communicative, day-to-day basis of
social life and the conduct of social relationships. Recent devel-
opments in the theory of social interaction that take account
of our common human nature (e.g., Goody 1995 ; Enfeld and
Levinson 2005 ; see also universal pragmatics ) ofer hope
for theoretical progress in this feld .
Penelope Brown
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brown , Penelope , and Stephen C. Levinson . [1978] 1987. Politeness: Some
Universals in Language Use . Cambridge : Cambridge University Press .
Eelen , Gino . 2005 . A Critique of Politeness Teories . Manchester : St.
Jerome .
Enfeld , Nick , and Stephen C. Levinson , eds. 2005. Roots of Sociality .
Cambridge : Cambridge University Press .
Gofman , Erving . 1967 . Te nature of deference and demeanor. In
Interaction Ritual , ed. Erving Gofman, 4795. New York : Anchor
Books .
. 1971 . Relations in Public . New York : Harper Colophon Books .
Politeness Politics of Language
637
Barthes ( 1957 ), who emphasizes the discursive routines and the
silences that are generated by the consumer-capitalist society. It
also underlies Foucaults ( 1984 ) notion of order of discourse,
and it is refected in Bourdieus ( 1991 ) notion of legitimate lan-
guage. In each case, macrosocial order manifests itself in dis-
course patterns, structures, both positively and negatively. Te
fact that some things can only be said in some ways is an efect of
the social and political order; the fact that some things cannot be
said at all is an efect of the same thing (Blommaert 2005 ).
Te fact is, however, that people rarely experience this shap-
ing of discourses as an efect of social and political forces. Mostly,
we perceive these discourse routines and absences as normal,
as just the way things are. It is at this point that we see scholars
refer to the Marxian notion of ideology an agentive notion in
which ideational complexes such as discourses have real mate-
rial efects as well as to the Gramscian notion of hegemony.
Hegemony is ideological dominance, that is, dominance that
is not perceived as dominance but as a neutral, normal state of
afairs. Social and political forces operate in language through
hegemony, that is, through naturalized, neutralized, and nor-
malized perceptions and forms of behavior.
Tese authors all provide frequently used key notions and
insights, all of which revolve around the same central node: that
language is not a neutral phenomenon but one that bears deep
traces of social and political structures and processes in society.
Te use of language, consequently, is always an activity that has
social and political dimensions: It can reproduce existing struc-
tures or challenge them, it can empower or disempower people,
and it can enfranchise and disenfranchise them .
State of Affairs
Te political load of language is one of the central concerns
for critical discourse analysis (CDA), an approach to dis-
course analysis that, especially since the 1990s, explicitly focuses
on the ways in which discourse refects power and social struc-
ture and constructs it (Fairclough 1989 , 1992 ; Blommaert 2005 ).
It is CDAs stated goal to analyze opaque as well as transparent
structural relationships of dominance, discrimination, power
and control as manifested in language (Wodak 1995 , 204) a
paradigmatic choice that is refected in numerous studies on rac-
ism, sexism, media, and political discourse and advertisements.
In all of these, linguistic and textual patterns are analyzed as con-
duits for hegemony and power abuse, and CDA has been infu-
ential in identifying registers and genres of power and control.
CDA clearly subscribes to a view of language as loaded (cf.
Bolinger 1980 ) and as invested with social and political interests
that steer discourse into particular, structural (i.e., nonarbitrary)
patterns of use and abuse. Te infuence of Foucault, Gramsci,
Bourdieu, and other critical theorists is explicitly acknowledged
in much CDA work.
Te same paradigmatic choice underlies work in what
could be called critical sociolinguistics: an approach in which
the distribution of language in society is also seen as a refec-
tion of power processes, often crystallized in normative (stan-
dards) discourses and invariably entailing judgments of users
through judgments of language use (e.g., Milroy and Milroy
1985 ; Cameron 1995 ; see standardization ). Variation in lan-
guage speaks to variation in society, and such forms of variation
Rather, I would suggest we read history backwards, starting from
the current approaches to the politics of language and look-
ing into those authors who are seen as formative now. From
that vantage point, two groups of authors stand out: authors
who developed a political view of language and authors whose
political-analytic work provides tools for scholars in the feld of
language. Te frst category is dominated by such scholars as M.
M. Bakhtin, V. N. Voloshinov, Roland Barthes, Michel Foucault,
and Pierre Bourdieu (see habitus); in the latter, Karl Marx (see
marxism) and Antonio Gramsci stand out.
Tis collection of authors and insights, it must be realized,
can only be discussed in a more or less coherent way when a
number of conditions are met. In particular, two presuppositions
are required:
(i) It is clear that refections of this kind are predicated on a
view of language as a social object (not a mental object); such
refections belong to the realm of a social theory of language.
(ii) Tey also are predicated on a view in which language
displays intricate connections with social structure: Either
language mirrors social structure (especially structures of
inequality) or it can become an instrument for changing
social structure.
Tese presuppositions ensure that the authors mentioned can
become interlocutors for current practitioners in the feld, and such
practitioners would then be clustered in applied felds, such as
discourse analysis (both linguistic and foucaultian),
sociolinguistics , and linguistic anthropology.
Te work of Bakhtin and Voloshinov has been infuential in
its emphasis on the social and political dimensions of a key fea-
ture of real language: its heteroglossic nature (see dialogism
and heteroglossia ). Heteroglossia stands for the presence
of multiple voices in an act of communication , and such
voices are intricately related to social formations and interests.
Whenever we communicate, thus, we engage with existing com-
plexes of social (and cultural) meaning, we insert ourselves in an
intertextual tradition in which such complexes make sense,
and we articulate interests, not only (neutral, self-contained)
meanings. In addition, the articulation of such interests is not
a unilateral and linear event. Bakhtin ( 1981 ) emphasizes the
importance of evaluative uptake in interaction his dialogical
principle in which every act of communication requires rati-
fcation by the other in order to be valid, that is, in order to be
meaningful. Tis process of ratifcation is evaluative : It is done
from within ordered complexes of forms-and-meanings in
which appropriateness, social roles, fuency, and other quality
attributes are specifed. Tus, even if I think I produce a cogent
story, my interlocutor may judge it to be of the mark because
what I say and how I say it do not qualify as good enough in his/
her evaluative framework. And evidently, such evaluative frame-
works are refections and instruments of the social and political
order (Voloshinov 1973 ) .
Tis social and political order penetrates language at a fun-
damental level: It shapes discourses. Discourses are complexes
of communicative forms (genres, styles) mapped onto thematic
and social domains, and what the social and political order does
is to create spaces in which particular discourses operate while it
eliminates other such spaces. Tis idea is central to the work of
Politics of Language
638
autonomous but cultural and social, and it displayed coherence
with other aspects of social and cultural patterning. In that sense,
grammatical form responded to collective patterns that orga-
nized social and cultural behavior, including linguistic behavior.
Te full richness of Whorfs approach was established by people
like Michael Silverstein ( 1979 ). Silverstein suggested that we
read Whorfs argument as follows: Linguistic form is indexical ;
it indexes aspects of context through ideological inferences: A
particular form stands for a particular social and cultural
meaning (also Silverstein 2003 ). Tus, in French, tu and vous
share a great deal of linguistic meaning but are diferentiated by
indexical meanings; tu indexes a low second person singular
addressee, while vous indexes a high second person singular
addressee. Te one who uses tu or vous would express indexically
his/her degree of respect and social distance toward the inter-
locutor, and the interlocutor would attribute conventional iden-
tity features, such as polite, proper, well educated, middle
class, and so on to the one using these forms. Tus, we select
linguistic (and wider semiotic ) forms in relation to socially and
culturally shared ideas about what would be appropriate, good,
useful, and salient communicative behavior in a specifc context,
and our use of semiotic means creates, supports, and manipu-
lates contexts.
Tis reconstruction of Whorfs foundational insight has signif-
icant implications. One efect is that it creates a new, but essen-
tially inseparable, layer to language structure: a metapragmatic
layer. Accepting that layer means that the analyst must accept
that whenever we communicate, we not only communicate in
our communication but also communicate about our commu-
nication: We always fag socially and culturally shared (ideo-
logical) indexical meanings while we talk, and these indexicals
make others perceive our talk as serious, arrogant, funny,
or knowledgeable. Te metapragmatics of language organizes
its pragmatics its meaning in society. And this, then, means
that approaches solely focused on a pragmatics of language risk
buying into commonly shared metapragmatic frames; in other
words, a normal linguistics always risks dragging along the
widespread language ideologies that dominate its object .
Another efect is that the range of variability in language is
vastly expanded, for the metapragmatic layer also provides an
enormous potential for social and cultural diferentiation ( dis-
tinction , to borrow Bourdieus term). In a nutshell, we can say
that every possible diference in language can become a socially
and culturally salient and important diference and that linguis-
tic diferences need not be big in order to generate important
social and cultural diferences .
Evaluation
Te idea that language is a politically invested object and that
people act politically in, through, and on language is by now a
well-established theoretical frame, the legitimacy of which no
longer requires debate. One reason is the fact that the diferent
approaches discussed here all have very strong empirical incli-
nations and that studies documenting the politics of language
often manage to transcend the slogans of a committed social
science and bring theoretical and methodological innovation to
the feld. CDA has done much to sensitize discourse analysts at
large about the fact that discourse matters to people because it
are evaluated given diferent value. Institutionalization, such
as, for example, in the education system (Rampton 1995 ) or in
bureaucracy (Sarangi and Slembrouck 1996 ) can stabilize and
reify such evaluative patterns and use them as normative, exclu-
sive, and excluding instruments of power and control. Sensitive
social identities, such as gender and immigrant identities, can
be especially vulnerable to exclusion or marginalization in such
reifed normative structures.
Both CDA and critical sociolinguistics seek an integration of
the linguistic or discourse-analytic method with social theory,
thus reversing the tendency toward autonomy and disciplinary
recognizability that characterized earlier phases in the develop-
ment of these disciplines (e.g., Cameron 1992 ; Chouliaraki and
Fairclough 1999 ). Tis move is aimed at strengthening the funda-
mental theoretical assumption: that language and social struc-
ture stand in an intricate relationship to each other and that one
cannot be understood without an understanding of the other .
From another theoretical angle, linguistic anthropology has
signifcantly contributed to the study of the politics of language.
In contrast to the previous schools, linguistic anthropology has its
roots in an integrated science of human behavior. Te anthropo-
logical notion of language, consequently, appears easier to inte-
grate into a mature social-theoretical framework than notions of
language that have their feet in twentieth-century linguistic tra-
ditions. Te fact that language forms and structures need to be
seen as refective and constructive of sociocultural and political
realities was central to Edward Sapirs work ( 1921 ), and the post
World War II reemergence of the ethnography of communication
(Gumperz and Hymes 1972 ; Gumperz 1982 ; Hymes 1996 ) started
from the assumption that there is no occurrence of language that
is not drenched in social, cultural, historical, and political con-
texts and that, consequently, can be understood without atten-
tion to these contexts (Duranti 1997 ). It is from within linguistic
anthropology that the paradigm of language ideologies devel-
oped (Schiefelin, Woolard, and Kroskrity 1998 ; Blommaert 1999 ;
Kroskrity 2000 ; Bauman and Briggs 2003 ) .
Language Ideologies
Language ideologies are beliefs, ideas, views, and perceptions
about language and communication. Such ideational complexes
pertain to every aspect of communication: about linguistic forms
and functions, as well as about the wider behavioral frames (often
called nonlinguistic) in which they occur. Tus, in the feld of
language ideologies, people are seen to perform meanings, and
language in the narrow sense of the term is seen as just one mode
of meaning production. People produce semiosis (meaningful
symbolic behavior) as performance , and they do so within a
regimented field in which language ideologies produce stability
and recognizability. Seen from that perspective, language ide-
ologies are of course not just ideational; they are practical in the
sense of Bourdieu, referring to the Marxian praxis , rather than to
the Mannheimian or Durkheimian notion of ideology .
Te study of language ideologies emerged out of the Whorfan
concern with connections between language form and world
view (Hill and Mannheim 1992 ). To recap Benjamin Whorfs
basic idea, he argued that grammatical categories encoded
and thus revealed aspects of collective perceptions of reality;
as such, grammatical organization was not random, logical, or
Politics of Language
639
Silverstein M . 1979 . Language structure and linguistic ideology. In
Te Elements , ed. P. Clyne , W. Hanks , and C. Hofbauer , 193247.
Chicago : Chicago Linguistic Society .
. 2003 . Indexical order and the dialectics of sociocultural life.
Language and Communication 23 : 193229.
Voloshinov , V. N . 1973 . Marxism and the Philosophy of Language .
Cambridge : Harvard University Press .
Wodak , R . 1995 . Critical linguistics and critical discourse analysis. In
Handbook of Pragmatics: Manual , ed. J. Verschueren , J. O. stman ,
and J. Blommaert , 20410. Amsterdam : John Benjamins
POSSIBLE WORLDS SEMANTICS
Possible worlds semantics is a family of semantic theories in
which the truth conditions of modal concepts and other inten-
sional locutions are expressed with the help of the concept of
possible world (scenario, possible state of afairs, possible course
of events). (See modality, intension and extension .)
Human beings constantly fnd themselves concerned with
what could happen or might have happened. Te modal notions
of possibility and necessity are used to cope with such situations.
Less directly, notions like knowledge, belief, obligation, permis-
sion, and so on serve the same purpose. Concepts behaving in
essentially the same way as necessity, knowledge, and so on are
known as intensional concepts. Modal notions have several vari-
eties, among them logical, conceptual, metaphysical, natural,
nomic, and physical modalities. When the diferent possibilities
can be weighted, one can also evoke the concept of probability.
It is nevertheless only relatively late that philosophers and
logicians came to think that in order to understand modal
notions (and other related notions), we have to consider unreal-
ized courses of events or states of afairs and, hence, merely pos-
sible worlds. Earlier philosophers usually did not think in such
terms. For one thinker, Aristotle, the only reality is the succession
of present moments outside of which there are no other possible
courses of events. Te idea of many worlds began its develop-
ment in the Middle Ages, encouraged by the famous condemna-
tion of 1277 of the view that God could not create other worlds.
Te notion of possible world was put to major metaphysical uses
by G. W. Leibniz for whom metaphysical truths are truths hold-
ing in all possible worlds.
In twentieth-century philosophical logic, the notion of possi-
ble world became prominent when modal logic was approached
from a model-theoretical or semantic point of view. Te use of the
notion of possible world in the study of modalities is analogous
to the measure-theoretical approach to probability theory, with
probability theorists sample-space points playing the same role
as logicians possible worlds. One of the pioneers of the seman-
tic study of modalities was Rudolf Carnap ( 1947 ), who explicitly
acknowledged the inspiration he received from Leibniz. Te
early treatments of the logic and semantics of modalities nev-
ertheless relied heavily on syntactical concepts and arguments.
For instance, Carnap represented possible worlds by sets of
sentences he called state-descriptions. A state-description is a
complete list of atomic sentences and the negations of atomic
sentences that are true in some model.
In such semisyntactical theorizing, interpretational questions
were neglected, relatively speaking. Fortunately, this neglect did
not initially matter. For what is the cash value of assuming that
is invested with power and social capital; critical sociolinguis-
tics has likewise drawn attention to the fact that sociolinguistic
distribution is not just a horizontal phenomenon but also a ver-
tical one: Sociolinguistic diference is complemented by socio-
linguistic inequality. And from within linguistic anthropology,
we have witnessed the emergence of a powerful ethnographic
paradigm that recovers the holistic and rich agenda developed
earlier by the likes of Sapir and Whorf and applies these insights
to an expanding feld of fundamental and applied topics of lan-
guage in society. Te language-ideological approach appears to
be the most promising one because of its compelling theoret-
ical coherence and empirical applicability, and it would beneft
adjacent disciplines if the central insight that every pragmat-
ics of language is accompanied by a metapragmatics would be
adopted .
Jan Blommaert
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bakhtin , M. M . 1981 . Te Dialogic Imagination: Four Essays .
Austin : University of Texas Press.
Barthes , R . 1957 . Mythologies . Paris : Seuil .
Bauman , R. , and C. Briggs . 2003 . Voices of Modernity: Language Ideologies
and the Politics of Inequality . Cambridge : Cambridge University Press .
Blommaert , J . 2005 . Discourse: A Critical Introduction .
Cambridge : Cambridge University Press .
Blommaert, J., ed. 1999. Language Ideological Debates . Berlin : Mouton de
Gruyter .
Bolinger , D . 1980 . Language: Te Loaded Weapon . London : Longman .
Bourdieu , P . 1991 . Language and Symbolic Power . Cambridge,
UK : Polity .
Cameron , D . 1992 . Feminism and Linguistic Teory . London : Macmillan .
. 1995 . Verbal Hygiene . London : Routledge.
Chouliaraki , L ., and N. Fairclough . 1999 . Discourse in
Late Modernity: Rethinking Critical Discourse Analysis .
Edinburgh : Edinburgh University Press .
Duranti , A . 1997 . Linguistic Anthropology . Cambridge : Cambridge
University Press .
Fairclough , N . 1989 . Language and Power . London : Longman.
. 1992 . Discourse and Social Change . Cambridge, UK : Polity .
Foucault , M . 1984 . Te order of discourse. In Language and Politics , ed.
M. Shapiro , 10838. London : Blackwell .
Gumperz , J . 1982 . Discourse Strategies . Cambridge : Cambridge University
Press
Gumperz , J ., and D. Hymes , eds. 1972. Directions in Sociolinguistics: Te
Ethnography of Communication . New York : Holt, Rinehart and
Winston
Hill , J ., and B. Mannheim . 1992 . Language and world view . Annual
Review of Anthropology 21 : 381 406.
Hymes , D . 1996 . Ethnography, Linguistics, Narrative Inequality: Toward
an Understanding of Voice . London : Taylor and Francis .
Kroskrity , P ., ed. 2000. Regimes of Language . Santa Fe, NM : SAR .
Milroy , J ., and L. Milroy . 1985 . Authority in Language: Investigating
Language Prescription and Standardisation . London : Routledge .
Rampton , B . 1995 . Crossing: Language and Ethnicity among Adolescents .
London : Longman .
Sapir , E . 1921 . Language: An Introduction to the Study of Speech . Orlando,
FL : Harcourt Brace .
Sarangi , S ., and S. Slembrouck . 1996 . Language, Bureaucracy and Social
Control . London : Longman.
Schiefelin , B ., K. Woolard , and P. Kroskrity , eds. 1998. Language
Ideologies: Practice and Teory . New York : Oxford University Press .
Politics of Language Possible Worlds Semantics
640
automatically help the identifcation of the same individual in
another world. Nor does the rest of the reference system help us
here. Tere must exist principles defning what counts as a single
individual across possible worlds. Teir totality can be called an
identifcation system .
Te nature of such identifcation has given rise to extensive
discussion and controversies. Te identifcation system codifed
in our language is largely independent of the reference system.
Indeed, there are two diferent kinds of identifcation actually
used in our conceptual system. An identifcation system can
be visualized as a kind of map shared by the possible worlds
between which the identifcation is to take place. In the most
common cases of identifcation, the map can be thought of as a
kind of universal registry of the relevant population. For instance,
if the fles of the Social Security System were to serve as such a
system, I would know who someone is if and only if I knew his or
her social security number. Such identifcation could be called
public. An idea of how the criteria of public identifcation could
work can be obtained by considering how we reidentify objects
over time. Continuity considerations obviously play a major role,
but questions as to how objects behave over time also come into
play .
An individuals position in someones perceptual space or
remembered role in someones past experiences can also serve
as a framework of identifcation. Te simplest framework of this
kind is someones visual space. Such forms of identifcation are
called perspectival . Among other expressions of our language,
demonstratives rely on perspectival identifcation. Teir oper-
ation is illustrated by Bertrand Russells onetime view that the
only logically proper names of English are this, that, and
I. Te explanation is that Russell tacitly presupposed only per-
spectival identifcation.
Te distinction between perspectival and public identifca-
tion gains further interest and robustness from the fact that these
two systems are, in the case of visual cognition, implemented
by diferent parts of the human brain (Vaina 1990 ; Hintikka and
Symons 2003). Since quantifers depend on identifcation, they
acquire a diferent meaning according to the kind of identifca-
tion presupposed.
Tese observations open the doors to extensive applica-
tions of logical languages with a possible worlds semantics. For
instance, a simple wh-statement Alonzo knows who (call him or
her x) is such that F[x] can be expressed by a sentence of the form
(x)K
Alonzo
F[x] where x ranges over persons. Tis shows how to
formalize simple knows who statements in general. For example,
(x)K
Alonzo
(Barbara = x) says that Alonzo knows of some particu-
lar individual x that Barbara is that x. Tis unmistakably means
that Alonzo knows who Barbara is. Such statements may be con-
trasted to K
a
(x)(b = x), which merely says that a knows that b
exists.
Tis kind of variation of operator ordering K versus (x)
cannot do the whole job, at least if we want to stay on the frst-
order level. In order to do so, we have to resort to a recently
introduced idea of operator independence (Hintikka 2003 ).
Since modalities are characterized by quantifcation over pos-
sible worlds, the same kind of independence can obtain between
modal operators and quantifers as between quantifers. Tis
independence can be expressed by a slash. Tus, we can express
possible worlds exist? According to Van Quine, such existence
means that we can quantify over them (see quantification ).
Te starting point of possible worlds semantics is the insight that
many modal and other intensional concepts can be construed as
quantifers over suitable classes of possible worlds. If NS means
it is necessary that S, it is true if and only if S is true in all pos-
sible worlds. It is possible that S, briefy PS, is true just in case
S is true in some possible world. If K
a
S means a knows that S,
it is true if and only if S is true in all the possible worlds not ruled
out by what a knows, and so on. Tus, the idea of possible worlds
was involved right from the beginning in the development of the
semantics of modal logic, following the work of Alfred Tarski and
his associates. (Cf. Copeland 2002 ; Kanger 1957 ; Hintikka 1957a ,
1957b ; Kripke 1959 .) Te frst to emphasize the role of possible
worlds semantics as the basis of general semantics seems to have
been Richard Montague (cf. Montague 1974 ).
Even though this is, for most purposes, an adequate expla-
nation of the meaning of K
a
S, the characterizations of necessity
and possibility need further specifcation, namely, an indication
of what kind of modality we are dealing with. For instance, not
all logically possible worlds are nomically (physically, naturally)
possible.
We thus seem to obtain a semantically interpreted language
by adding to a frst-order language the operator or those opera-
tors we are interested in. On the basis of this idea, we can develop
much of a viable modal logic, epistemic logic, and so on, as well
as the required methods of proof.
Tis procedure is not suf cient alone, however. For one thing,
the possible worlds that fgure in these explanations are relative
to the world w in which NS, PS, K
a
S, and so on are evaluated
semantically. Tey will be called alternatives to w . To deal with
iterated or multiple modalities, we have to consider alternatives
to alternative worlds, and so on. Te alternativeness relation
involved here is sometimes called the accessibility relation .
Tis does not yet completely determine the possible worlds
semantics. To see what is missing, consider how the references
of linguistic expressions are determined. Te guiding principle of
possible worlds semantics is that the application of a language,
including the reference of any expression e , in a given world w
must depend only on that world. Te way in which the reference
of e in w is determined is, therefore, codifed in the function f that
determines the reference of e as a function f ( w , e ) only. We could
call the totality of these functions the reference system of the
language. For instance, the reference of the 44th president of
the United States is whoever wins the 2008 election.
Tese ideas of reference and meaning are, in fact, the corner-
stone of the version of possible worlds semantics most exten-
sively used in linguistics, known as Montague semantics (see
montague grammar ). It was developed by Montague ( 1974 )
and applied in linguistics most vigorously by Barbara Partee
( 1976 , 1989 ).
Tere are further problems in the development of possible
worlds semantics, however. When we use a quantifer, we con-
sider each of its values as being the same individual in diferent
possible worlds. But how can such identities be recognized?
Tey cannot be established by examining the diferent possible
worlds in question independently of one another. For instance,
a name-conferring (dubbing) ceremony in one world does not
Possible Worlds Semantics
641
. 2003 . A second generation epistemic logic and its general signif-
cance. In Knowledge Contributors , ed. by V. Hendricks , K. F. Jrgensen ,
and S. A. Pedersen , 3356. Dordrecht, the Netherlands : Kluwer
Academic .
Hintikka , Jaakko , and John Symons . 2003 . Systems of visual identifca-
tion and neuroscience: Lessons from epistemic logic . Philosophy of
Science 70 : 89104.
Kanger , Stig . 1957 . Provability in Logic . Stockholm Studies in Philosophy.
Vol. 1. Stockholm: Almqvist and Wicksell.
Kripke , Saul A . 1959 . A completeness theorem in modal logic . Journal
of Symbolic Logic 24 : 114.
Montague , Richard . 1974 . Formal Philosophy: Selected Papers by
Richard Montague , ed. by Richmond Tomason. New Haven, CT : Yale
University Press .
Partee , Barbara H . 1989 . Possible worlds in model-theoretic seman-
tics: A linguistic perspective. In Possible Worlds in Humanities, Arts
and Sciences: Proceedings of Nobel Symposium , ed. by S. Alln , 93123.
Berlin and New York : Walter de Gruyter .
Partee , Barbara H ., ed. 1976. Montague Grammar . New York : Academic
Press .
Vaina , Lucia . 1990 . What and Where in the Human Visual System: Two
Hierarchies of Visual Modules . Synthese 83 : 4991.
POSSIBLE WORLDS SEMANTICS AND FICTION
Te applications of the philosophical concept of possible world
to narrative and to fction were frst developed in the late 1970s
and early 1980s as a reaction to structuralist poetics, a
movement that adhered to Ferdinand de Saussures conception
of language as a self-enclosed system of signs. As Tomas Pavel
has argued, this theoretical position led to a moratorium on
representational topics (1986, 6) and on the notion of reference
to a world external to language. In its literary applications, pos-
sible worlds (hence, PW) semantics is an attempt to restore the
relevance of mimesis, reference, and the question of truth with-
out reducing the fctional text to an image of reality.
Te logician Jaakko Hintikka ( 1989 ) describes the concep-
tion of language to which PW semantics seeks an alternative as
language as the universal medium. According to this view, all
that language is good for is to enable us to talk about this world
(Hintikka 1989 , 54). Te primary target of his description is posi-
tivist philosophies that limit reference to an objectively existing
external reality, such as those of Gottlob Frege, Bertrand Russell,
and the early Ludwig Wittgenstein. For the positivist, a statement
concerning a nonexisting entity, such as Santa Claus or Emma
Bovary, is either false or indeterminate. It is, therefore, impossible
to diferentiate the validity of statements made about imaginary
beings. Structuralism and deconstruction go even further in
their interpretation of language as universal medium by regard-
ing it as the unique reality to which it is capable of referring.
To the conception of language as universal medium, Hintikka
opposes what he calls language as calculus. In this framework,
you can so to speak stop your language and step of. In less met-
aphoric terms, you can discuss the semantics of your language
and even vary systematically its interpretation. Te operative
word highlights the thesis that language is freely interpretable
like a calculus (1989, 54). By virtue of this reinterpretability, lan-
guage can be directed toward diferent domains of reference and
the truth value of propositions established separately for each
(x)K
a
(b = x) equivalently as K
a
(x/K
a
)(b = x), where x/K
a
means that is independent of K
a
. (Notice that by so doing, we
can stay closer to the structure [word order] of the corresponding
English knowledge statements.) In more complicated cases, we
can, for instance, express a knows which function g(x) is by
K
a
(x)(y/K
a
)(g(x) = y).
We cannot stay on the frst-order level here without the inde-
pendence indicator.
Tis does not clear up all interpretational problems, however.
We can still ask: What are the relevant possible worlds in difer-
ent applications? Tis question is connected with the question as
to what kinds of modalities and other intensional notions there
are.
Te characterization of possible worlds as represented by
maximal consistent classes of sentences of a given language has
encouraged the idea that what is intended by possible worlds are
indeed worlds in the sense of entire universes. However, a com-
parison with probability theory shows that such grandiose inter-
pretations are neither unavoidable nor even preferable. In most
applications of probability theory, the possible worlds (sample
space points) are not worlds in any ordinary sense of the word.
Tey usually are what might be called scenarios , namely, courses of
events involving a small region of space-time, for example, tosses
of a die. Some probability theorists speak of small worlds , and
practically all realistic applications of possible worlds semantics
are to such small worlds. In some of his work, Montague, in fact,
operates with contexts of use, rather than possible worlds.
Tere remains the question of diferent modalities. Are they
all viable in the light of possible world semantics? Tere are no
major unsolved conceptual problems about epistemic or dox-
astic modalities or other similar intensional concepts. Te class
of alternative epistemic worlds has a clear meaning or at least
as clear a meaning as our language has. Logical (conceptual)
modalities are interpretable only if we look at the structure of
the possible worlds that they involve. If we begin to speak of all
possibilities of individual existence, the alternatives to a given
world do not form a viable class anymore than the set of all sets
in set theory. Te idea of natural possibility has a clear sense if it
is taken to mean conformity with natural laws (nomic necessity).
But when it is claimed that there exists a metaphysical necessity
separate from nomic (physical) and conceptual (logical) neces-
sities, it is hard to see what is being meant. It is not enough to
claim that we have intuitions about them, for the very notion of
intuition in its current philosophical use is highly suspect .
Jaakko Hintikka and Risto Hilpinen
WORKS CITED AND SUGGESTIONS FUR FURTHER READING
Carnap , Rudolf . 1947 . Meaning and Necessity: A Study in Semantics of
Modal Logic . Chicago : University of Chicago Press . An enlarged edi-
tion appeared in 1956.
Copeland , B. Jack . 2002 . Te genesis of possible worlds semantics .
Journal of Philosophical Logic 31 : 99137.
Hintikka , Jaakko . 1957 a. Quantifers in Deontic Logic . Helsinki: Societas
Scientiarum Fennica (Commentationes Humanarum Litterarum, Vol.
13.4).
. 1957 b. Modality as referential multiplicity . Ajatus 20 : 4964.
Possible Worlds Semantics Possible Worlds Semantics and Fiction
642
hypothetical statement describes another world from the point
of view of the actual world.
For the duration of our immersion in a work of fction, we
regard or, rather, pretend to regard its world as actual. Tis
pseudoactuality is produced by a gesture of imaginative recen-
tering of the reader, spectator, or player into the fctional world
(Ryan 1991 , 212). Te experience of fction has been compared
by Kendall Walton ( 1990 ) to a game of make-believe, but what
exactly is it that we pretend to believe when we immerse our-
selves in a work of fction? PW theory, and more specifcally the
indexical theory of actuality, spells out the rules of this game as
pretending to believe that fction describes a world that is both
real and actual. Pretending that this world is real means pretend-
ing that it exists independently of the text, while pretending that
it is actual means transporting oneself in imagination into this
world and adopting the point of view of one of its members .
Another of Lewiss contributions to the theory of fction is the
elaboration of an algorithm for determining the truth or falsity
of interpretive statements made by the reader about fctional
worlds (1978). Tis algorithm is adapted from his famous analysis
of the truth conditions of counterfactual statements (1973).
Lewiss criterion for establishing the truth value of statements
about fction, such as Emma Bovary was a devoted mother, can
be paraphrased as follows: A sentence of the form in the fction
f, p is true when some world where f is told as known fact difers
less, on balance, from the actual world than does any world where
f is told as known fact and p is false. Tis formula tells us that
Emma was not a model mother because, in order to accept this
interpretation, we would have to assume that the fctional world
adheres to a set of standards of good motherhood vastly diferent
from the values of our cultural corner of the actual world, though
nothing in the text authorizes such an assumption.
Lewiss algorithm entails a fundamental principle for the
phenomenology of reading. Tis principle, called by Marie-
Laure Ryan the principle of minimal departure, states that
when readers construct fctional worlds, they fll in the gaps in
the text on the basis of their experience of the actual world, and
they will not make gratuitous changes. Te principle of minimal
departure can only be overruled by the text itself. For instance, if
a work of fction mentions an elephant, the reader will imagine
the elephant as huge and gray unless the text describes it as a
polka-dotted pet the size of a chihuahua. Even then, the reader
will imagine that the elephant has thick skin, big ears, and tusks .
Te assimilation of fctional worlds a concept often used
by critics informally to the more technical notion of possible
worlds can claim Aristotle as its forefather: As he writes in the
Poetics , the function of the poet is not to say what has happened,
but to say the kind of thing that would happen, i.e., what is pos-
sible in accordance with probability and necessity (1996, 16,
par. 5.5). But even if one extends the notion of possibility beyond
what could happen in our world so as to include the logically
coherent but nomologically impossible worlds of science fction
and fantasy, the straightforward assimilation of fctional worlds
to possible worlds encounters dif culties .
Logicians consider possible worlds to be maximal states of
afairs, meaning by this formula that every proposition is either
true or false in a given world. But fctions are created by texts,
and texts can only assert a limited number of propositions. To
of these domains. A statement can consequently be false in one
domain and true in another, and it becomes possible to assign a
positive truth value to the statement Emma Bovary committed
suicide by swallowing arsenic for the world of Gustave Flauberts
novel, even though the sentence is false in the real world (unless
we prefx it with in Flauberts novel) .
Hintikkas conception of language as calculus relies on an
ontological model made of a plurality of worlds. A common justif-
cation for the postulation of multiple worlds is the intuitive notion
that things could have been diferent from what they are. Saul
Kripke formalized this intuition through a model that describes
reality the sum of the thinkable as a set of elements hierarchi-
cally structured by the opposition of one element, which can be
interpreted as the actual or real world, to all the other members of
the system. Kripke envisions a relation of accessibility that links
the actual world to those worlds that are possible but not actual.
Worlds not linked to the central element of the system are consid-
ered impossible worlds, but when the relation of accessibility is
interpreted as respect for the laws of logic (noncontradiction and
excluded middle), one may debate whether they are worlds at
all, rather than incoherent collections of propositions. Tere are,
however, other interpretations of accessibility that preserve the
world status of the inaccessible elements: for instance, nomologi-
cal (respect of the laws of nature), epistemic (distinguishing what
is known, believed, and ignored), and deontic (based on what is
allowed, obligatory, and forbidden).
A question raised by Kripkes model is what distinguishes
the actual world from all the other members of the system.
According to a widespread view that may be called absolutist, the
actual world difers in ontological status from merely possible
ones in that this world alone presents an autonomous existence;
all the other worlds are the product of a mental activity, such
as dreaming, imagining, foretelling, promising, or storytelling.
David Lewis ( 1986 , 8491) proposes an alternative to the absolut-
ist view known as modal realism. For Lewis, all possible worlds
are equally real, and all possibilities are realized in some world,
independently of whether somebody thinks of them or not. But if
all possible words are real, how does one pick one of these worlds
as actual? Lewis answers this question through an indexical con-
ception of actuality. Te reference of the expression the actual
world varies with the speaker, like the reference of the deictics I ,
you , here , and now (see deixis ). All possible worlds are conse-
quently actual from the point of view of their inhabitants.
Te indexical conception of actuality is very important for
the description of the readers experience of fction. We nor-
mally think of fctional worlds as imaginary and as nonexisting.
We know that in contrast to our world, they are produced by a
human mind, the mind of the author. But this does not explain
how we relate to them. In contrast to hypothetical and counter-
factual statements, whose reference to an imaginary world is
stressed by the conditional mode or by an if then construct,
fctions are narrated in the indicative mode and, therefore, hide
the nonactual status of their reference world. Lewis accounts for
the formal similarity between fction and statements of facts by
characterizing fction as a story told as true about a world other
than the one we regard as actual by a narrator situated within
that other world. A nonfctional story, by contrast, is told as true
about our world by one of its members, and a counterfactual or
Possible Worlds Semantics and Fiction
643
. 1986 . On the Plurality of Worlds . Oxford : Blackwell .
Matre , Doreen . 1983 . Literature and Possible World s. London: Middlesex
University Press . Proposes a typology of fctions based on types of
possibility.
Martin , Tomas . 2004 . Poesis and Possible Worlds . Toronto : University
of Toronto Press . Investigates the consequences of Jaakko Hintikkas
notions of language as universal medium versus language as calculus
for literary theory.
Pavel , Tomas . 1986 . Fictional Worlds . Cambridge : Harvard University
Press .
Ronen , Ruth . 1994 . Possible Worlds in Literary Teory .
Cambridge : Cambridge University Press . A critique of the use of the
concept of possible world by literary theorists.
Ryan , Marie-Laure . 1991 . Possible Worlds, Artifcial Intelligence and
Narrative Teory . Bloomington : University of Indiana Press .
Walton , Kendall . 1990 . Mimesis as Make-Believe: On the Foundations of
the Representational Art . Cambridge : Harvard University Press .
PRAGMATIC COMPETENCE, ACQUISITION OF
Te acquisition of pragmatic competence involves the devel-
opment by the frst or second language learner of a wide range
of skills and capacities in using language to interact. Together
with linguistic competence, it forms the core of communicative
competence, as defned by D. Hymes ( 1967 , 1992 ; Foster-Cohen
2001 ) and includes how learners develop the ability to convey
and interpret linguistic (and nonlinguistic) messages; how they
expand their repertoire of communicative acts; how they come
to modulate those acts for such features as directness, polite-
ness , and informativeness; and how they design their messages
so that they conform to the social norms of their community. It
also includes the developing ability to construct linguistic units
larger than an utterance, including extended conversational
exchanges, stories, descriptions, explanations of procedures,
and so on, as well as the very basic capacity to enter into inter-
actions with other people and the turn-taking, topic initiation,
topic maintenance, and exchange opening and closing behav-
iors upon which any interaction depends.
Work in the acquisition of pragmatics has often resulted in
extended taxonomies of the speech-acts and types of talk that
learners acquire. A. Ninio and P. Wheeler ( 1984 ), for example,
developed a taxonomy of 70 distinct types of talk interchanges
acquired by children, categorized at an exchange level as negoti-
ations , discussions , performances , and so on and at the utterance
level in terms of communicative acts, such as directives, declara-
tions, statements, questions, and the responses to each of these.
Te taxonomy represents a hypothesis about the organization
of the mental representation of communicative intents in the
mind (Ninio and Snow 1996 , 39) and has its roots in speech-act
theory, pioneered by Austin and Searle (see Searle 1969 ) .
Other work has attempted to explore what is needed for chil-
dren to be able to behave in ways that others accept as natural
and efective language use in a given community. E. Andersen
( 1990 ), for example, examined childrens growing knowledge of
the language appropriate to roles with signifcant power diferen-
tials, such as doctor versus patient or teacher versus pupil; others
have explored issues such as how children develop gender-dif-
ferentiated ways of speaking (see gender and language ).
Other studies have explored what kinds of social, cognitive, and
take a famous example, the tragedy of Shakespeare implies that
Lady Macbeth had children, but it does not specify the number
of these children. Should one regard fctional worlds as radically
incomplete, as does Lubomr Doleel ( 1998 , 223), a view imply-
ing that Lady Macbeth is a creature who lacks the feature having
a determinate number of children? Or should one apply mini-
mal departure and assume that by presenting her as a human
being, the text invites the reader to regard the number of her
children as unavailable information as would be the case for
a fesh-and-blood woman rather than as an ontological lack?
Waltons concept of make-believe ofers a compromise between
these two interpretations: While readers know that fctional
worlds are the product of a fnite number of textual assertions,
they imagine these worlds and their inhabitants as ontologically
complete.
Another problem with regarding fctional worlds as possible
worlds is the existence of fctions that do not respect the laws of
logic and consequently fail to satisfy the broadest notion of pos-
sibility. In logic , a single contradiction in a group of proposi-
tions renders the system radically inconsistent because this
contradiction allows everything (and its opposite) to be inferred.
Tis makes it impossible to imagine a world. But in a fctional
text, transgressions of logic are not totally incompatible with the
mental construction of worlds. Logically impossible objects or
events can be limited to certain areas, comparable to the holes in
a Swiss cheese, and the reader remains capable of drawing infer-
ences for the solid parts of the cheese. We can still imagine the
world of a time-travel story that presents impossible causal loops
or of a fantastic tale situated in an inconsistent space. But some
fctions generalize contradiction by systematically negating what
they assert or by presenting dreamlike situations that continually
morph into other situations. Te reader of these texts can only
construe fragments of worlds that do not ft together. All fctions
project a set of meanings, but if we conceive worlds as relatively
stable totalities populated by individuals whose evolution main-
tains some continuity, the extent to which these meanings form
a world is variable. A fction made of incompatible world frag-
ments blocks the experience of immersion because it does not
ofer a target for the recentering of the imagination .
Marie-Laure Ryan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aristotle . 1996 . Poetics . Trans. and intro. Malcolm Heath. New
York : Penguin .
Doleel , Lubomr . 1998 . Heterocosmica: Fiction and Possible Worlds .
Baltimore, MD : Johns Hopkins University Press .
Eco , Umberto . 1984 . Te Role of the Reader: Explorations in the Semiotics
of Texts . Bloomington : Indiana University Press . Outlines a semantics
of narrative based on possible worlds.
Hintikka , Jaakko . 1989. Exploring possible worlds. In Possible Worlds in
Humanities, Arts and Sciences: Proceedings of Nobel Symposium 65 , ed.
Sture Alln, 5273. Berlin : de Gruyter .
Kripke , Saul . 1963 Semantical considerations on modal logic . Acta
Philosophica Fennica 16 : 83 94.
Lewis , David . 1973 . Counterfactuals . Cambridge : Cambridge University
Press .
. 1978 . Truth in fction . American Philosophical Quarterly
15 : 37 46.
Possible Worlds Semantics and Fiction Pragmatic Competence, Acquisition of
644
the interlocutor, rather than for its ability to seek reasons and
motivations, which at that stage are beyond the childs level of
cognitive development .
Not long after children acquire the ability to produce individ-
ual communicative acts, they begin to combine them into larger
language acts and events, such as stories, explanations, and
complex observations. Narratives range from the simple state-
ment of a problematic event in the past followed by a solution
(Te baby cried; the mummy picked it up) to heavily scafolded
productions such as the following:
Ross (aged 2;6) and his mother are sitting together eating a
snack.
R: Sometimes Ross come out bed bed come out night.
M: What are you talking about? What about that bed at night?
Sometimes you what the bed at night?
R: Mmm.
M: What did you say?
R: In the dark.
M: In the dark!
R: Ross, erm, Ross runs in the dark.
M: Run in the dark?!
R: Ross runs.
M: You get out of the bed in the night did you and ran around
in the dark. Tat sounds a daft thing to do! (Foster-Cohen
1990 )
Here, Rosss mother helps him get his story out piece by piece
and puts it together for him.
As children develop their story skills, we start seeing complex
depictions with recognizable phases and characteristic packag-
ing of information (Labov and Walestsky 1967 ). Te literature on
childrens narratives has been enhanced by several large-scale
studies, such as that by Ruth Berman and Dan Slobin , whose col-
lection of stories told in response to wordless books about a small
boy and a frog (known, naturally enough, as Te Frog Stories)
has provided a cross-linguistic, cross-cultural view of how chil-
dren develop the ability to tell a story (Berman and Slobin 1994 ;
Strmqvist and Verhoeven 2004 ) .
Stories and other large discourse units are held together via
the coherence of their informational structure and by the
markers of cohesion that link individual utterances to each other.
Te presentation of new information in relation to assumed or
known information is one key aspect of coherence, and requires
children to be able to infer what their interlocutor knows and
to structure the information provided accordingly. As such, the
development of coherence in childrens narratives, and in their
language use generally, is dependent on the evolving under-
standing of other minds (see theory of mind and language
acquisition ). When there is diferent and conficting informa-
tion held by the child narrator and by a protagonist in the story,
it places considerable strain on the young childs pragmatic
competence. An example occurs in the Frog, where are you?
story. At one point, the boy in the story grabs what he believes to
be branches but the narrator knows to be the antlers of a deer.
Children struggle with how to represent this conficting informa-
tion, as the following representative samples from Berman and
Slobins ( 1994 ) work suggest:
linguistic developments are needed before children can engage in
particular acts. J. Bernicot and V. Laval ( 2004 ), for example, have
explored how children learn to make a promise or understand
one. Tis kind of work addresses childrens growing understand-
ing of how the interpretation of linguistic expressions depends
crucially on inferencing by correctly processing both the language
that is spoken and a wide range of contextual characteristics that
go well beyond the words actually uttered. In fact, the exploration
of childrens capacity for drawing inferences in communication
is a healthy area of research in its own right. I. Noveck ( 2001 ), for
example, has explored childrens interpretations of scalar impli-
catures (see conversational implicature ) and suggested
that in certain ways, children are more logical in their interpre-
tations of words, such as some, than are adults .
Pragmatic Acquisition from Birth to Adolescence
Infants are surprisingly communicative, even from birth. Perhaps
because of mirror neurons (Brten 2007 ), newborns respond
in kind to the social advances of adults and are soon able to initi-
ate social exchanges with others through eye gaze, movement,
and vocalization (see communication, prelinguistic ), as
well as to engage in elementary turn-taking. Tis is the beginning
of the development of pragmatic competence.
Elizabeth Bates and colleagues (Bates, Camaioni, and Volterra
1975 ) suggested that Austins tripartite distinction between per-
locution , illocution , and locution might be harnessed to
describe what happens in the frst year of life. Tey suggested that
when an infant communicates a message (such as Im hungry)
through a cry but without the intention to communicate, it can
be seen as a sort of perlocutionary event (albeit without either
illocution or locution). Ten, with the emergence of gestures
and specifc patterns of eye gaze, illocutionary forces, such as
requests (perhaps achieved via a reaching hand coupled with
a glance at the interlocutor and a glance at the desired object)
or making an observation (perhaps through a similar eye gaze
pattern, but this time coupled with a pointing hand shape), can
be observed while still in the prelinguistic phase and therefore
prelocutionary. Finally, the emergence of recognizable words
adds the capacity for locution. Other researchers, such as John
Dore ( 1975 ) and Michael Halliday ( 1975 ), have also added to our
understanding of just how rich very young childrens capacity
for pragmatic competence is before the emergence of signifcant
expressive language.
As productive expressive language emerges, from around
the frst birthday, the communicative resources of the child
expand and go on expanding over the next two decades and,
for some, throughout life. Some communicative acts are eas-
ily accomplished with quite simple linguistic resources (greet-
ings, leave-takings, agreements, and labeling of objects, for
example). Others, such as giving explanations, asking questions,
providing defnitions, or making hypotheses, are dependent on
a more sophisticated level of language development. So, while
yes/no questions can be asked with only a rise of intonation
(Out?, for example), open-ended questions require at least
some control of the wh-questions forms, for example, What
Daddy do? and How you go there? Interestingly (and often
irritatingly for the parent), the ubiquitous Why? seems to be
used in the frst instance for its ability to elicit a response from
Pragmatic Competence, Acquisition of
645
tages and those clinching business deals know, they may hone
their pragmatic skills for the rest of their lives .
Developing Pragmatic Competence in a Second Language
Most of the work on the acquisition of pragmatics has been car-
ried out within frst language research circles. However, there is
now a thriving research stream in second language pragmatics,
pioneered most notably by Gabriele Kasper (Kasper and Rose
2002 ). Te signifcance of this work for those working in devel-
opmental pragmatics with children lies in the help it provides
for teasing apart those aspects of pragmatics that are pancultural
and part of the human makeup and those that are specifc to par-
ticular language and cultural groups.
Research suggests that second language pragmatics is noto-
riously dif cult to learn. Tere are a number of possible reasons.
One is that, unlike grammar, inappropriate pragmatics is gener-
ally perceived by the other party as another pragmatic message.
So, inability to respond to a compliment as a native speaker
might is perceived as ungratefulness or rudeness. Overlapping
another persons speech in a way that is not native can be per-
ceived as interrupting. As a result, learners may not receive the
kind of feedback they actually need to adjust their pragmatics
toward the native-speaker norms. Another reason is that while
learning the grammar of a second language can be perceived
by the learner as simply the learning of a code (another way of
saying something), learning the pragmatics of another language
group is learning that groups culture and, as such, is felt more
deeply and more personally. As a French learner of English
once said when I was trying to teach the pragmatics of English,
I want to learn English; I dont want to be English. Finally, and
relatedly, the reason second language pragmatics is so hard to
acquire is because researchers and pedagogues are not very
good at describing it in such a way that learners can, in fact,
learn it .
Disorders of Pragmatic Competence
Because the development of pragmatic competence depends
on a number of intersecting developments, there are multiple
ways in which it can be derailed or curtailed. Dif culties with
understanding the nature of social engagement, of inferring the
knowledge and intentions of another person, or of processing
the subtle cues of verbal and nonverbal communication can all
impact on a learners capacity to become pragmatically compe-
tent. Te most well known disorder of pragmatic competence is
autistic spectrum disorder, a condition which comes in a variety
of forms and degrees of severity. It can impact all of the afore-
mentioned prerequisites for pragmatic competence . A variety
of other developmental conditions (including Down syndrome,
Williams syndrome, and global developmental delay) also
impact on pragmatic development. More avoidable are those
disruptions to pragmatic development that come about as a
result of neglect or abuse and the consequent failure to attach
efectively to one or more key people. It is now quite clear that
the development of social relationships, and with it pragmatic
competence, is dependent on experience in responsive and
pragmatically appropriate relationships. Even an intact child
can have his or her pragmatic competence derailed by poor
experiences .
He hops on the deer. (4.7; no understanding of the boys mis-
judgment of the branches)
Ten he got on a reindeer, because the reindeer was hiding
there. (5.2; understanding that the reindeer was not initially
visible to the boy, but no attention to the boys state of mind)
He got picked up by a reindeer. (5.8; use of get-passive sug-
gests the narrator is aware that the boy was not an intentional
agent)
Hes holding on to some sticks. But they arent really sticks.
When uh something came up, and the little boy was on it.
Um it was a father deer, Id call it. (5.10; explicit recogni-
tion of the boys misperception, though from the point of view
of the narrator, rather than the boy)
He thought it was sticks and he got on that and the deer
came and carried him. (5.11; explicit attribution of misper-
ception to the boy; groping for means of encoding the unin-
tentionality of the consequences)
And then he stands up on the rock and hangs onto some
branches. Ten it turns out theyre a deers antlers. So and
he gets he lands on his head. (9.11; the turns-out construc-
tion provides a means of encoding the switch in perspective,
and the interrupted he gets suggests a groping for a passive
construction)
And fnally, here is an adult version: When he gets to the top
of the rock, he holds onto something that he app thinks are
branches, and calls to the frog. And what the boy took to be
branches were really antlers of a deer on which he gets caught.
Tese samples can also be used to illustrate cohesion markers.
In particular, we can see how the deer is introduced by the youn-
gest child with a defnite article, as if its presence were already
known. Almost all the other older children use the indefnite
article appropriate to a frst mention. All the children use he
to refer to the boy, which works in these examples. However, in
another example, taken from elsewhere in the database And
he [the deer] starts running. And he [the deer] tips him of over a
clif into the water. And he [the boy] lands we can see how the
use of pronouns undergoes development in order for the hearer
to keep track of the protagonists reliably. If the simple story
referred to had been Te baby cried; the Mummy picked Johnny
up, we would be forgiven for wondering whether Johnny is the
baby or someone else .
As they develop, children are able to rely more and more
on their own skills in pragmatics and depend less and less on
a cooperative other to make sense of what they are trying to
say. As a result, their interactions with their peers can begin to
mature, and they can develop the skills for working and learn-
ing cooperatively with children their own age. Tese conversa-
tions are often much more combative than any conversation
between nurturing parent and child, and children need to, and
do, develop important skills for repairing the misunderstandings
that inevitably arise. However, unlike grammatical development,
which is largely complete by the age of fve, pragmatic compe-
tence keeps on developing. Teenagers continue to develop their
skills of staying on topic, interrupting appropriately, showing
empathy, and entertaining others by telling jokes and acting out
stories and events (Nippold 2000 ). Moreover, as professional
orators, stand-up comics, negotiators seeking the release of hos-
Pragmatic Competence, Acquisition of
646
Nippold , M . 2000 . Language development during the adolescent
years: Aspects of pragmatics, syntax, and semantics . Topics in
Language Disorders 20.2 : 15 28.
Noveck , I. 2001 . When children are more logical than adults: Investigations
of scalar implicature . Cognition 78 : 165 88.
Searle , J . 1969 . Speech Acts . Cambridge : Cambridge University Press .
Sperber , D. 1996 . Explaining Culture: A Naturalistic Approach .
Cambridge, MA : Blackwell .
Strmqvist , S. , and L. Verhoeven . 2004 . Relating Events in Narrative.
Vol. 2. Typological and Contextual Perspectives. Mahwah, NJ : Lawrence
Erlbaum .
PRAGMATICS
Pragmatics refers to the study of meaning in context. Consider,
for example, the following exchange between two close friends:
Harvey: Are you going to the big party tonight?
Molly: Didnt you hear that Jason would be there?
How does Harvey interpret Mollys response to his question?
Although Mollys response is itself a question, it is considered
an appropriate answer to Harveys original question, at least
in this context, given the assumption that Harvey knows what
Molly feels about Jason. Of course, listeners who do not know
how Molly feels about Jason would be unable to infer whether
Molly implies yes or no by her response. But the informa-
tion that Harvey and Molly share about Jason, and particularly
Mollys thoughts about Jason, such as that he is an ex-boyfriend
whom she wishes to avoid, should allow Harvey to easily infer
what Molly means by what she says.
Peoples pragmatic understanding of speakers utterances
in context is assumed to rely on their general knowledge of the
world, the specifc discourse context, and what they know about
their interlocutors. Pragmatics is seen as distinct from semantics
in referring to contextual meaning, as opposed to context-invari-
ant word meaning or sentence meaning , and is also viewed as
being associated with what speakers imply, as opposed to what
they literally say . Philosophers interested in ordinary lan-
guage use , and not more narrowly semantic meaning, launched
the study of pragmatics in the late 1950s. For instance, J. L. Austin
( 1962 ) described the ways in which people use words to accom-
plish diferent social actions, and he demonstrated that speakers
typically intend to communicate diferent or additional meanings
beyond what their words literally say (see performative and
constative ). Tus, when a speaker says, Ill lend you fve dol-
lars, she communicates a promise to actually give the listener $5.
Te philosopher John Searle ( 1975 ) later argued that there were
only fve major types of speech-acts by which speakers perform
acts with diferent illocutionary force , including:
Representative or Assertive : Te speaker becomes committed
to the truth of the propositional content of an utterance; such
as asserting Te sun is shining today.
Directive : Te speaker tries to get the hearer to fulfll what
is represented by the propositional content of an utterance,
such as Please stop talking.
Commissive : Te speaker commits to act in the way repre-
sented by the propositional content of an utterance, such as
Ill lend you fve dollars.
How Pragmatics Is/Are Acquired
As with any human development, pragmatic competence is
acquired as a result of both biological design and social experi-
ence. As already indicated, the acquisition of pragmatic compe-
tence depends crucially on nonlinguistic factors, such as innate
social responsiveness, the development of real-world knowledge,
and general problem-solving ability. It has also been claimed to
depend more heavily on overt instruction. Children are observed
being taught to be polite, to follow the rules for trick or treat
at Halloween, or to adjust their messages to take account of the
other persons knowledge and seem more able to learn from
correction and overt modeling than they are when their gram-
matical errors are corrected. However, given the complexity of
pragmatic development, the huge amount of unconscious infer-
encing it requires, and the dif culty understanding what has
gone wrong when pragmatic rules and expectations have been
violated, it is unlikely that more than the most codifed of prag-
matic skills (politeness formulae, terms of address, fxed events
such as trick or treat) are acquired in this manner. Rather, prag-
matic skills are caught through cultural contact and spread
by epidemiological principles (Sperber 1996 ) .
Susan Foster-Cohen
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Andersen , E . 1990 . Speaking with Style: Te Sociolinguistic Skills of
Children. London : Routledge, Kegan and Paul .
Bates , E ., L. Camaioni , and V. Volterra . 1975 . Te acquisition of perfor-
matives prior to speech . Merrill-Palmer Quarterly 21 : 205 26.
Berman , R ., and D. I. Slobin 1994 . Relating Events in Narrative: A
Crosslinguistic Developmental Study . Hillsdale, NJ : Lawrence
Erlbaum .
Bernicot , J ., and V. Laval . 2004 . Speech acts in children: Te example
of promises. In Experimental Pragmatics, ed. Ira Noveck and Dan
Sperber , 20727. Basingstoke, UK : Palgrave Macmillan .
Brten , S ., ed. 2007. On Being Moved: From Mirror Neurons to Empathy .
Amsterdam : John Benjamins .
Dore , J . 1975 . Holophrases, speech acts and language universals .
Journal of Child Language 2 : 20 40.
Foster-Cohen , S . 1990 . Te Communicative Competence of Young
Children. Harlow, England : Longman.
. 2001 . Communicative competence: Linguistics aspects. In
International Encyclopedia of the Social and Behavioral Sciences, ed.
N. J. Smelser and P. B. Baltes , 231923. Amsterdam : Elsevier Science .
Halliday , M . 1975 . Learning How to Mean: Explorations in the Development
of Language. London : Arnold.
Hymes , D . 1967 . Models of the interaction of language and social set-
ting . Journal of Social Issues 23 : 8 28.
. 1992 . Te concept of communicative competence revisited. In
Tirty Years of Linguistics Evolution: Studies in Honor of Ren Dirven ,
ed. M. Ptz , 3158. Philadelphia: Benjamins .
Kasper , G. , and K. Rose . 2002 . Pragmatic Development in Second
Language . Oxford : Blackwell .
Labov , W ., and J. Waletsky . 1967 . Narrative analysis: Oral versions of
personal experience. In Essays on the Verbal and Visual Arts, ed.
J. Helms , 1244. Seattle : University of Washington Press.
Ninio , A ., and C. Snow . 1996 . Pragmatic Development . Boulder,
CO : Westview.
Ninio , A ., and P. Wheeler . 1984 . A manual for classifying verbal com-
municative acts in mother-infant interaction. Working Papers in
Developmental Psychology, no. 1, Hebrew University, Jerusalem.
Pragmatic Competence, Acquisition of Pragmatics
647
entire expression. Second, they assess whether this literal inter-
pretation is appropriate for the specifc context. Tird, if the literal
meaning is contextually inappropriate, as is the case for fgurative
language, listeners must then derive the intended fgurative (e.g.,
metaphorical, ironic) meaning via the cooperative principle. Tis
view suggests, then, that fgurative language should be more dif-
fcult to comprehend than corresponding literal speech, because
fgurative speech requires an additional processing step in which
the literal meanings are rejected and the intended fgurative
meanings are subsequently inferred.
Many pragmatic theories, especially in philosophy, embrace
all or some of the Gricean view of conversational implicature and
his specifc proposals on understanding indirect and fgurative
language. Indeed, much of the focus in philosophical and linguis-
tic studies on pragmatics is devoted to demonstrating how clas-
sic semantic phenomena, such as reference, indexicals , and
demonstratives, can be explained in terms of an understanding of
the specifc facts about the speaker, time, and location of an utter-
ance (Kaplan 1989 ; Stalnaker 1999 ). But psychological experiments
have raised important questions about Grices theory. Although
there is considerable evidence showing that speakers generally
aim to be cooperative, with talk being primarily organized around
the recovery of speakers pragmatic intentions (Clark 1996 ; Gibbs
1999 ), it is less clear that meaning is processed in the serial manner
that Grice and other pragmatists assume. For instance, numerous
psycholinguistic studies indicate that many kinds of fgurative lan-
guage, including novel metaphors, can be understood as quickly
as literal speech when these expressions are encountered in rich
linguistic contexts (Gibbs 1994 ). Tus, pragmatic knowledge may
be immediately accessed and applied in order to understand what
speakers imply by what they say, without listeners frst having to
analyze the literal, semantic meaning of utterances.
A diferent proposal on the pragmatics of utterance inter-
pretation assumes that speakers aim to be optimally relevant
in saying what they do. Optimizing relevance is a fundamental
tenet of relevance theory (Sperber and Wilson 1995 ). Under
this optimally relevant view, every act of ostensive behavior
communicates a presumption of its own optimal relevance,
that is, a presumption that it will be relevant enough to war-
rant the addressees attention and as relevant as compatible
with the communicators own goals and preferences (the com-
municative principle of relevance ). Speakers design their utter-
ances to maximize the number of cognitive efects that listeners
infer, while minimizing the amount of cognitive efort to do so.
Listeners understand speakers communicative intentions via
the relevance-theoretic comprehension procedure (Sperber
and Wilson 2002 ), by following a path of least efort in computing
cognitive efects. Tey do this by testing interpretive hypotheses
(e.g., disambiguations, reference resolutions, implicatures) in
order of accessibility, and then stopping when their expectations
of relevance are satisfed.
For example, consider the following exchange between two
university professors (Sperber and Wilson 2002 , 19):
Peter: Can we trust John to do as we tell him and defend the
interests of the Linguistics Department in the University
Council?
Mary: John is a soldier!
Expressive : Te speaker expresses an attitude toward the
propositional content of an utterance, such as Im sad your
wallet was stolen.
Declarative : Te speaker performs an action just representing
himself or herself as performing that action, such as We fnd
the defendant guilty of murder in the frst degree.
Philosophers have explored the various social and institu-
tional facts that must hold for an utterance to be faithfully seen
as an example of any of these speech-acts, such as whether an
individual must be capable of fulflling the act represented in
an utterance for it to be seen as a sincere promise (i.e., that the
speaker actually has $5 to loan and can give this money to the
listener).
Pragmatic theory, however, has paid greater attention in the
last 40 years to the process by which listeners infer what speakers
mean by what they say. Recall the conversation between Harvey
and Molly. Understanding that Mollys comment is meant as
a particular answer to Harveys question requires that Harvey
go through a chain of reasoning regarding Mollys intentions
because her answer does not logically follow from his ques-
tion. Te philosopher H. Paul Grice called the intended message
behind Mollys utterance a conversational implicature ,
which is a natural outcome of speakers and listeners tacit
adherence to the cooperative principle . Tis states that a
speaker must make your conversational contribution such as
is required, at the stage at which it occurs, by the accepted pur-
pose or direction of the talk exchange in which you are engaged
(Grice, 1975 : 45). Te cooperative principle carries with it four
maxims:
Maxim of Quantity : Make your contribution as informative as
is required, but not more so, for the current purposes of the
exchange.
Maxim of Quality : Do not say anything you believe to be false
or for which you lack adequate evidence.
Maxim of Relation : Say only what is relevant for the current
purposes of the conversation.
Maxim of Manner : Be brief, but avoid ambiguity and obscu-
rity of expression.
Grice noted that speakers do not always uphold these max-
ims. So long as speakers generally adhere to the overall cooper-
ative principle, they can fout any of these maxims to produce
certain implicatures. For example, Mollys response to Harveys
question fouts the maxim of manner to implicate that she is not
going to the party because of Jason. According to Grices analy-
sis, Harvey would not consider Mollys response to be uncoop-
erative. Instead, Harvey would continue to assume that Mollys
rhetorical response was cooperative and would seek an interpre-
tation given what he assumes about Molly, and what he believes
Molly assumes about him, in order to derive an acceptable and
authorized interpretation .
One place where speakers fout conversational maxims is in
their use of fgurative language, such as metaphor (e.g.,Lawyers
are sharks) and irony (e.g., A fne friend you are!). Grices
theory assumes that fgurative language is understood in a series
of steps ( 1975 ). First, listeners analyze the literal meaning of the
Pragmatics
648
(Horn 2004 ; Levinson 2000 ) to psychologists who fault relevance
theory for the circularity in its proposed trade-of between maxi-
mizing cognitive efects and minimizing cognitive efort (Giora
1997 ).
Psychological studies on pragmatics have primarily exam-
ined fgurative language understanding and the degree to
which speakers and listeners coordinate during conversational
exchanges. Neither the Gricean nor relevance theory perspec-
tive assumes that speakers and listeners rely on some defnitive
common ground in order for conversation to proceed smoothly.
Some psychologists, however, have demonstrated through vari-
ous empirical means that speakers and listeners actively col-
laborate and coordinate their beliefs and knowledge to achieve
mutual understandings in diferent contexts (Clark 1996 ; Gibbs
1999 ).
For example, research shows that speakers take the address-
ees perspective into account when designing their utterances
in naturalistic, task-oriented dialogue. One set of studies had
two people, who could not see each other, collaborate over the
arrangement of Tangram fgures (geometric shapes that are
vaguely suggestive of silhouettes of people and other objects)
(Clark and Wilkes-Gibbs 1986 ). One person (the director)
had an ordered array of these fgures and had to explain their
arrangement to the other (the matcher) so that the other per-
son could reproduce the arrangement. Each director-matcher
pair did this six times. Te main hypothesis is that as common
ground is established between the director and matcher during
the conversation, it should be easier for them to mutually deter-
mine where each fgure should go. As expected, the number of
words used per Tangram fgure fell from around 40 in the frst
trial to around 10 in the last. For instance, a speaker referred
to one fgure in Trial 1 by saying All right, the next one looks
like a person whos ice skating, except theyre sticking two arms
out in front, while in Trial 6 the speaker said Te ice skater.
A similar decline was observed in the number of turns required
to complete the arrangement task, showing that the interchange
became more economical as common ground was established.
Other studies using this experimental paradigm indicate that
speakers and listeners can also coordinate to hide information
from overhearers without damaging their own understanding
of each others communicative meanings (Clark and Schaefer
1987 ).
Tese data demonstrated that the assessment of common
ground has an integral part in determining what speakers spe-
cifcally say and in facilitating listeners recovery of speakers
intentions. One implication of these fndings is that utterance
interpretation is a joint activity of both speakers and listeners,
and not solely the responsibility of listeners. Indeed, psycho-
logical studies also demonstrate that conversational participants
typically try to reach the mutual belief that the addressees have
understood what the speaker meant to a criterion suf cient for
current purposes. Tus, when Molly speaks, she looks for evi-
dence from Harvey that he has understood her. Harvey, in turn,
tries to provide that evidence by saying, oh right, nodding his
head, or taking the relevant next turn. Of course, the collabora-
tion and coordination between speakers and listeners refects
the operation of rapid, mostly unconscious, comprehension
processes. Conversational participants are rarely aware of the
How does Peter understand Marys metaphorical assertion
about John? Peters mentally represented concept of a soldier
includes many ideas that may be attributed to John. Among
these are a) John is devoted to his duty, b) John willingly follows
orders, c) John does not question authority, d) John identifes
with the goals of his team, e) John is a patriot, f) John earns a
soldiers pay, and g) John is a member of the military. Each of
these ideas may possibly be activated to some degree by Marys
use of soldier in relation to John. However, certain of these
attributes may be particularly accessible given Peters preced-
ing question where he alludes to trust, doing as one is told, and
defending interests. Following the relevance-theoretic com-
prehension procedure, Peter considers these implications in
order of accessibility, arrives at an interpretation that satisfes
his expectations of relevance at d, and stops there. He does not
even consider further possible implications, such as eg, let
alone evaluate and reject them. In particular, Peter does not
consider g, the literal interpretation of Marys utterance, con-
trary to what is advanced by the Gricean view, and consistent
with the psychological evidence on inferring metaphorical
meaning.
Relevance theory has also advanced the idea that signif-
cant aspects of what speakers say, and not just what they totally
communicate, are deeply dependent upon enriched pragmatic
knowledge. Essentially, the same sorts of inferential processes
used to determine conversational implicatures also enter into
determining what speakers say (Carston 2002 ; Recanati 2004 ;
Sperber and Wilson 1995 ). Consider a case where a speaker says
to you I havent eaten in response to a question about whether
she found time for breakfast that morning. Once the indexical ref-
erences and the time of the utterance are fxed, the literal mean-
ing of the sentence determines a defnite proposition, with a
defnite truth condition, which can be expressed as Te speaker
has not eaten prior to the time of the utterance. Tis paraphrase
refects the minimal proposition expressed by I havent eaten.
However, a speaker of I havent eaten is likely to be communi-
cating not a minimal proposition but some pragmatic expansion
of it, such as I havent eaten today. Tis possibility suggests
that signifcant pragmatic knowledge plays a role in enabling
listeners to expand upon the minimal proposition expressed in
order to recover an enriched pragmatic understanding of what
a speaker says. Several experimental studies indicate that prag-
matics plays a major role in peoples intuitions of what speak-
ers say (Gibbs and Moise 1997 ). Tus, the distinction between
what speakers say and imply may possibly be orthogonal to any
distinction between semantics and pragmatics, contrary to the
traditional Gricean view .
Te vast number of studies on a wide assortment of linguistic
and nonlinguistic phenomena conducted within the relevance
theory framework makes it the most salient model of pragmat-
ics and utterance interpretation available today. At the very least,
part of relevance theorys signifcant appeal in interdisciplinary
language studies is its explicit aim to situate pragmatics within
broader concerns of human cognition and communication,
through its embrace of the principles of relevance. Not surpris-
ingly, relevance theory has its critics, ranging from scholars,
primarily in linguistics, who assume that utterance meaning is
determined by heuristics of default or preferred interpretations
Pragmatics
649
Levinson , Stephen . 2000 . Presumptive Meanings . Cambridge, MA : MIT
Press .
Recanati , Francois . 2004 . Literal Meaning . New York : Cambridge
University Press .
Searle , John . 1975 . A taxonomy of illocutionary acts. In Language, Mind,
and Knowledge, ed. Keith Gunderson , 34469. Minneapolis: University
of Minnesota Press .
Sperber , Dan , and Deirdre Wilson . 1995 . Relevance: Cognition and
Communication. 2d ed. Oxford : Blackwell.
. 2002 . Pragmatics, modularity, and mind-reading. Mind and
Language 17: 323.
Stalnaker , Robert . 1999 . Context and Content . Oxford : Oxford University
Press .
PRAGMATICS, EVOLUTION AND
For at least 100,000 years, human beings have been talking the
way we do. Language is universally used by most individuals in
every culture several hours each day, primarily during conver-
sational chatter (Dunbar 1998 ). How did our species come to
adopt such a strange behavior in the course of its evolution? Te
question has been considered in turn as obvious and baf ing.
A proper approach to the reasons why we talk requires that the
biological function of language be understood, and pragmatics is
the right place to seek out that function.
If we adopt the perspective of an ethologist, then language
appears as a distinctive feature of our species that, like any other
fnely designed characteristic, must have a defnite function to
have been selected through the repeated efect of diferential
reproduction (Pinker and Bloom 1990 ). For decades, language
was thought to be essentially a means for organizing, responding
to, and manipulating the behavior of others (Brown 1991 , 130) or
a tool for sharing knowledge (Pinker 1994 , 367), and it was con-
sidered obvious that it had been selected for these purposes. Tis
traditional view has now lost most of its obviousness for two rea-
sons: 1) Its logic contradicts Darwinian principles, and 2) what
people spontaneously do with language corresponds to a quite
diferent picture. We examine these two issues in turn, before
considering more plausible alternatives.
Any evolutionary account of the existence of language must
make clear what biological advantage both speakers and listen-
ers get out of speaking. In many traditional accounts, the fact that
listeners take advantage from receiving information is taken as
suf cient explanation for the existence of language, but language
cannot evolve if there is no direct or indirect advantage on the
speakers side. If language is a way of infuencing others behav-
ior, the speakers advantage is now obvious, but Darwinian selec-
tion should have led to resistance on the listeners side: Tere is
an advantage in ignoring signals aiming at bringing you to serve
the interest of others.
One of the most striking and incomprehensible facts about
human language is that it relies on a positive attitude from speak-
ers. Speakers bear all the burden of designing appropriate (Grice
1975 ) or even optimal (Sperber and Wilson 1986 ) messages to
convey intentional meaning. If they do so spontaneously and
often quite profusely, it must be because they gain some beneft
from it. Listeners, on the other hand, show much trust in what
they hear. Knowing that language is cheap, the fact that listen-
ers give credence to most of what they hear is hard to explain
cognitive and linguistic processes that underlie their under-
standing of others pragmatic intentions, unless the attempt to
coordinate fails and leads to misunderstandings .
Not all psychologists agree that speakers and listeners always
aim to be cooperative in conversation by taking the other per-
sons perspective into account during speaking and listening.
Some experiments show, for example, that speakers and listen-
ers can each adopt an egocentric bias as they speak and com-
prehend, particularly when they experience additional cognitive
load or stress (Horton and Keysar 1996 ). Speakers also some-
times overestimate how efective they are communicating their
messages to listeners, with listeners also sometimes assuming
that they correctly understood speakers when in fact they did not
(Keysar and Henly 2002 ). Tese studies show how there are at
least some systematic sources of misunderstanding attributable
to what might be best characterized as an egocentric bias in com-
munication efectiveness.
Te study of pragmatics will undoubtedly continue to have a
strong interdisciplinary favor in the future. Scholarly intuitions
about how knowledge of the world and context shape utterance
interpretation must be supplemented by experimental stud-
ies that examine fast, unconscious cognitive and linguistic pro-
cesses operating when speaker meaning is understood. We need
to understand not only what pragmatic information shapes con-
textual meaning but also when that knowledge in recruited in the
psychology of ordinary language interpretation.
Raymond W. Gibbs , Jr. and Gregory A. Bryant
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Austin , John . 1962 . How to Do Tings with Word s . Oxford : Clarendon.
Carston , Robyn . 2002 . Toughts and Utterances: Te Pragmatics of Explicit
Communication . Oxford : Blackwell .
Clark , Herbert . 1996 . Using Language . New York : Cambridge University
Press .
Clark , Herbert , and Edward Schaefer . 1987 . Concealing meaning from
overhearers . Journal of Memory and Language 26 : 20925.
Clark , Herbert , and Deanna Wilkes-Gibbs . 1986 . Referring as a collab-
orative act . Cognition 22 : 139.
Giora , Rachel . 1997 . Discourse coherence and theory of rele-
vance: Stumbling blocks in search of a unifed theory. Journal of
Pragmatics 27 : 1734.
Gibbs , Raymond . 1994 . Te Poetics of Mind: Figurative Tought, Language,
and Understanding . New York : Cambridge University Press .
. 1999 . Intentions in the Experience of Meaning . New York : Cambridge
University Press.
Gibbs , Raymond , and Jessica Moise . 1997. Pragmatics in understanding
what is said . Cognition 62 : 5174.
Grice , H. Paul . 1975 . Logic and conversation. In Syntax and Semantics .
Vol 3: Speech Acts . Ed. Peter Cole and Jerald Morgan , 4158. New
York : Academic Press.
Horn , Larry . 2004 . Implicature. In Te Handbook of Pragmatics , ed.
Larry Horn and Gregory Ward , 328. Oxford : Blackwell.
Horton , William , and Boaz Keysar . 1996 . When do speakers take into
account common ground? Cognition 59 : 91117.
Kaplan , David . 1989 . Demonstratives. In Temes from Kaplan , ed.
Joseph Almog , John Perry , and Howard Weinstein , 481563. New
York : Oxford University Press.
Keysar , Boaz , and Anne Henly . 2002 . Speakers overestimation of their
efectiveness. Psychological Science 13 : 20712.
Pragmatics Pragmatics, Evolution and
650
Brianne: we were like oh wow. It was really weird. But it was
her. (laughs)
Addie: Oh no. Weird.
Brianne: I mean, thats weird when you run into somebody in
Chicago.
Addie: yeah.
Stories come in chunks, the so-called story rounds (Tannen
1984 , 100), which may last for tens of minutes. Te biological
signifcance of this systematic and universal tendency to report
emotional and unexpected events lies quite far away from any
immediate utilitarian efect, like behavioral infuence or vital
knowledge transfer .
During argumentation, in contrast with narration, individu-
als are not bound to mention fully instantiated states of afairs.
Tey may even utter quite general statements to make a point.
Argumentation can be described, at the cognitive level, as an
oscillation between problems and tentative solutions (Dessalles
2007). During conversation, any inconsistency between beliefs
or between beliefs and desires is likely to be signaled, and it trig-
gers a collective search for solutions. In the following example
(adapted from Tannen 1984 , 62), two participants wonder
how the third one came to know about the sociologist Erving
Gofman .
Deborah: But anyway. How do you happen to know his
stuf?
Chad: Cause I read it.
Peter: What do you do?
Deborah: Are you in sociology or anything?
Chad: No.
Deborah: You just heard about it, huh?
Chad: Yeah. No. I heard about it from a friend who was a
sociologist, and he said read this book, its a good book and I
read that book n
Deborah: I had never heard about him before I started study-
ing linguistics.
Chad: Really?
Te argumentative process is the same, with its characteristic
alternation between problems and solutions, regardless of the
social situation in which it occurs: a discussion about a famous
sociologists work, the planning of some forthcoming travel, or
a harsh dispute. Te biological signifcance of this systematic
and universal propensity to mention inconsistencies and then
to make every attempt to solve them cannot be reduced to the
pursuit of some immediate practical beneft. Quite often, casual
discussions are about futile matters that are unlikely to change
the interlocutors fate .
Why do human beings devote most of their speaking time
telling stories and dealing with apparent inconsistencies? What
utilitarian models fall short of explaining is directly addressed
by models, like the grooming hypothesis, that emphasize the
role of language in the establishment of social bonds (Dunbar
1996 ; Dessalles 2007). Language acts would not be biologically
motivated by their immediate beneft but because they are
in a Darwinian world in which creatures are designed to favor
their own success, not the success of others (Knight 2002 ). Te
absence of trust is what explains the repetitiveness, the cost, and
the poverty of most animal communication (Zahavi and Zahavi
1997 ).
Tese concerns about the speakers willingness to speak and
the listeners to trust have no known solution within frameworks
in which language acts are supposed to provide immediate
beneft to either party. It has been suggested that information
exchange through language could be based on reciprocity (Pinker
2003 , 28; Nowak and Sigmund 2005 , 1293). Te reciprocation
model, however, functions under strict limits: good beneft-to-
cost ratio and strict control of reciprocity. It is at odds with sev-
eral observations about spontaneous language, such as the fact
that many conversational utterances are about futile topics, or
the fact that talkative behavior is far from being an exception: On
average, individuals typically talk to two persons simultaneously
(Dunbar, Duncan, and Nettle 1995).
Te utilitarian conceptions of language that inspired most
traditional ideas about its biological role are dictated mainly by
theoretical considerations. Some theories emphasize the role
of language in performing actions; it is thus natural to imagine
language as having emerged from simple directives (Holdcroft
2004 ). Other theories see in language a process through which
individuals actively try to infuence the beliefs of others (Sperber
and Origgi 2010). A natural strategy, to decide which aspect of
language use is most likely to have given a biological advantage
both to speakers and listeners, is to observe how current human
beings spontaneously talk.
Conversation constitutes by far and universally the main
occasion in which language is used. Conversational activity,
however, is not monolithic. When chatting, individuals show
essentially two forms of behavior: Tey tell stories and they
pursue argumentative discussions. Even if both are often inter-
twined, it is important to distinguish narration and argumen-
tation, as they involve quite diferent cognitive processes and
might have arisen successively during evolution. Conversational
narrative analysis shows that narratives fll up to one-half of
our speaking time (Eggins and Slade 1997, 265) and may rep-
resent some 10 percent of our awake time. Speakers take time,
sometimes several minutes, to recount some past situation in
minute detail (Norrick 2000 ). Not all situations are likely to be
reported: Only those that can elicit specifc emotions , especially
surprise , are recounted (Dessalles 2007). Te following exam-
ple, adapted from (Norrick 2000 , 556), is about an unexpected
encounter:
Brianne: It was just about two weeks ago. And then we did
some fgure drawing. Everyone was kind of like, oh my God,
we cant believe it. We- yknow, Midwest College, yknow,
Brianne: like a nude models and stuf. And it was really
weird, because then, like, just last week, we went downtown
one night to see a movie, and we were sitting in [a restaurant],
like downtown, waiting for our movie, and we saw her in the
[restaurant], and it was like, thats our model (laughing) in
clothes
Addie: (laughs) Oh my God.
Pragmatics, Evolution and
651
Johansson , Sverker . 2005 . Origins of Language Constraints on
Hypotheses . Amsterdam : John Benjamins.
Knight , Chris . 2002 . Language and revolutionary consciousness. In Te
Transition to Language , ed. A. Wray , 13860. Oxford : Oxford University
Press .
Norrick , Neal R . 2000 . Conversational Narrative: Storytelling in Everyday
Talk . Amsterdam : John Benjamins .
Nowak , Martin A ., and Karl Sigmund . 2005 . Evolution of indirect reci-
procity . Nature 437. 27: 1291 8.
Pinker , Steven . 1994 . Te Language Instinct . New York : Harper
Perennial .
. 2003 . Language as an adaptation to the cognitive niche. In
Language Evolution , ed. M. H. Christiansen and S. Kirby , 1637.
Oxford : Oxford University Press .
Pinker , Steven , and Paul Bloom . 1990 . Natural language and natural
selection. Behavioral and Brain Sciences 13. 4: 707 84.
Sperber , Dan , and Gloria Origgi . 2010. A pragmatic perspective on the evo-
lution of language. In Te Evolution of Human Language: Biolinguistic
Perspectives , ed. Richard Larson, Viviane Dprez, and Hiroko Yamakido,
12432. Cambridge: Cambridge University Press.
Sperber , Dan , and Deirdre Wilson . 1986 . Relevance: Communication and
Cognition . Oxford : Blackwell .
Tannen , Deborah . 1984 . Conversational Style Analyzing Talk Among
Friends . Norwood, NJ : Ablex .
Zahavi , Amotz , and Avishag Zahavi . 1997 . Te Handicap Principle . New
York : Oxford University Press .
PRAGMATICS, NEUROSCIENCE OF
What Is Pragmatics and What Is the Neuroscience of
Pragmatics?
Te neuroscience of pragmatics has not been extensively stud-
ied. Tis is understandable, inasmuch as pragmatics came
into focus fairly late as an area of study in linguistics. Te feld
is not dominated by one specifc theoretical framework. Tere
are controversies concerning its delimitation with respect to,
for example, semantics and nonlinguistic behavior. Finally,
since pragmatic phenomena are crucially related to connected
discourse and communicative interaction, they do not lend
themselves easily to investigation by established experimental
approaches or to such methods as neuroimaging, electroen-
cephalography (EEG), and so on that focus on limited, often
decontextualized, linguistic units. Tis entry is an attempt to
summarize some typical approaches in relating pragmatic
phenomena to neural processing, and what they have found to
date.
Te term pragmatics is used here in accordance with C. W.
Morris ( 1938 ), who posited a framework in which syntax deals
with the formal relations among signs, semantics adds the rela-
tions of signs to objects, and pragmatics further adds the rela-
tions of signs to the interpreter. Pragmatics is about language use
or communication, in a broad sense and in context. It is assumed
here that there is no clear sense in which pragmatics can be sep-
arated from semantics. Te focus here, however, is on phenom-
ena that are typically considered to be part of pragmatics .
In terms of the relation of pragmatics to neuroscience , difer-
ent types of approaches have to be considered, since this is far
from a uniform feld of study. One set of approaches constitutes
experimental studies of phenomena that are considered to be
reliable indicators of some speaker quality that is valued in the
establishment of solidarity networks. In these models, language
is display . In the political niche of our species, individuals who
are aware of their physical and social environment make better
coalition partners. Hence, individuals demonstrate that they are
able to witness unusual situations by reporting facts that elicit
surprise and emotion. By recounting the weird encounter with
the nude model, Brianne obeys this urge to show her ability to
surprise others .
From this perspective, language is a competition for inter-
est . On the friendship marketplace, where solidarity bonds are
established and dissolved, individuals who report the most inter-
esting events are, all other things being equal, the most appreci-
ated. Now, the biological role of argumentation becomes clear.
Without the ability to detect inconsistencies, individuals would
easily shine by reporting incredible events that never occurred.
Argumentation presumably emerged as an anti-liar device,
besides checking for oneself (Dessalles 1998 ). As it is preferable
to have nongullible members in ones coalition, argumentation
became a way to demonstrate this quality. Hence, Deborahs and
Peters refexes show that they could spot an apparent inconsis-
tency during their conversation with Chad.
Recently, there have been various attempts to account for the
existence of language (Johansson 2005 ). Te one emphasized
here highlights the political importance of talking. Language per-
formance is indirectly vital: Tose who recount in boring fashion
or who are unable to build sensible arguments are rapidly left
aside. In the world of our hominine ancestors, lonely individuals
were defenseless and likely to be exploited. Language emerged
as a way for human beings to show to their conspecifcs that they
have the required qualities to be valuable friends .
Jean-Louis Dessalles
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brown , Donald E . 1991 . Human Universals . Philadelphia : Temple
University Press .
Dessalles , Jean-Louis . 1998 . Altruism, status, and the origin of relevance.
In Approaches to the Evolution of Language: Social and Cognitive
Bases , ed. J . R. Hurford , M. Studdert-Kennedy , and C. Knight , 13047.
Cambridge : Cambridge University Press .
. 2007 . Why We Talk Te Evolutionary Origins of Language . Trans.
James Grieve. Oxford : Oxford University Press .
Dunbar , Robin I. M . 1996 . Grooming, Gossip, and the Evolution of
Language . Cambridge : Harvard University Press .
. 1998 . Teory of mind and the evolution of language. In
Approaches to the Evolution of Language: Social and Cognitive Bases ,
ed. J. R. Hurford , M. Studdert-Kennedy , and C. Knight , 92110.
Cambridge : Cambridge University Press .
Dunbar , Robin I. M. , N. D. C. Duncan , and Daniel Nettle . 1995 . Size and
structure of freely forming conversational groups. Human Nature
6. 1: 67 78.
Eggins , Suzanne , and Diana Slade . 1997 . Analysing Casual Conversation .
London : Equinox .
Grice , H. Paul . 1975 . Logic and conversation. In Syntax and Semantics .
Vol. 3: Speech Acts . Ed. P. Cole and J. L. Morgan , 4158. New
York : Academic Press .
Holdcroft , David . 2004 . Pragmatics and evolution. Pragmatics and
Beyond 127 . 117 27.
Pragmatics, Evolution and Pragmatics, Neuroscience of
652
the acceptance of pragmatics as a discipline within linguistics,
anthropology, sociology, and communication sciences and
its introduction into clinical linguistics and medical settings
during the last 20 to 30 years. Some approaches have involved
the extension of the classical model of aphasia syndromes
and the application of cognitive neuropsychology models. But
there is also a recognition that other theoretical frameworks
must be applied in the study of pragmatics: the increasing use
of connectionist modeling, the growing community apply-
ing pragmatic theories and methods of analysis to studies of
communication involving persons with brain damage, and the
increasing interest in embodied cognition and communica-
tion. Te rapid development of neuroimaging techniques, also
during the last 20 to 30 years, has coincided with the develop-
ment of pragmatic approaches to neurolinguistics; until very
recently, however, the two research streams have not joined
forces. Neuroimaging studies focus on phenomena that are
easily studied in an experimental context. But the rapid devel-
opment of fMRI, PET (positron emission tomography), and
MEG (magnetoencephalography) techniques, as well as EEG/
ERP, has paved the way for recent and ongoing attempts to
actually capture pragmatic phenomena as well in this type of
research .
Current State of the Field
Following are descriptions and examples of some of the domi-
nant types of studies in this area.
MEASUREMENT OF BRAIN ACTIVITY. Some of the recurring abili-
ties or functions attributed to brain areas are
inhibition
selection and ordering of speech, behavior, and logic
formation and execution of plans of action
memory processes (working memory, episodic memory
retrieval, emotional modulation of memory processes and
executive processing, and cues for long-term memory)
theory of mind (ToM), mental inferencing, attribution of
mental states, simulation for comprehension, visuospatial
imagery, abstraction
TYPICAL STUDIES I EEG/ERP. Studies of brain activity using
EEG/ERP involve the correlation of specifc temporal compo-
nents of brain activity with performance. Conversation is not
easily studied with this type of method. One of the most fre-
quently used components is the N400, a negative ERP response
to semantic anomaly.
One of the fndings from studies of the N400 is that there
is rapid incremental processing all the way through; in other
words, listeners start early to respond to unfolding words as
infuenced by topic, how the speech is produced, and by whom.
Listeners also use discourse information to automatically make
predictions about semantics, syntax, phonology, and referents;
discourse information can overrule local constraints. No evi-
dence of context-free sentence-internal interpretation has been
found, and the conclusion is that the brain does not engage in
a two-step interpretation of semantics and pragmatics. What
seems to be happening is that an initial quick and superfcial
important in pragmatics. Tis may involve trying to isolate such
phenomena in an experimental setting that allows for neuroim-
aging techniques, such as fMRI (functional magnetic resonance
imaging), or EEG/ERP (event-related potentials) to be used
to measure brain activity. Lesion data can be handled either
by experimental group studies, comparing, for example, indi-
viduals with left hemisphere damage (LHD), right hemisphere
damage (RHD) and no brain lesion, or by case studies com-
paring specifc phenomena in one individual or a number of
individuals with lesions and communication disorders. Lesion
studies in the area of pragmatics often also include a focus on
communicative interaction in context, involving persons with
brain damage, and use such methods as videorecording of face-
to-face interaction, transcription, and coding and microanaly-
sis of sequences and patterns of interaction. Tese studies also
include more social constructivist empirical approaches, such
as conversation analysis . Lesion studies can also be done with
brain activity measurements, but this is more dif cult, espe-
cially when homogeneous groups and many repeated events
are needed and when an interaction between two persons is
being studied.
Te feld of pragmatics primarily needs models of connected
speech, for example, of story structure and topic fow, and mod-
els of linguistic communicative interaction, that is, models
involving two (or more) participants and the interactive fow,
co-construction, activation, and so on between them, including
diferent levels of conscious control. Tese models can vary con-
siderably in degree of detail and specifcity. Typically in studies
of pragmatics, communicative, cognitive, and emotive factors
are included. Multimodality, such as body communication and
prosody in speech, is considered important. Data often consist
of sequences longer than words and sentences monologue and
dialogue samples, for example. Overall structure and the course
of communication are studied, and interactive phenomena are
often in focus. Specifc pragmatic phenomena that can be stud-
ied from a neuroscience perspective are listed in Table 1 .
Main Topical Subdivisions of the Field
Table 1 illustrates typical actual (x) and potential combinations
of phenomena in pragmatics with methods in neurocognitive
studies. Some of the combinations have also been attempted
but, in general, there is a dividing line between phenomena
that can be studied both in monologue and dialogue conditions
and those that can only be fruitfully studied in dialogue. Studies
involving the measurement of brain activity have generally been
limited to monologue situations. It is, of course, possible to
combine more than one method and more than one pragmatic
phenomenon in a given study. Te neuroscience of pragmatics
faces the challenge of unifying the fairly rich fndings from natu-
ralistic and experimental behavioral studies of monologue and
dialogue/interaction with studies of brain activity, which so far
have been related to monological and experimental tasks only.
Tis requires i) extensive work on models and theories, and ii)
continued development of techniques and methodologies .
A Brief History of Modern Developments in the Field
A number of important milestones can be mentioned in the
development of the neuroscience of pragmatics. Te frst is
Pragmatics, Neuroscience of
653
comparison to single words and sentences, have most frequently
been used to measure brain activity in studies of pragmatic phe-
nomena. In general, many-to-many mappings of structures and
functions are found, and this points to the need to develop theo-
ries and models .
TYPICAL STUDIES III BEHAVIORAL STUDIES OF RIGHT HEMI-
SPHERE DAMAGE. Studies of the behavior of LHD, RHD, and
control subjects with no brain damage are perhaps the most
prototypical ones in behavioral studies both experimental ones
and studies of naturalistic conversation. A number of such stud-
ies over the last 20 or 30 years have shown that RHD subjects,
in spite of their good performance on traditional aphasia tests,
defnitely perform worse on many diferent aspects of pragmat-
ics than LHD or control subjects. Tese fndings have placed
RHD at the center of the neuroscience of pragmatics. Studies of
this type have used experimental group designs, as well as case
studies and microanalysis of communicative interaction. Tey
have given us a picture of RH functions in lexical semantics,
the semantics of connected speech and writing, prosody, body
communication, holistic processing, spatial imagery, ToM,
topic management, sensitivity to interactive cues, inferencing
(especially about emotions), and a number of other pragmatic
abilities.
Some of the limitations to this approach are the (so far) rela-
tively broad and uncertain mapping of specifc areas in the RH
to specifc functions, the fact that groups of RH subjects are
not homogeneous, and the relative lack of good instruments
for measuring pragmatic functions in an experimental context.
interpretation is followed or partially overlapped by a more pre-
cise one .
TYPICAL STUDIES II NEUROIMAGING fMRI. Using fMRI
involves the comparison of brain activity in one task with
activity in another via subtraction. Naturalistic interaction
is not easily studied with fMRI. Labels used for the behavior
studied are often too broad for specific comparisons between
studies to be useful, and the method only shows activation
across trials and participants, and not individual, diffuse, and
weak signals that could also be important for modeling prag-
matic processing.
One relevant fMRI study had (non-brain-damaged) subjects
listen to connected versus nonconnected sentences. In this
experiment by D. Robertson and colleagues (2000), no diference
in brain activity between the two conditions was found for the left
hemisphere (LH), whereas the right hemisphere (RH) showed
increased activation for connected discourse in the middle and
superior frontal regions. Te same fndings were replicated for
picture stories. In passive listening only, however, it disappears.
In addition, fMRI studies have shown that increased dif culty
leads to more difuse activation of brain regions. Activation of
the temporal poles was also found only for this task. One sug-
gested interpretation is that RH frontal lobe activation may only
show up when the subject is activating memory to create coher-
ence in a story representation. Te same areas have been linked
to abilities such as ToM, episodic memory, and integration.
As we have seen, studies of the comprehension and pro-
duction of connected units of language, such as narratives, in
Table 1. Possible and typical (x) combinations of pragmatic objects of study and neuroscientifc methods
Pragmatic phenomena/methods Lesion-based studies studies of behavior Brain activity studies
Empirical studies of
behavior in context
Experimental studies Neuroimaging, esp. fMRI
EEG/ERP
Comprehension and production of longer
spoken contributions and texts, such as
narratives
x x x x
Inference x x x x
Cognitive semantics x x x x
Emotion in communication x x
Own communication management
(e.g., hesitation, self-repair)
x x
Interactive communication management
(e.g., turn-taking, feedback)
x x
Speech-acts, language games x x
Conversational principles x x
Flexibility and adaptation;
alignment, coupling, holistic patterns
x x
Body communication,embodiment,
multimodality
x x
Pragmatics, Neuroscience of
654
(Hymes 1982 ). But with the growth of linguistic typology and
the empirical search for language universals, it has become
increasingly clear that real universals in the straightforward
sense, properties that all languages have are vanishingly rare
(at least beyond the basic organizational principles outlined by
Hockett 1960 , and some of the architectural properties sketched
by Jackendof 2002 ). Instead, linguistic typologists have found
that empirical generalizations are nearly always of the kind
Across all languages, if a language has property X, then it prob-
ably also has property Y. Meanwhile, generative grammarians
have hoped to account for the diversity in terms of a limited set
of variants (see principles and parameters theory ), but
such variants are not manifested in grammars in any straight-
forward way, and the whole attempt does not appear successful
to many dispassionate observers (Newmeyer 2004 ). Te reality
is that there is an extraordinary diversity of linguistic types, in
which both shared patterns and diferences seem best under-
stood historically and geographically (see, e.g., Haspelmath et al.
2005 ).
With the waning of hopes for straightforward grammatical
universals, the case for pragmatic universals looks, in contrast,
stronger and stronger. Te distinct possibility now arises that
while grammatical patterns are in large part a matter of histori-
cal and cultural evolution, principles of language usage con-
stitute the foundational infrastructure for language, to which
commonalities across languages can be partially attributed.
Tis inverts the traditional view (as in Hymes 1982 ) that gram-
mar is universal and language usage variable. If this inverted
picture is even partially correct, then we would expect sig-
nifcant absolute (unconditional) universals across the sub-
domains of pragmatics (see absolute and statistical
universals ). Te following sections lay out the case for prag-
matic universals .
Deixis
Te fundamental use of language is in face-to-face conversa-
tion, where participants take turns at speaking. Aspects of this
context are built into languages in many detailed ways. All spo-
ken languages have a grammatical category of person , that is,
a grammatical refection of the diferent roles that participants
(and nonparticipants) have in an utterance (speaker, addressee,
third party), which is likely to be refected in personal pro-
nouns, verbal infections, imperatives, vocatives (as in address
forms), and so forth. Likewise, all languages have at least one
demonstrative, a special form for indicating entities in the con-
text typically, there are contrastive forms (like this and that )
associated with pointing. Tey also have ways to distinguish the
time and place of speaking (they may not have tense , but they
will have forms denoting now, today, here, etc.). Tese
aspects of language structure are pragmatic in the sense that
they refer to aspects of the context of utterance, and their inter-
pretation is relative to that context. Te peculiarity of these sys-
tems is that as speakers alternate, the reference of these terms
also alternates (my I is your you , and my this may be your that ),
a fact that children can fnd dif cult when learning a language.
Since artifcial languages (logics, programming languages) suc-
cessfully purge their structures of such items, it is clear that
It should also be stressed that when fne-grained methods are
used, pragmatic defcits stemming from LH aphasia, traumatic
brain injury, and other brain damage conditions are also found.
Concerning the analysis of face-to-face interaction, the general-
izability of results tends to be fairly low. Still, these types of analy-
sis are extremely important to the neuroscience of pragmatics, as
they provide studies of important pragmatic phenomena, which
can also serve as input for further development of theories and
methods. Most of the theoretical claims made on the basis of
brain activity studies today in the area of pragmatics were already
made much earlier on the basis of empirical studies of behavior
following brain damage .
Elisabeth Ahlsn
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brownell , H ., and O. Friedman . 2001 . Discourse ability in patients with
unilateral left and right hemisphere brain damage. In Handbook
of Neuropsychology . 2d ed. Vol. 3. Ed. R. S. Berndt, 189203.
Amsterdam : Elsevier .
Gernsbacher , M. A ., and M. P. Kashak . 2003 . Neuroimaging stud-
ies of language production and comprehension. Annual Review of
Psychology 54 : 91 114.
Mar , R. A . 2004 . Te neuropsychology of narrative: Story comprehen-
sion, story production and their interrelation . Neuropsychologia 42:
1414 34.
Morris , C. W . 1938 . Foundations of the theory of signs. Chicago: Chicago
University Press.
Robertson , D.A. , M. Gernsbacher , S. Guidotti , R. Robertson , W. Irwin ,
B. Mock , and M. Campana . 2000 . Functional neuroanatomy of the
cognitive process of mapping during discourse comprehension .
Psychological Science 11 .3: 255 60.
van Berkum , J. J. A . 2005 . Te electrophysiology of discourse and con-
versation. In Te Cambridge Handbook of Psycholinguistics , ed. M.
Spivey , M. Joanisse , and K. McRae . Cambridge : Cambridge University
Press .
PRAGMATICS, UNIVERSALS IN
Changing Prospects for Universals in Pragmatics
The term pragmatics has come to denote the study of
general principles of language use. It is usually understood
to contrast with semantics , the study of encoded mean-
ing, and also, by some authors, to contrast with sociolin-
guistics and the ethnography of speaking, which are more
concerned with local sociocultural practices. Given that prag-
maticists come from disciplines as varied as philosophy, soci-
ology, linguistics, communication studies, psychology, and
anthropology, it is not surprising that definitions of pragmat-
ics vary. Nevertheless, most authors agree on a list of topics
that come under the rubric, including deixis, presupposi-
tion , implicature (see conversational implicature),
speech-acts , and conversational organization (see con-
versational analysis ). Here, we can use this extensional
definition as a starting point (Levinson 1988; Huang 2007).
With the rise of generative grammar , and the insis-
tence on universals of grammar (see universal grammar ),
anthropologists began to emphasize the diversity of language
use, implicitly accepting the underlying uniformity of grammar
Pragmatics, Universals in
655
Te maxims of quantity and manner, in particular, seem to
be responsible for detailed cross-linguistic patterns of inference
(Horn 1984 ; Levinson 2000 ). For example, the cofee is warm
suggests that it is not hot, or Ibn Saud had 22 wives suggests
that he did not have 23 even though if cofee is hot it is cer-
tainly warm, and if you have 23 wives you certainly have 22. Te
reasoning seems to be that if you know the stronger quantity
holds, you should have said so not saying so implicates that it
does not hold. In a similar cross-linguistically general way, Its
not impossible that the war will still be won implicates greater
pessimism that the war will be won than the logically equiva-
lent Its possible the war will still be won. Te reasoning seems
to be that since the speaker has avoided the positive by using
a double negative, by the maxim of manner he must have had
some reason to do so. Tese cross-linguistic patterns seem to
have systematic efects on grammar and lexicon (Levinson 2000 ;
Sperber and Wilson 1995 ).
Speech-Acts
Te speech acts of questioning, requesting, and stating are found
in conversation in any language, and they have grammatical
repercussions in all language systems for example, in inter-
rogative, imperative, and declarative syntax (Sadock and Zwicky
1985 ). Languages difer, of course, in how, and the extent to
which, these acts are grammatically coded, but they always are
at least partially refected in grammar. John Searle ( 1976 ) sug-
gested that there are fve major kinds of speech-acts: directives (a
class including questions and requests), representatives (includ-
ing statements), commissives (promising, threatening, ofering),
expressives (thanking, apologizing, congratulating, etc.), and
declarations (declaring war, christening, fring, excommunicat-
ing, etc.). Te types are individuated by diferent preconditions
and intended efects, known as their felicity conditions .
Te broad taxonomy ofers plausible universal classes, while
subsuming culture-specifc actions like declarations, such as
divorce by announcement in Moslem societies or magical spells
in a Melanesian society .
Despite the fact that there is an association between, for
example, interrogative form and questioning, the link between
form and action performed is often complex. In English, for
example, requests are rarely done in the imperative, but typi-
cally in the interrogative, as in Can you help me get this suitcase
down? It has been noticed that if a distinctive felicity condi-
tion for a successful request is stated or requested, this will itself
serve as a request (the addressee being able to get the suitcase
down being a precondition to a felicitous request). Tis seems
to have general cross-linguistic application, suggesting that the
action performed is in fact implicated by what is said (Brown
and Levinson 1987 , 136 f). However, in many cases, less regu-
lar strategies link what is said to the actions performed, and the
mapping from utterances to actions remains a serious theoreti-
cal problem in pragmatics .
Conversation Structure
Te organization of conversation seems likely to provide some
of the most robust pragmatic universals. As far as we know,
in all societies the most informal type of talk involves rapid
natural languages could be diferent and, thus, that deictic orga-
nization constitutes a nontrivial universal aspect of language
built for interactive use .
Presupposition
Languages have various ways to foreground and background
information, and this is crucial if the speakers current point
is to be identifed. Information that is presumed in the context
(either because it has already been mentioned or is taken for
granted) is typically not asserted but presupposed, and this is
refected in language structure. Te contrast between defnite
and indefnite articles, in those languages that have them, is a
simple example: Both Te ninth planet has a peculiar orbit and
Te ninth planet does not have a peculiar orbit presuppose that
there is a ninth planet. Tis constancy under negation is often
taken to be a defning property of presupposition it shows
that the presupposed content is not what is being asserted.
Note that unlike what is asserted, presuppositions are defeasi-
ble (fall away) in certain contexts, as in If there is one, the ninth
planet must have a peculiar orbit . Many structures have been
identifed that signal this presuppositional property: factive
verbs like regret in he regrets publishing it (which presupposes
he did publish it), cleft-sentences like It was the police who hid
the crime (which presupposes that someone hid the crime), or
comparatives like Hes a better golfer than Tiger (which pre-
supposes that Tiger is a golfer). Although this might seem
to be purely a matter of the arbitrary conventions of a single
language, in fact structures with similar semantics also tend
to carry similar presuppositions in other unrelated languages
(Levinson and Annamalai 1992 ), suggesting that it is proper-
ties of the semantic representation that trigger the presuppo-
sitional inferences. It is thus possible to make an inventory of
types of structure that tend to universally signal presupposi-
tional content .
Implicature
A conversational implicature is an inference that comes about
by virtue of background assumptions about language use,
interacting closely with the form of what has been said. H.
Paul Grice ( 1975 , 1989 ) outlined a cooperative principle
instantiated in four such background maxims of use: Speak
the truth ( quality ), provide enough but not too much infor-
mation ( quantity ), be relevant ( relevance ), and be perspicu-
ous ( manner ). For example, if A says Have you seen Henk?
and B says His office door is open, we read Bs utterance
as a partial answer (by relevance), which B chooses because
he hasnt seen Henk but wishes to provide information that
is both true (quality) and relevant, and sufficient to be use-
ful (quantity) and clear enough (manner). By virtue of the
assumption that B is following these maxims, Bs utterance
can suggest, or conversationally implicate , in Grices termi-
nology, that Henk is somewhere close by. Despite the fact
that we often have reasons or cultural conventions for being
obscure or economical with the truth (Sacks 1975 ; Ochs 1976 ),
such indirect answers seem to be universal, suggesting that
the background assumption of cooperation holds right across
the cultures of the world .
Pragmatics, Universals in
656
children learn a language and use it in strikingly parallel ways.
Te strong universals of use suggest that language, in fact, rides
on a rich, language-independent infrastructure. A crucial ele-
ment is the ability to infer intentions from actions. Grice ( 1957 )
outlined a psychological theory of non-natural meaning or
communication along the following lines: A communicator
intends to cause an efect in an addressee by producing an
action or utterance that is designed to cause that efect just
by having that intention recognized (see communicative
intention ). Consider a nonverbal signal: A mother makes as
if to smooth her own hair, thereby signaling to her daughter in
a school concert that the daughters hair is in disarray if the
child recognizes her intent, communication has succeeded. No
conventional symbols are necessarily involved. Such a mode of
communication, which can be observed in nonconventional
sign languages like home-sign (Goldin-Meadow 2003 ), relies
on some form of reciprocal mind-reading abilities (Levinson
2006 ). It plausibly forms the basis for the learning of language,
as communication is evident in infancy (e.g., through point-
ing) prior to language acquisition (see communication,
prelinguistic ).
If a mind-reading ability is part of the infrastructure for
language, there are also other aspects of the pragmatic infra-
structure that are potentially independent of linguistic commu-
nication. For example, systematic turn-taking is discerniable in
infantcaretaker interaction long before verbal interchange is
possible. Similarly, the use of gesture, facial expression, gaze,
and posture in interaction appears early in child development.
All of this points to a large raft of abilities and inherited dispo-
sitions that makes language use possible in the form that we
know it. It is this infrastructure that infants use to bootstrap
themselves into language. What is now observable in ontogeny
was no doubt true also in phylogeny for this infrastructure
no doubt preceded the evolutionary specializations in anat-
omy and brain that now drive language (Enfeld and Levinson
2006 ).
Tere are yet other universals of language use that are refec-
tions of a common human ethology. We are one of the few species
that shows evidence of cooperative instincts that are not based
on kin selection. Tis cooperation is made possible by the subtle
linguistic and paralinguistic expression of solidarity, dominance,
and the juggling for position (see paralanguage ), much of
this explored by pragmaticists under the rubric of politeness
(Brown and Levinson 1987 ). Again, there seem to be systematic
universals here, both in the underlying dimensions expressed
(e.g., power, solidarity, degree of imposition) and in the basic
strategies used to express them (e.g., modulations of deference
or camaraderie) .
In sum, then, an understanding of universals in pragmatics
promises to give us deep insights into the infrastructure that lies
behind human communication and the language that is so dis-
tinctive of it. Tis infrastructure is arguably what lies behind the
development of language in infancy, as well as the evolution of
language in the species. Taken as a core part of human ethology,
it also tells us much about human nature and how it came to be
the way it is .
Stephen C. Levinson
alternation of speaking roles (Sacks, Scheglof, and Jeferson
1974 ). Tis turn-taking, of course, motivates the deictic system
already mentioned. Such informal talk is also characterized by
the immediacy of conversational repair ; that is, if address-
ees do not hear or understand what is said, they may query either
the whole or part, getting immediate feedback in the next turn
(Scheglof, Jeferson, and Sacks 1977 ). Such talk is structured
locally in terms of sequences (Scheglof 2006 ) in the simplest
case, adjacency pairs , that is, pairs of utterances perform-
ing actions like question-answer, ofer-acceptance, request-
compliance, greeting-greeting, and so forth. Sequences can be
embedded, as in A: Do you have Marlboros? B: You want
20s? A: Yes. B: Ah sorry, no. We do have 10s. Tey can also
be extended over more turns, for example by adding a prese-
quence as in: A: Do you mind if I ask you something? B: No.
A: Why did you give up that amazing job? B: Burnout. Given
the general expectation for rapid turn-taking, any participant
wishing to have an extended turn at talk is likely to negoti-
ate this, for example, through a prestory of the kind Have you
heard what happened to Bonny? During such an extended turn
at talk, feedback of restricted types ( mmhm, uhuh , etc.) may be
expected. In addition to these local levels of organization, con-
versations also generally have overall structures for example,
they are likely to be initiated by greetings and ended with part-
ings, each with its distinctive structure.
All of this detailed structure seems entirely general across
cultures and languages, although there may be constraints of
many local kinds about who can talk to whom and where in
this informal way. Ethnographic reports to the contrary do not
seem to stand the test of close examination. Tere are, though,
many aspects of cultural patterning that can be very distinctive.
For example, although in all cultures conversation makes use
of multimodal signals (gaze, gesture, facial expression, etc.) in
face-to-face interaction, the details can difer strikingly, whereas
Tzeltal speakers avoid gaze and the signals that would be thus
made available, Rossel Islanders presume mutual gaze and so
can systematically signal responses like yes, no, amazing!
and so on by facial expression.
In addition to these general observations about conversa-
tional universals, there seem to be very detailed generalizations
about specifc actions. For instance, in a wide sample of lan-
guages, it seems that reference to persons follows a precise set
of expectations about the form of reference expressions, as well
as the procedures to follow when the expression proves inade-
quate (Stivers and Enfeld 2007). Tus, utterances of the follow-
ing kind, where specifc components are added incrementally
and in order until recognition is signaled, can be expected
in any language: John (.) Wilkins (.) Te man you met at the
party.
Human Ethology and Communication
Human language is unique in the animal world by virtue of its
complex internal structure, its potential displacement across
modalities (as in sign languages ), and its wide range of
functions. It is also the only animal communication system that
exhibits great diversity in structure and meaning across social
groups. Tis diversity shows that it is heavily interdependent
with historical and cultural processes. Nevertheless, all normal
Pragmatics, Universals in
657
Sperber , Dan , and Deirdre Wilson . 1995 . Relevance: Communication and
Cognition . 2d ed. Oxford : Blackwell .
Stivers , Tanya , and Nick J. Enfeld , eds. 2007. Person Reference
in Interaction: Linguistic, Cultural and Social Perspectives .
Cambridge : Cambridge University Press .
PRAGMATISM AND LANGUAGE
According to a very infuential conception of language, lan-
guage functions by tracing truth conditions. Individual words
denote objects, properties, and relations, and combinations of
words in sentences represent possible states of afairs (see
truth conditional semantics ). Te pragmatist rejects this
conception of language, arguing that we must focus not on what
is the case if a sentence is true but, instead, on what follows if it
is true.
As originally formulated by Charles Sanders Peirce in the late
nineteenth century, pragmatism is motivated by a nonfounda-
tionalist conception of scientifc inquiry. Rather than taking
inquiry to begin with some basic truths, deriving consequences
in light of ones understanding of what follows from what, Peirce
takes inquiry to begin with a hypothesis, asking what would fol-
low on that hypothesis. If various consequences are true, then
one has reason to believe that ones hypothesis is true as well; if
any consequence is false, then the hypothesis must be rejected.
Because it is impossible to exhaust all the consequences of a
given claim, it follows immediately that there is no certainty, no
indubitable truth. Anything we think we know, however self-ev-
ident it may seem, can turn out to have been mistaken; nothing
is (as Wilfrid Sellars would say) given . But although nothing is
given as the frm and indubitable foundation for inquiry, there
is, at any stage in inquiry, much that one has no reason to doubt.
Tat is where we must start, from where we are, while at the same
time recognizing that in our inquiries we do not stand upon the
bedrock of fact but are instead walking upon a bog, and can
only say, this ground seems to hold for the present (Peirce 1992 ,
1767). Judgment, on such a view, is inherently provisional; it
not only corrects its conclusions, it even corrects its premises
(Peirce 1992 , 165). For the pragmatist, such a nonfoundationalist
and fallibilist conception of inquiry motivates, in turn, the idea
that meaning is to be understood not by reference to truth but by
reference to consequences.
Tis pragmatist conception of meaning in terms of conse-
quences is especially plausible for the case of mathematical and
natural scientifc concepts. Whereas the standard foundation-
alist view would seem to require some special insight into the
basic truths of mathematics, the pragmatist takes mathematics
to proceed experimentally, by axiomatizing some domain,
thereby making explicit our (current) understanding of the con-
cepts relevant to that domain and deriving theorems as a means
of testing the adequacy of that understanding. Similarly, in the
empirical sciences, we form theories, the empirical adequacy
of which is determined by reference to the observable efects
of the theory. Te pragmatist conception of meaning is much
less plausible in the case of the everyday prescientifc concepts
of natural language, concepts of sensory qualities such as red-
ness, say, or even of a substance such as water as it is presci-
entifcally understood, concepts the contents of which seem
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brown , Penelope , and Stephen C. Levinson . 1987 . Politeness: Universals
in Language Usage . Studies in Interactional Sociolinguistics 4.
Cambridge : Cambridge University Press .
Enfield , Nick J ., and Stephen C. Levinson , eds. 2006. Roots of Human
Sociality: Culture, Cognition and Human Interaction . Oxford :
Berg .
Goldin-Meadow , Susan . 2003 . Te Resilience of Language: What Gesture
Creation in Deaf Children Can Tell Us About How All Children Learn
Language . New York : Psychology Press .
Grice , H. P . 1957 . Meaning . Philosophical Review 67 : 377 88.
. 1975 . Logic and conversation. In Syntax and Semantics. Vol.
3: Speech Acts . Ed. P. Cole and J. Morgan , 4158. New York : Academic
Press .
. 1989 . Studies in the Way of Words . Cambridge : Harvard University
Press
Haspelmath , Martin , Matthew Dryer , David Gil , and Bernard Comrie ,
eds. 2005. Te World Atlas of Language Structures . Oxford : Oxford
University Press .
Hockett , C. F . 1960 . Te origin of speech . Scientifc American
203 : 89 96.
Horn , Laurence . 1984 . Toward a new taxonomy for pragmatic infer-
ence: Q- and R-based implicature. In Meaning, Form and Use in
Context , ed. Deborah Shifrin , 1142. Washington, DC : Georgetown
University Press .
Huang , Yan . 2007 . Pragmatics . Oxford : Oxford University Press .
Hymes , Dell . 1982 . Models of the interaction of language and social life.
In Directions in Sociolinguistics , ed. John J . Gumperz and Dell Hymes ,
3571 . New York : Holt, Rinehart and Winston .
Jackendof , Ray . 2002 . Foundations of Language . Oxford : Oxford
University Press .
Levinson , Stephen . 1983 . Pragmatics . Cambridge : Cambridge University
Press .
. 2000 . Presumptive Meanings: Te Teory of Generalized
Conversational Implicature . Cambridge, MA : MIT Press .
. 2006 . On the human interaction engine. In Roots of Human
Sociality , ed. N. Enfeld and S. Levinson , 3969. Oxford : Berg .
Levinson , Stephen C ., and E. Annamalai 1992 . Why presuppositions
arent conventional. In Language and Text: Studies in Honour of Ashok
R. Kelkar , ed. R. N. Srivastava , Suresh Kumar , K . K. Goswami , and
R. V. Dhongde , 22742. Dehli : Kalinga Publications .
Newmeyer , Fredrick . 2004 . Typological evidence and universal gram-
mar . Studies in Language 28 : 527 48.
Ochs , Elinor . 1976 . Te universality of conversational postulates .
Language in Society 5.1/3 : 67 80.
Sacks , Harvey . 1975 . Everyone has to lie. In Sociocultural Dimensions
of Language Use , ed. M. Sanches and B. Blount , 5780. New
York : Academic Press .
Sacks , Harvey , Emanuel Scheglof , and Gail Jeferson . 1974 . A sim-
plest systematics for the organization of turn-taking in conversation .
Language 50 : 696 735.
Sadock , Jerrold , and Arnold Zwicky . 1985 . Speech act distinctions in
syntax. In Language Typology and Syntactic Description. Vol. l: Clause
Structure . Ed. Timothy Shopen , 15596. Cambridge : Cambridge
University Press .
Scheglof , Emanuel . 2006 . Sequence Organization in Interaction: A
Primer in Conversation Analysis . Cambridge : Cambridge University
Press .
Scheglof , Emanuel , Gail Jeferson , and Harvey Sacks . 1977 . Te prefer-
ence for selfcorrection in the organization of repair in conversation .
Language 53 : 361 82.
Searle , John . 1976 . Speech Acts: An Essay in the Philosophy of Language .
Cambridge : Cambridge University Press .
Pragmatics, Universals in Pragmatism and Language
658
is no given meaning to the signpost (or utterance or apple), no
meaning that it has independent of the ways we actually go on
in light of it, we suppose instead that the signpost has the mean-
ing it does because we respond to it in a certain way, because
it is taken a certain way, because, as Wittgenstein puts it, it is
interpreted. But it can then be argued that this response, too,
has no given meaning, no meaning independent of the ways
we actually go on in light of it. So in order for it to be a nor-
matively signifcant response, a taking of the sign as meaning
such and such, a further response would seem to be needed.
But this clearly starts a vicious regress. Perhaps, then, we need
the notion of a response that is inherently meaningful, that just
is normatively signifcant. Such responses are not answerable
to things as they actually are (as this would require a given) but
only to things as they are taken to be; and they are essentially
social because otherwise whatever seems right to one will be
right, and then we cannot speak of right (see private lan-
guage argument ) .
According to this Jamesian reading (rehearsed, for instance,
by Brandom), the fact that nothing is given requires turning
instead to takings, socially articulated, normatively signifcant
responses to things such as signposts, utterances of other speak-
ers, objects, and states of afairs, in virtue of which the things
responded to have the signifcances they are taken to have. A
more radical, Peircean response jettisons not merely the given
but also the whole framework relative to which we must choose
between a mythic given and a merely socially articulated taking
(see, for example, McDowell 2002 ). Te model here is the way
animals interact with things in their environments. Grass, for
example, is food for some animals. But grass is not simply given
as food; that it serves as food depends on there having evolved
animals for whom grass is nourishing and is eaten for nourish-
ment. Nor is grass food merely by being taken to be so, by in
fact being taken up and eaten. (An animal might on occasion
eat something that is not nourishing for it, a piece of plastic,
say; and it may at times be unable to digest food, that is, stufs
that generally are nourishing for it.) Instead, grass has the sig-
nifcance of being food only relative to the kind of animal for
which it is food; and contrariwise, the animal is intelligible as
the sort of animal it is, as an instance of a particular form of
life , only in light of its environment providing opportunities
(such as food) and hazards for it. Being an animal and having
an environment are correlative notions; neither is intelligible
without the other.
Similarly, being a speaker and having the world in view as the
stuf of ones talk and the standard of the correctness of ones
judgments are correlative notions, neither intelligible without
the other. Tere is no given if by that one means things revealed
as meaningful to us independent of the evolution and acquisi-
tion of language, but nor is language use merely a matter of tak-
ings. Rather, through ones acculturation into natural language,
one comes to have the world in view as that about which one
speaks, much as an animal, through its development into the
kind of animal it is, comes to have an environment to which it
is perceptually sensitive and through which it moves. And it is
the world that is in view for a speaker, on this reading, precisely
because meaning and truth are not given but are instead the
fruits of inquiry.
not to be exhausted by their (observable) consequences but
ineliminably to involve also a particular phenomenal quality
(see language, natural and symbolic ). Pragmatists have
nonetheless tended to understand the contents of all concepts,
whether belonging to natural or to symbolic language, in terms
of their consequences .
Te pragmatist conception of meaning in terms of con-
sequences shifts attention away from truth as the product of
inquiry toward the process of inquiry, the striving for truth;
and it does so because (in the absence of a given foundation)
it is not settled in advance how conficts, as they arise, are to
be adjudicated, which of the competing claims are to be jet-
tisoned, and which retained, if only provisionally. Suppose,
for example, that we fnd some stuf that looks like water but,
on analysis, is shown to be not H
2
O but some other chemical
stuf, call it XYZ. What should we conclude? Tere are many
options. Perhaps the fault lies with our analytic procedure
or in the execution of it. Perhaps water is not inevitably H
2
O.
Perhaps the stuf is not really water. And other responses are
possible as well. At any given point in our ongoing inquiry,
some responses will seem more plausible than others; never-
theless, it is not simply given what the correct response is. It is
only the way we actually go on in the course of inquiry that will,
retrospectively and defeasibly, settle what our words mean. In
a slogan, meaning lies in use.
As originally conceived by Peirce, and defended more recently
by Sellars , this pragmatist conception of meaning enables a
fully robust notion of objective truth, a conception of scientifc
inquiry as answering to things as they are. As interpreted by
William James , and defended more recently by Richard Rorty
and (more subtly) by Robert Brandom , pragmatism entails rela-
tivism, a conception of scientifc inquiry as answering only to our
interests, to what the community of, say, scientists takes to be the
case. Tere is, on this latter view, no objective standard govern-
ing the correctness of our judgments but only a social one, no
truth but only solidarity. And it is not hard to understand how
the pragmatist conception of meaning can seem to entail such a
view. If, as the pragmatist thinks, there is no given foundation of
meaning and truth, then it can seem to follow that we have only
our takings, our subjective conceptions of things to go on. But if
so, then objectivity would seem to require the impossible: that
we step outside of language, outside of our subjective concep-
tions, to see how things are independent of those conceptions.
If there is no given but only taking, then our inquiries cannot be
answerable to things as they are.
In his Philosophical Investigations (1953), Ludwig
Wittgenstein argues for what is, in efect, the pragmatist concep-
tion of meaning in terms of use and against the representation-
alist conception of meaning. And here again, both a Peircean,
realist reading and a Jamesian, relativistic reading are possible.
Language use is essentially normative, subject to standards of
correctness that speakers in some way grasp or understand. It
is, as Wittgenstein thinks of it, a matter of rule-following .
Te task is to understand how exactly this works. We begin with
an expression of the rule, a signpost, for instance, that shows
the way. (We could equally well begin with a persons utterance
showing what that person thinks or even with something like an
apple that shows itself to a perceiver as an apple.) Because there
Pragmatism and Language
659
PREFERENCE RULES
Preference rule systems constitute a form of rule interaction
related to default logic and harmonic grammar (Smolensky and
Legendre 2006 ). Tey are introduced in semantic theory in
Jackendof ( 1983 ) and in generative music theory in Lerdahl and
Jackendof ( 1983 ) and argued to be ubiquitous in cognition.
A standard example is the meaning of the verb climb . A stereo-
typical case such as John climbed for hours is interpreted as John a)
moving upwards on a surface, b) with an efortful clambering man-
ner of motion. Both conditions are violable. John climbed down the
mountain and John climbed across the clif do not involve upward
motion; the airplane climbed steadily entails upward motion but
not clambering. However, both conditions cannot be violated at
once: * Te airplane climbed down 5,000 feet .
Tese examples make it impossible to analyze the meaning
of climb in terms of necessary and sufficient conditions ,
as assumed in the philosophical and formal logic traditions.
Neither condition is necessary, but either one is suf cient for
an action to count as climbing. At frst blush, this suggests that
the conditions are simply disjunctive. However, there is a further
wrinkle: Satisfying both constraints results in a more stereotypi-
cal use of climb , and in cases where there is no evidence to the
contrary, both conditions are assumed by default. Tus, prefer-
ence rule systems provide a formal characterization of Ludwig
Wittgensteins ( 1953 ) and E. Rosch and C. Merviss ( 1975 ) notion
of categories displaying a family resemblance : Tere is no
single criterial condition for members of the category, stereotypi-
cal members satisfy all or most conditions, and marginal mem-
bers satisfy fewer conditions .
Preference rule systems difer from optimality- theoretic
rule systems in that the constraints, though violable, are not
ranked: Under proper conditions, either rule can dominate
the other. A classical example comes from gestalt principles of
visual grouping (Wertheimer [1924] 1938 ), where grouping of
units can be based either on their relative distance (1a) or their
relative similarity (1b). Tus, either condition is suf cient for
grouping:
(1) a. x x x x x x x x x [identical units with variable spacing]
b. x x xX X Xx x x [diferent units with identical spacing]
In displays with variable units and variable distances, alignment
of the two conditions produces stronger grouping judgments
(2a). If the two conditions are not aligned, a judgment can be
forced by suf cient disparity either in distance (2b) or in form
(2c).
(2) a. x x x X X X x x x [stronger judgment]
b. x x X X X x x x x [distance overrules size]
c. x x
X
X
X
x x x x [size overrules distance]
When attempting to state the conditions on a category or
rule, then, one should suspect the presence of a preference rule
system when a) every condition one can think of as criterial has
important counterexamples, b) there are diferent counterexam-
ples to each condition, and c) satisfaction of all (or most) condi-
tions produces stereotypical instances of the category or rule .
Ray Jackendof
Te pragmatist critique of the representationalist conception
of language in terms of unquestionable, or given, wordworld
relations of denotation does not show merely that the founda-
tion is diferent than we had thought, that it is social rather than
objective, but instead, and more radically, that the objectivity
of inquiry requires not a foundation but the capacity for critical
refection: [E]mpirical knowledge is rational, not because it
has a foundation but because it is a self-correcting enterprise
which can put any claim in jeopardy, though not all at once
(Sellars 1997, 38). Not only do we revise our beliefs about things,
but we also revise our conceptions of the kinds of things there
are and can be. Indeed, we even revise our most fundamental
understanding of the nature of reality as a whole, and in so doing,
we come to ever more adequate languages with which to address
things as they are, the same for all rational beings .
Danielle Macbeth
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brandom , Robert . 1994 . Making It Explicit: Reasoning, Representing, and
Discursive Commitment . Cambridge : Harvard University Press .
James , William . 1975 . Pragmatism. Cambridge : Harvard University
Press.
Macbeth , Danielle . 1995 . Pragmatism and the philosophy of language .
Philosophy and Phenomenological Research 55 : 501 23.
. 2007 . Pragmatism and objective truth. In Te New Pragmatists ,
ed. C. Misak, 16992. Oxford : Clarendon .
McDowell , John . 1994 . Mind and World . Cambridge : Harvard University
Press .
. 2002 . How not to read philosophical investigations: Brandoms
Wittgenstein. In Wittgenstein and the Future of Philosophy: A
Reassessment after 50 Years , ed. R. Haller and K. Puhl , 25162.
Vienna: Holder, Pichler, and Tempsky .
Peirce , Charles Sanders . 1931 58. Collected Papers of Charles Sanders
Peirce. 8 vols . Cambridge : Harvard University Press.
. 1992 . Reasoning and the Logic of Tings: Te Cambridge Conference
Lectures of 1898 , ed. K. Ketner. Cambridge : Harvard University Press .
Rorty , Richard . 1982 . Consequences of Pragmatism (Essays: 19721980) .
Minneapolis : University of Minnesota Press.
Sellars , Wilfrid . 1997 . Empiricism and the Philosophy of Mind .
Cambridge : Harvard University Press.
Wittgenstein , Ludwig . 1953 . Philosophical Investigations . Oxford :
Blackwell.
PREDICATE AND ARGUMENT
In all languages, the vocabulary consists of two basic types of
words : those that denote entities, such as pronouns and proper
names, and those (such as verbs, adjectives, and adverbs) that
present information about entities, such as their properties,
states, or transformations. In a terminology derived from logic,
the relational words are called predicates, and the entities that
they relate to are called their arguments. Predicates are like func-
tions in mathematics, with their arguments serving as variables.
In traditional grammar, the term predicate is used also for one of
two constituent parts of a sentence , the other being the sub-
ject. (See also quantification, categorial grammar , and
montague grammar .)
Anat Ninio
Predicate and Argument Preference Rules
660
or what people think its record to have been (1989, 4), not only
describing the status quo but also including a diachronic dimen-
sion (see synchrony and diachrony ). While prestige is
mostly associated with speaker evaluations, its role as a catalyst
for other characteristics, such as functional specialization, liter-
ary heritage, and standardization , has also been acknowl-
edged. In a dynamic model of language prestige and prestige
change, Susanne Mhleisen ( 2002 ) looks at the interaction
among societal, institutional and interactional, and sociopsy-
chological dimensions of prestige. Te dynamics of this inter-
action may result in various types and directions of changes in
language prestige .
Susanne Mhleisen
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Ammon , Ulrich . 1989 . Towards a descriptive framework for the status/
function (social position) of a language within a country. In Status
and Function of Languages and Language Varieties , ed. Ulrich Ammon ,
21106. Berlin : de Gruyter.
Bilaniuk , Laada . 1993 . Diglossia in fux: Language and ethnicity in
Ukraine . Texas Linguistic Forum 33 : 79 88.
Ferguson , Charles A . 1959 . Diglossia . Word 15.2 : 325 40.
. 1966 . National sociolinguistic profle formulas. In Sociolinguistics ,
ed. William Bright, 30924. Te Hague : Mouton.
Mackey , William F . 1989 . Determining the status and function of lan-
guages in multinational societies. In Status and Function of Languages
and Language Varieties , ed. Ulrich Ammon, 320. Berlin : de Gruyter .
Mhleisen , Susanne . 2002 . Creole Discourse: Exploring Prestige
Formation and Change Across Caribbean English-Lexicon Creoles .
Amsterdam : Benjamins .
Rickford , John . 1983 . Standard and nonstandard attitudes in a Creole
community. Society for Caribbean Linguistics (Occasional Paper 16),
University of the West Indies, St. Augustine, Trinidad.
Stewart , William A . 1968 . A sociolinguistic typology for describing
national multilingualism. In Readings in the Sociology of Language ,
ed. Joshua A. Fishman, 53145. Te Hague : Mouton .
PRESUPPOSITION
A presupposition is a precondition of a sentence such that the
sentence cannot be uttered meaningfully unless the presuppo-
sition is satisfed. Te concept of a presupposition originated
with Gottlob Frege ( 1892 ), but the English term was coined by
Peter F. Strawson ( 1950 ). Presupposition theory is an area of
active research at the semantics/pragmatics interface. A
related term is conventional implicature. H. Paul Grice
(1975) distinguished between presuppositions and conven-
tional implicatures, however, it is still under debate whether
such a distinction is necessary (cf. Potts 2007 and ensuing
discussion).
definite descriptions have played a major role in the
development of presupposition theory and are still generally
analyzed as introducing a presupposition. Consider for example
(1). No entity that satisfes the description biggest natural num-
ber exists. What is the status of (1)? Is it true or false?
(1) Te biggest natural number is prime.
Presupposition theory says (1) is neither true nor false: A
defnite description the NP (noun phrase) presupposes the
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Jackendof , R . 1983 . Semantics and Cognition . Cambridge, MA : MIT
Press .
Lerdahl , F. , and R. Jackendof . 1983 . A Generative Teory of Tonal Music .
Cambridge, MA : MIT Press .
Rosch , E ., and C. Mervis . 1975 . Family resemblances: Studies in the
internal structure of categories . Cognitive Psychology 7 : 573 605.
Smolensky , P. , and G. Legendre . 2006 . Te Harmonic Mind . Cambridge,
MA : MIT Press .
Wertheimer , M. [1924] 1938. Laws of organization in perceptual
forms. In A Source Book of Gestalt Psycholog y , ed. W. D. Ellis, 7188.
London : Routledge and Kegan Paul.
Wittgenstein , L. 1953 . Philosophical Investigations . Oxford : Blackwell.
PRESTIGE
Language prestige refers to the social position of a language,
especially in multilingual settings (see bilingualism and
multilingualism ), and the purposes it is used for, as well as
peoples beliefs and feelings about it. Te prestige of a variety is
unrelated to its structure and can only be determined in a social
context. Te prestige of a particular language may, therefore,
difer greatly from one speech community to another and may
also be subject to change. Typically, the languages of immigrant
groups (e.g., Turkish in Germany) have a relatively low prestige
in comparison to that in the country of origin (e.g., Turkish as
the national and of cial language in Turkey). An example of
a prestige change is the move of Ukrainian from low to high
prestige in post-Soviet Ukraine, a position that had been previ-
ously held by Russian as the exclusive high-prestige language
(Bilaniuk 1993 ).
Charles A. Ferguson ( 1959 ) used the term prestige to describe
the functional distribution of two language varieties of the same
language, a H(igh prestige) language and a L(ow prestige) lan-
guage in diglossic situations. Te term often serves as an
umbrella notion encompassing status and functions of lan-
guages, on the one hand, and language attitudes, on the other.
As of the 1960s, various classifcations of status and function of
languages had been proposed (see, for instance, Ferguson 1966 ;
Stewart 1968 ), mostly for the description of national sociolin-
guistic profles in multilingual societies . Some of these sug-
gestions were later taken up and redefned in other detailed
frameworks (see Ammon 1989 ; Mackey 1989 ) where prestige
stands for an important sociocultural dimension. Along with
features such as demographic factors, institutional support, and
status (see language policy ), prestige is also seen as a fac-
tor of the ethnolinguistic vitality of a linguistic group, that
is, that which makes it behave as a distinctive entity within mul-
tiethnic and multilingual settings. In language attitude studies,
we can distinguish between speaker evaluations in terms of overt
prestige (i.e., as languages of authority) and covert prestige (i.e.,
as languages of solidarity). In a diglossic language situation, for
instance in Guyana, the L-language (Guyanese Creole) is attrib-
uted a high-solidarity and a low-authority value, whereas the
H-language (English) holds a low-covert and a high-overt pres-
tige (Rickford 1983 ).
Te prestige of a language is often explained in historical
terms. Tus, William F. Mackey speaks of a languages record,
Prestige Presupposition
661
presupposition projection that has given rise to dynamic seman-
tics (see also Beaver 2001 ; Kadmon 2001 ). However, the projec-
tion problem is still subject to lively debate (see Schlenker 2007 ) .
Uli Sauerland
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Abusch , Dorit 2005 . Triggering from alternative sets and projection
of pragmatic presuppositions. Unpublished manuscript, Cornell
University.
Beaver , David . 2001 . Presupposition and Assertion in Dynamic Semantics .
Stanford, CA: CSLI Publications.
Frege , Gottlob . [1892] 1952. ber Sinn und Bedeutung. Zeitschrift fr
Philosophie und philosophische Kritik 100 : 2550. English transla-
tion: On sense and reference. In Translations from the Philosophical
Writings of Gottlob Frege , ed. Peter T. Geach and M. Black , 5678.
Oxford : Blackwell .
Gazdar , Gerald . 1979 . Pragmatics: Implicature, Presupposition, and
Logical Form . New York : Academic Press .
Grice, Herbert Paul. 1975. Logic and conversation. In Syntax and
Semantics . Vol. 3. Ed. Cole and Morgan, 4158. New York: Academic
Press.
Heim , Irene . 1983 . On the projection problem for presuppositions.
In Proceedings of WCCFL 2, ed. Dan Flickinger, 11425. Stanford,
C A: CSLI .
. 1991 . Artikel und Defnitheit. In Semantik: Ein internationales
Handbuch der zeitgenssischen Forschung , ed. Arnim von Stechow and
Dieter Wunderlich , 487535. Berlin : Mouton de Gruyter .
Kadmon , Nirit . 2001 . Formal Pragmatics: Semantics, Pragmatics,
Presuppositions and Focus . Malden, MA, and Oxford : Blackwell .
Karttunen , Lauri . 1974 . Presuppositions and linguistic context .
Teoretical Linguistics 1 : 181 94.
Karttunen , Lauri , and S. Peters . 1979 . Conventional implicature. In
Presupposition , ed. C. Oh and D. Dineen , 156. New York : Academic
Press .
Levinson , Stephen C . 1983 . Pragmatics . Cambridge : Cambridge
University Press .
Potts , Christopher 2007 . Te expressive dimension . Teoretical
Linguistics 33 .2: 16597. Includes commentary by other scholars.
Russell , Bertrand . 1905 . On denoting . Mind, n.s., 14 , 479 93.
Sauerland , Uli . 2008 . Implicated presuppositions. In Sentence and
Context , ed. A. Steube , 581600. Berlin : Mouton de Gruyter .
Schlenker , Philippe . 2007 . Transparency: An incremental theory of
presupposition projection. In Presupposition and Implicature in
Compositional Semantics , ed. Uli Sauerland and Penka Stateva , 21442.
Basingstoke, UK : Palgrave Macmillan .
Stalnaker , Robert . 1973 . Presuppositions . Journal of Philosophical Logic
2 : 447 57.
Strawson , Peter F . 1950 . On referring . Mind 59 : 320 44.
PRIMATE VOCAL COMMUNICATION
Nonhuman primate vocal communication is one of our few
links to understanding the evolution of human speech and its
underlying physiological bases. Since the vocal tract and the
brain do not fossilize, insights into the origins of human com-
munication occur through a comparison of the vocal behavior of
extant primates with humans. Tis comparative approach forms
a framework upon which testable hypotheses on the evolution of
speech can be based. Also, since the brain is critical for the pro-
duction and perception of vocalizations, nonhuman primates
existence of an individual that satisfes NP in other words,
defnite descriptions carry an existence presupposition .
Presupposition failure describes the case wherein a presuppo-
sition is not fulflled like (1). Presupposition failures are ana-
lyzed as being neither true nor false, but as being truth value
gaps. Presupposition theory, therefore, relies on a distinction
among three possible truth values a sentence may have: true,
false, and undefned. One important argument in support of
a third truth value has been the interaction between negation
and presuppositions: A presupposition failure in many cases
remains a presupposition failure even when the sentence is
negated:
(2) Te biggest natural number is not prime.
It follows that (2), like (1), is a presupposition failure just if nega-
tion does not change the conditions under which a sentence
has a truth value. Negation can be used in this way as a presup-
position test: A presupposition follows from a sentence and its
negation. Te assertion, on the other hand, only follows from the
sentence itself, and not from its negation.
Just as the existence presupposition of the sentences in (1)
and (2) is triggered by the defnite article the , many other words
trigger presuppositions. Stephen C. Levinsons ( 1983 ) textbook
lists several pages of presupposition triggers in English. A par-
ticularly interesting paradigm is that in (3) (cf. Abusch 2005 ): (3a)
has no relevant lexically triggered presupposition, whereas (3b)
presupposes that it is actually raining outside and asserts that
Bill thinks so, too. Finally, (3c) presupposes that Bill thinks that
it is raining outside, and asserts that it actually is raining outside.
It is particularly interesting that be right and know have the same
truth conditions, but difer on which part of them is presup-
posed. Paradigm (3) shows that part of our specifc knowledge
about think , know , and be right is whether they trigger a presup-
position and which one.
(3) a. Bill thinks that its raining outside.
b. Bill knows that its raining outside.
c. Bill is right that its raining outside.
Some presuppositions are not lexically triggered. For exam-
ple, (3a) cannot be used if it is known that it really is raining
outside. Tis presupposition, however, has been analyzed as an
implicated presupposition (Sauerland 2008 ). It can be derived in
a similar way to conversational implicatures as arising from the
avoidance of a presupposition trigger and a principle of presup-
position maximization (Heim 1991 ).
One central problem of presupposition theory is the question
of how to predict the presuppositions of complex sentences the
problem of presupposition projection . Lauri Karttunen and S.
Peters ( 1979 ) show that while negation does not afect presuppo-
sitions, in other complex sentences presupposition triggers can
occur, but the presupposition may not project to the entire sen-
tence: In example (4), the conditional clause blocks projection of
the existence presupposition of the biggest natural number .
(4) If there was a biggest natural number, the biggest natural
number would be prime.
Building on work by Robert Stalnaker ( 1973 ) and Karttunen
( 1974 ), Irene Heim ( 1983 ) has developed an infuential account of
Presupposition Primate Vocal Communication
662
most meaningful unit from the perspective of socially isolated
receivers.
Tere is also some evidence for a lexical syntax in which
diferent combinations of acoustic units are used to trans-
mit diferent meanings to listeners (Zuberbuhler 2002 ; Arnold
and Zuberbuhler 2006 ). Tese data come from studies of two
African forest monkey species. First, not only do Diana monkeys
( Cercopithecus diana ) perceive the leopard and alarm calls of a
sympatric species, Campbells monkeys ( C. campbelli ), but they
also seem to understand that if these two calls are preceded by
a Campbells monkeys boom call, the threat is less urgent.
Importantly, if a Campbells monkeys boom call occurs before
a Diana monkeys species-specifc alarm call, it has no efect on
the Diana monkeys behavior. Second, putty-nosed monkeys ( C.
nictitans ) produce two diferent alarm calls: pyows for leopards
and hacks for eagles. When produced, each elicits a stereotypi-
cal escape response from the listeners. However, when males
combine the calls to form pyow-hack sequences, the combina-
tion does not elicit escape responses but, instead, elicits general
group movement .
REFERENTIAL COMMUNICATION. Te species-specifc vocaliza-
tions of many primates, as well as many other animals, can
be used by a listener as a source of information about objects,
events, and the status of peers in their environment. Tese
vocalizations are important since, on the basis of acoustic struc-
ture alone, listeners can extract functional (referential) infor-
mation about a vocalizations meaning (see reference and
extension ).
A classic example of referential communication signal-
ing is use of predator alarm calls by the vervet monkeys
( Cercopithecus aethiops ) (Seyfarth, Cheney, and Marler 1980 ).
Vervets produce unique alarm calls for three diferent preda-
tors: snakes, leopards, and eagles. When an alarm call is pro-
duced, it initiates predator-appropriate behaviors in listeners.
For example, when vervets hear an eagle-alarm call, they scan
the sky for visual cues of the airborne predator, and in some
cases, run to locations that provide overhead coverage. In con-
trast, when they hear a snake-alarm call, they stand up and
scan the ground. Finally, a leopard-alarm call initiates a third
distinct behavior: Vervets run up the nearest tree while scan-
ning the horizon for the leopard.
Te capacity to process referential signals successfully also
allows animals to use the referential information that is trans-
mitted by the vocalizations of other species (Zuberbuhler 2000 ).
For example, female Diana monkeys elicit a predator alarm call
when they hear a male Diana monkey producing a leopard alarm
or when they hear the leopard-alarm call of a crested guinea fowl.
Tis observation is important since it suggests that Diana mon-
keys can form abstract categorical representations of a vocaliza-
tions functional meaning that is independent of acoustics and
the species generating the signal.
Another example of the categorization of referential infor-
mation is the food-associated calls of rhesus macaques
( Macaca mulatta ) (Hauser 1998 ; Giford, Hauser, and Cohen
2003 ). When free-ranging rhesus monkeys encounter low-
quality food, they elicit one of two acoustically distinct vocal-
izations, coos or grunts. In contrast, when they encounter
(hereafter, primates ) are the ideal model system through which
we can directly monitor neurons and neural ensembles to fnd
the causal links between brain activity and vocal behavior.
Vocal Perception In Nonhuman Primates
To date, this comparative approach has been most fruitful when
scientists have examined how primates perceive their own vocal-
izations and then compare how these perceptions relate to those
occurring in human speech production. In this chapter, we
review some of these fndings.
SYNTACTIC PROCESSING OF VOCAL SEQUENCES. Many primate
species produce bouts of vocalizations that contain sequences of
similar acoustic units and/or diferent-sounding acoustic units.
Do these units separately code meaningful information (akin
to words in a sentence )? Or do they need to be combined to
form a meaningful utterance (akin to syllables in a word)? For
example, the chimpanzee ( Pan troglodytes ) pant-hoot con-
sists of a series of hoot calls followed by a series of screams.
Since both hoots and screams are produced individually in other
contexts, the pant-hoot could either be a single vocalization or a
bout of several vocalizations.
Most of our insights into the units of perception come from
studies of the orderly arrangement of the sound units in pri-
mate long calls (Marler 1968 ; Waser 1982 ). Long calls serve
as localization cues for conspecifcs and are produced in the
context of territorial encounters, mate attraction, and isola-
tion/group cohesion. Tese long calls provide evidence of
phonological syntax in which individual acoustic units are
assembled to form a larger, more functional (meaningful) unit.
For example, male titi monkeys and gibbons produce multi-
unit long calls that are used to demarcate and defend their ter-
ritories (Robinson 1979 ; Mitani and Marler 1989 ). When one
of these long calls is rearranged by a human experimenter and
presented to a conspecifc (in the form of a playback experi-
ment), the primate recognizes this novel vocalization and
responds as if there is a new male in the adjacent territory.
Gibbons produce signifcantly more squeak calls (given dur-
ing intergroup encounters) when hearing these novel stimuli,
whereas titi monkeys produce signifcantly more moaning
responses (also given in response to interspecies and inter-
group encounters). Tese data suggest that, at least in these
species, the global order of syllable sequences represents a cue
to individual recognition.
Te vocalizing behavior of cotton-top tamarins also provides
evidence of phonological syntax (Ghazanfar et al. 2001 ). When
socially isolated, tamarins elicit a long call that begins with one
to two chirps and ends with two to fve whistles. When conspe-
cifcs hear these vocalizations, they respond with their own calls,
a behavior called antiphonal calling . Do the individual chirps
or whistles provide functional information? Or is information
provided only by chirp-whistle combinations? Playback experi-
ments in which chirps, whistles, or the entire long call (chirp-
whistle combinations) are presented to tamarins were used to
address these questions. Tese experiments have shown that
the entire long call is more efective in eliciting antiphonal long
calls than isolated chirps and whistles, an observation consis-
tent with the hypothesis that in this species, the whole call is the
Primate Vocal Communication
663
across the superior temporal gyrus without reference to any
neuroanatomical subdivisions.
Recent anatomical and neurophysiological experiments
(Hackett, Preuss, and Kaas 2001 ; Rauschecker and Tian 2004 ;
Tian et al. 2001 ; Kaas and Hackett 1998 ) in rhesus monkeys have
identifed the serial and parallel processing that occurs between
the primary auditory cortex (A1) and secondary levels of audi-
tory processing along the superior temporal gyrus. Tese high-
er-order areas are called the caudolateral , middle lateral , and
anterolateral belt areas (CL, ML, and AL, respectively). Tese
studies suggest that belt neurons responded to more complex
sounds than A1 neurons. For instance, neurons in all three lat-
eral belt areas seem to prefer vocalizations to energy-matched
pure tone stimuli.
Other studies, however, suggest that A1 neurons are also
sensitive to the complex acoustic features needed in speech
and other types of auditory-object processing. For example, in
common marmosets ( Callithrix jacchus ) and squirrel monkeys,
A1 neurons are selective for species-specifc vocalizations and
phase-lock their fring pattern to the functional acoustic units
that comprise a vocalization, as opposed to the fner-grain
acoustic features that are not functionally meaningful (Bieser
1998 ; Wang et al. 1995 ; Lu, Liang, and Wang 2001 ). Also, A1
neurons, as well as those in belt and parabelt regions, are sensi-
tive to the pitch of an auditory stimulus, which may be used to
infer a vocalizations afective content (Bendor and Wang 2005 ).
Tese studies highlight the fact that the fow of information
from primary auditory areas to belt and parabelt regions is not
strictly serial but is parallel with both feedforward and feedback
interactions .
PREFRONTAL CORTEX. The frontal lobes contain an audi-
tory responsive region that responds robustly to vocaliza-
tions (Romanski, Bates, and Goldman-Rakic 1999 ). This
region, the ventrolateral prefrontal cortex (vPFC), has been
hypothesized to play an important role in processing the
more abstract components of vocalizations. Specifically, it
has been suggested that the vPFC plays an important role
in processing the referential information transmitted by a
vocalization. Indeed, in one set of experiments, it was dem-
onstrated that vPFC neurons were modulated more by differ-
ences between the food-related referential information (see
section on referential communication) that is transmitted
by a vocalization than by differences between their acous-
tic features (Gifford et al. 2005 ). These data suggested that,
on average, vPFC neurons are modulated preferentially by
transitions between presentations of food vocalizations that
belong to functionally meaningful and different categories.
Consistent with the proposed role of vPFC in categorical pro-
cessing, vPFC neurons in a second experiment responded in
the same way to different vocalizations that transmit infor-
mation about different types of food quality (i.e., high-qual-
ity and low-quality food) (Cohen, Hauser, and Russ 2006 ).
However, these same vPFC neurons responded differently to
different vocalizations that transmitted information about
different nonfood events .
Yale E. Cohen and Asif A. Ghazanfar
rare, high-quality food, they elicit one of two acoustically dis-
tinct vocalizations, harmonic arches or warbles. However,
despite the fact that these four vocalizations are all acoustically
distinct, rhesus do not discriminate among the vocalizations on
the basis of diferences in their acoustics but instead discrimi-
nate and categorize these vocalizations on the basis of the type
of referential information transmitted (e.g., low-quality versus
high-quality food) .
TEMPORAL CUES FOR VOCAL RECOGNITION. Duration, inter-
val, the order of acoustic features, and other temporal cues are
important components in the capacity of humans to distinguish
between diferent speech sounds. Te diference between the
two phonemes /pa/ and /ba/ is due to diferences in voice-on-
set time. Similarly, the diference between /sa/ and /sta/ is due
to diferences in the silent time between the consonants and the
vowels.
Primates can also use temporal information to distinguish
between diferent vocalizations. As discussed, cotton-top tam-
arins antiphonally call preferentially when they hear entire
long calls versus portions of the long calls (Ghazanfar et al.
2001 ). Further studies revealed that while tamarins did not
distinguish between normal calls and time-reversed or pitch-
shifted long calls (Ghazanfar et al. 2002), normal response
rates did require the species-specifc temporal structure of the
amplitude envelope (Ghazanfar et al. 2001). Finally, the num-
ber of acoustic units and the presentation rate may also infu-
ence antiphonal calling. Along similar lines, rhesus respond
diferently to shrill barks and grunts than to copulation calls
when the interval between the acoustic units is expanded or
contracted beyond the normal range (Hauser, Agnetta, and
Perez 1998 ). Finally, at least for shrill barks and harmonic
arches, there is evidence suggesting that rhesus are sensitive
to the temporal progression of these vocalizations amplitude
envelope: When shrill barks or harmonic arches are time-re-
versed, which changes their temporal structure but not their
spectral content, rhesus monkeys act as if they do not recog-
nize these stimuli as species-specifc vocalizations (Ghazanfar,
Smith-Rohrberg, and Hauser 2001 ) .
Neural Bases of Primate Vocal Communication
Overall, the referential, syntactic, and temporal features of pri-
mate vocalizations, from many diferent species, suggest strik-
ing parallels with human speech processing. How are these
features represented and processed at the level of neurons
and neural assemblies? Here, we review some relevant recent
fndings.
AUDITORY CORTEX. Despite the ethological importance of pri-
mate vocal communication in the lives of primates, we lack a
complete understanding of how biologically relevant features of
complex sounds are processed at the level of single neurons or
small populations of neurons. For many years, the squirrel mon-
key was the only primate model for investigating the role of audi-
tory cortex in processing species-specifc vocalizations. Studies
using this species found that many cells in the superior temporal
gyrus responded to species-specifc vocalizations. However, one
of the drawbacks of these studies was that recordings were made
Primate Vocal Communication
664
Romanski , L. M. , J. F. Bates , and P. S. Goldman-Rakic . 1999 . Auditory
belt and parabelt projections to the prefrontal cortex in the rhesus
monkey . J Comp Neurol 403. 2: 141 57.
Seyfarth , R. M ., D. L. Cheney , and P. Marler . 1980 . Monkey responses
to three diferent alarm calls: Evidence of predator classifcation and
semantic communication. Science 210. 4471: 801 3.
Tian , B. , D. Reser , A. Durham , A. Kustov , and J. P. Rauschecker . 2001 .
Functional specialization in rhesus monkey auditory cortex . Science
292 : 290 3.
Wang , X ., M. M. Merzenich , R. E. Beitel , and C. E. Schreiner . 1995 .
Representation of a species-specifc vocalization in the primary audi-
tory cortex of the common marmoset: Temporal and spectral charac-
teristics. J Neurophysiol 74. 6: 2685 2706.
Waser , P. M. 1977 . Individual recognition, intragroup cohesion, and
intergroup spacing: Evidence from sound playback to forest monkeys .
Behaviour 60 : 28 74.
. 1982 . The evolution of male loud calls among mangabeys
and baboons. In Primate communication , ed. C. T. Snowdon ,
C. H. Brown , and M. R. Petersen . New York : Cambridge University
Press .
Zuberbuhler , K . 2000 . Referential labelling in Diana monkeys . Anim
Behav 59 : 917 27.
. 2002 . A syntactic rule in forest monkey communication . Anim
Behav 63 : 293 9.
PRIMING, SEMANTIC
priming is used to describe a situation with two words (or other
entities) that are related, whereby an encounter with the one will
either facilitate or inhibit recovery of the other, either as regards
the speed or the accuracy with which it is recovered (see also
spreading activation ). Tere are various kinds of priming,
of which semantic priming is perhaps the most discussed in the
psycholinguistic literature. An example of semantic priming
is that the word dog will be processed more quickly if the word
bark has just been encountered. If the two words are unrelated
or an encounter with a non-word precedes the encounter with
a word, an encounter with the former will have no efect on the
speed with which the latter is processed; thus, the processing
of dog will be unafected if either teaching or prif has just been
encountered. Tis is automatic and independent of any inten-
tion or task-related motivation. When processing is afected,
the afected word is known as the target ; the word or entity that
afects the process is known as the prime . Primes do not have to
be words; they can be groups of words or complete sentences, or
they can be pictorial or aural. Te prime is assumed to partially
activate circuits that include the target .
Te assumption is that the degree of priming is proportional
to the semantic relatedness of the items (though this is not accu-
rate for all types of priming) and that priming is an automatic
process. Another assumption, which gives rise to the assump-
tion of proportionality, is that related words are stored closer
in semantic space to one another than are nonrelated words
(non-words are, of course, stored nowhere prior to encounter).
Indeed, one of the major values of priming studies is that they
permit investigation of the interconnections of the mental lexi-
con (Rumelhart and Norman 1985 ). Interpreting the results of
such studies, though, is not unproblematic; Kenneth I. Forster
( 1976 ) shows that the facilitation of processing is afected in
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Arnold , K. , and K. Zuberbuhler . 2006 . Language evolution: Semantic
combinations in primate calls . Nature 441. 7091: 303 .
Bendor , D ., and X. Wang . 2005 . Te neuronal representation of pitch in
primate auditory cortex . Nature 436. 7054: 1161 5.
Bieser , A . 1998 . Processing of twitter-call fundamental frequencies
in insulas and auditory cortex of squirrel monkeys . Exp Brain Res
122 : 139 48.
Chapman , C.A ., and D.M. Weary . 1990 . Variability in spider monkeys
vocalizations may provide basis for individual recognition. American
Journal of Primatology 22 : 279 84.
Cleveland , J ., and C.T. Snowdon . 1982 . Te complex vocal repertoire
of the adult cotton-top tamarin, Saguinus oedipus. Zeitschrift fr
Tierpsychologie 58 : 231 70.
Cohen , Y. E ., M. D. Hauser , and B. E. Russ . 2006 . Spontaneous process-
ing of abstract categorical information in the ventrolateral prefrontal
cortex . Biology Letters 2 : 261 5.
Ghazanfar , A. A ., J. I. Flombaum , C. T. Miller , and M. D. Hauser . 2001 .
Te units of perception in the antiphonal calling behavior of cotton-
top tamarins (Saguinus oedipus): Playback experiments with long
calls. J Comp Physiol [A] 187. 1: 27 35.
Ghazanfar , A. A ., D. Smith-Rohrberg , and M. D. Hauser . 2001 . Te role
of temporal cues in rhesus monkey vocal recognition: Orienting asym-
metries to reversed calls . Brain Behav Evol 58 : 163 72.
Ghazanfar , A. A ., D. Smith-Rohrberg , A. A. Pollen , and M. D. Hauser .
2002 . Temporal cues in the antiphonal long-calling behaviour of cot-
tontop tamarins . Animal Behavior 64 : 427 38.
Giford , III, G. W ., M. D. Hauser , and Y. E. Cohen . 2003 . Discrimination
of functionally referential calls by laboratory-housed rhesus
macaques: Implications for neuroethological studies. Brain Behav
Evol 61 : 213 24.
Giford , III, G. W ., K. A. MacLean , M. D. Hauser , and Y. E. Cohen .
2005 . Te neurophysiology of functionally meaningful catego-
ries: Macaque ventrolateral prefrontal cortex plays a critical role in
spontaneous categorization of species-specifc vocalizations . J Cogn
Neurosci 17 : 1471 82.
Hackett , T. A ., T. M. Preuss , and J. H. Kaas . 2001 . Architectonic identif-
cation of the core region in auditory cortex of macaques, chimpanzees,
and humans . J Comp Neurol 441. 3: 197 222.
Hauser , M. D . 1998 . Functional referents and acoustic similar-
ity: Field playback experiments with rhesus monkeys . Anim Behav
55. 6: 1647 58.
Hauser , M. D. , B. Agnetta , and C. Perez . 1998 . Orientation asymmetries
in rhesus monkeys: Efect of time-domain changes on acoustic percep-
tion . Anim Behav 56 : 41 7.
Kaas , J. H. , and T. A. Hackett . 1998 . Subdivisions of auditory cortex
and levels of processing in primates . Audiology and Neuro-otology
3. 2/3: 73 85.
Lu , T ., L. Liang , and X. Wang . 2001 . Neural representations of tempo-
rally asymmetric stimuli in the auditory cortex of awake primates.
J Neurophysiol 85. 6: 2364 80.
Marler , P . 1968 . Aggregation and dispersal: Two functions in primate
communication. In Primates: Studies in Adaptation and Variability ,
ed. P. C. Jay . New York : Holt, Rinehart, and Winston .
Mitani , J. C . and Marler , P . 1989 . A phonological analysis of male gibbon
singing behavior . Behaviour 109 : 20 45.
Rauschecker , J. P ., and B. Tian . 2004 . Processing of band-passed noise in
the lateral auditory belt cortex of the rhesus monkey. J Neurophysiol
91. 6: 2578 89.
Robinson , J. G . 1979 . An analysis of vocal communication in the
titi monkey Callicebus moloch . Zeitschrift fur Tierpsychologie
49 : 46 79.
Primate Vocal Communication Priming, Semantic
665
connection occurs when a textual co-text is used as prime and
the completion or continuation of the co-text as target. So its
time to go will accelerate recognition of home , but inhibit recog-
nition of feet . It is interesting to note that even where the target
is semantically unpredictable, the co-text prime will still accel-
erate, rather than inhibit, recognition of the target so long as
it is syntactically predictable from the co-text (as in its time to
go hang-gliding ) (Wright and Garrett 1984 ; West and Stanovich
1986 )
Fairly obviously (but importantly), if people are primed
with a particular word, then they will recognize it more quickly
when it occurs again as target; this is known as repetition prim-
ing. (So dog accelerates the processing of a second instance
of dog. ) Repetition priming afects both accuracy of response
(Jacoby and Dallas 1981 ) and speed of response (Scarborough,
Cortese, and Scarborough 1977 ). Te priming may be over very
short intervals (tiny fractions of a second) or long intervals
(minutes, hours, or even days), and many believe that long-
term repetition priming is explicable in terms of quite diferent
mechanisms from short-term priming. Te way that repeti-
tion priming works has been disputed (Jacoby 1983 ; Tulving
and Schacter 1990 ), but it can be assumed to be a key factor
in priming for cohesion (see following discussion) particularly
as repetition priming has been shown to last for several hours.
Long-term repetition priming is often referred to as implicit
memory .
We have seen that psycholinguists have interested them-
selves in the way a prime may accelerate or retard the processing
of a target. Te target itself must, however, be assumed to have a
preexistent relationship with the prime in advance of the partic-
ular priming efect. Otherwise, the relationship would have to be
created at the time of processing the target and the efect would
presumably be one of retardation. Tis prior relationship is inves-
tigated by Michael Hoey ( 2005 ) using corpus-linguistic rather
than psycholinguistic methodology; he assumes the relationship
to have been created by a type of repetition priming, such that
repeated encounters of the same items within the same envi-
ronment result in the creation of an association between them.
Hoey terms these relationships lexical primings and uses them to
account for a wide range of linguistic phenomena. Although he
does not explicitly relate his model to the claims of connection-
ism, there are clear points of parallel.
As noted, lexical priming draws on a diferent evidential base
from the semantic (and repetition) priming research described
so far, drawing on corpus-linguistic evidence to demonstrate
the probability of particular psycholinguistic associations. Te
frst of these types of associations, and perhaps the most funda-
mental, is that of collocation (Sinclair 1991 ). A collocation such
as dog and bark is created for a speaker whenever each word is
semantically primed by the other as a result of repetition priming
of word combinations, such as the dog barked. Te implication is
that repetition priming is primary both because it is long lasting
and because its existence accounts for (some) semantic priming.
From these primings, semantic associations (or preferences)
(Hoey 2005 ; Sinclair 2004 ), such as that of bark with spaniel-Alsa-
tian-poodle-Labrador (etc.), are created. Members of the set may
be stored close to each other, but their presence in the set is a
result initially of the repetition priming of one or more members
complex ways by whether the prime and target are low frequency
or high frequency (or combinations of these).
Diferent kinds of priming have diferent kinds of efect. If the
prime is formally similar but semantically unrelated to the target
(e.g., from and frog ), the prime inhibits recognition of the target,
presumably because one is momentarily mistaken for the other
in the recognition process (Colombo 1986 ). Where the prime
and the target are drawn from the same semantic set (e.g., dog
and cat ) or are otherwise related semantically, as in the dog-bark
example, the target is retrieved more quickly. Tis phenomenon
was frst demonstrated (though not frst noticed) by D. E. Meyer
and R. W. Schvaneveldt ( 1971 ) and has become a cornerstone of
psycholinguistic methodology.
Semantic priming automatically accelerates the recogni-
tion process if the time between sight of the prime and sight
of the target is short, but if the gap is greater, other factors
may impact. In particular, even if the prime and the target
are semantically related, there may be inhibition rather than
acceleration if the informant has been led to expect some-
thing else to occur (Neely 1977 ), though this inhibitory efect
does not occur if there is a very short time between prime and
target display (technically known as the stimulus-onset asyn-
chrony , or SOA) . Tis leads one to conclude that there are two
types of priming, one of which is automatic and short-lived
and the other of which attentional priming is longer last-
ing and more available for conscious inspection. Te former
can be interfered with but will persist in the speaker on sub-
sequent occasions; the latter can be created and equally can
disappear . So in an infuential experiment, Neely ( 1977 ) told
informants that whenever they were primed with body , they
should expect a word associated with buildings to be the tar-
get. He then sometimes gave a word associated with the body
as target instead. He found that with an extremely short SOA,
both heart and door were processed equally quickly after use
of body as a prime. With a slightly longer SOA, however, the
processing of the unexpected heart actually took longer than
the (in the circumstances) expected door. Outside the context
of the experiment, though, there are no reasons to suppose
that the body-heart connection would be afected, and equally
it seems unlikely that the attentional priming would persist
much beyond the end of the experiment.
Successful processing has an efect here, too, and further
confrms the existence of attentional priming as a separate kind
of priming from automatic priming. If informants fnd that the
primes they are given relate regularly and reliably to the targets,
the efect of the priming gets increasingly strong. If, on the other
hand, they fnd little connection between the primes and the tar-
gets, the priming efect gets weaker (Den Heyer 1985 )
It is important to note that priming of either the automatic
or attentional kind presupposes a causeefect relationship: Te
prime afects the target. However, it is implicit in this causal rela-
tionship that the word that is the target has some prior mental
connection with the word (or other linguistic, or indeed nonlin-
guistic, entity) that is chosen as prime, and it is this prior con-
nection that the prime activates, whether the connection takes
the form of the words being stored near to each other because of
their semantic or pragmatic proximity or because they regularly
co-occur for some reason. An exploitation of the latter kind of
Priming, Semantic
666
Hoey , Michael . 2005 . Lexical Priming: A New Teory of Words and
Language . London : Routledge. Te key work on lexical priming.
Jacoby , Larry L . 1983. Perceptual enhancement: Persistent efects of an
experience . Journal of Experimental Psychology: Learning, Memory
and Cognition , 15 : 930 40.
Jacoby , L. L ., and M. Dallas . 1981 . On the relationship between autobio-
graphical memory and perceptual learning . Journal of Experimental
Psychology: General 110 : 306 40.
Meyer , D. E ., and R. W. Schvaneveldt . 1971 . Facilitation in recognising
pairs of words: Evidence of a dependence between retrieval opera-
tions . Journal of Experimental Psychology 90 : 227 35.
Neely , J . 1977 . Semantic priming and retrieval from lexical mem-
ory: Roles of inhibitionless spreading activation and limited capacity
attention. Journal of Experimental Psychology: General 106 : 226 54.
. 1991 . Semantic priming efects in visual word recognition: A
selective review of current fndings and theories. In Basic Processes in
Reading: Visual Word Recognition , ed. D. Besner and G. Humphreys ,
264336. Hillsdale, NJ : Erlbaum . An extensive review of semantic prim-
ing from a word recognition perspective.
Rumelhart , D. E ., and D. A. Norman . 1985 . Representations of knowl-
edge. In Issues in Cognitive Modeling , ed. A. M. Aitkenhead and J. M.
Slack , 1562. Hillsdale, NJ : Lawrence Erlbaum .
Scarborough , D. L. , C. Cortese , and H. S. Scarborough . 1977 . Frequency
and repetition efects in lexical memory. Journal of Experimental
Psychology: Human Perception and Performance 3 : 1 17.
Sinclair , John . 1991. Corpus, Concordance, Collocation . Oxford : Oxford
University Press .
. 2004 . Trust the Text: Language, Corpus and Discourse .
London: Routledge .
Tulving , E ., and D. L. Schacter . 1990 . Priming and human memory sys-
tems . Science 247 : 301 6.
West , R. F ., and K. E. Stanovich . 1986 . Robust efects of syntactic struc-
ture on visual word processing. Memory and Cognition 14 : 104 12.
Wright , B ., and M. Garrett . 1984. Lexical decision in sentences: Efects of
syntactic s tructure Memory and Cognition 12 : 31 45.
PRINCIPLES AND PARAMETERS THEORY
The Framework
Principles and parameters (P&P) theory has been the prevail-
ing approach to natural language syntax within transfor-
mational grammar and generative grammar since the
beginning of the 1980s. According to the P&P theory, the initial,
innate state of the human faculty of language FL
0
is characterized
as a fnite set of general principles complemented by a fnite set
of variable options, called parameters . Tese principles and
parameters together constitute universal grammar (UG),
a model of FL
0
. FL
0
functions as a language acquisition
device : It imposes severe constraints on attainable languages,
thereby facilitating the process of language acquisition, the core
of which lies in fxing the open parameter values of FL
0
. On this
view, competence in a given language is the result of a par-
ticular specifcation of the parameters of FL
0
(called parameter-
setting ), which determine the range of possible variation among
languages.
Interpreted broadly, the P&P framework can be seen as
a general model of the interaction of nature and nurture
(genetic endowment and experience) in the development of any
module of human cognition. Accordingly, it has come to be
applied beyond syntax both inside and outside of linguistics. An
example of the former case is the theory of phonology called
of the set in conjunction with bark. Tus, the mind stores sev-
eral instances of the Alsatian barked and of the poodle barked , as
well of course the dog barked , and from this creates an associa-
tion of bark with all types of dog. Tis set remains sound until
conficting evidence is encountered. So Chihuahua may never
be encountered with barked but with yelped. Te confict persists
until the speaker modifes the original priming or treats the new
priming as either an exception or an anomaly. So, for instance,
bark might be placed in a semantic set of nonverbal noises, which
would include growl as well as bark and yelp. According to Hoey,
the same processes result in the establishment for the speaker
of quasi-grammatical relations associated with the lexical item
( colligational relations ; Hoey 2005 ; Sinclair 2004 ), and grammar
is argued to be an output from the primings, rather than having
an existence independent of them .
Hoey (2005) also claims that speakers are primed to associate
words with textual position and cohesive patterning. All prim-
ings are assumed to be genre and domain specifc, and this is
particularly noticeable of the textual primings. Tus, for exam-
ple, British newspaper readers are primed by British news stories
to associate yesterday with text-initial sentences and to associate
Mr with frst-word position in paragraph-initial sentences. Te
words Blair and Bush are primed for most readers to be followed
cohesively by pro-forms ( he , his ), while Pluto is primed to be fol-
lowed by co-hyponyms ( Neptune, Saturn, etc.).
Te term priming is used by Hoey to describe both the pro-
cess and the product of that process. Te association created
by priming may itself be subject to priming (which Hoey terms
nesting ). Tus, we are primed to collocate Bush with George , and
then to collocate the nested pair George Bush (with or without W )
with President. As a further step, we are then primed to associate
(President) (George)(W) Bush as a combination with pronominal
cohesion. No assumptions are made, however, about the order
in which the primings may occur. Primings necessarily vary from
individual to individual.
Te accumulation of collocational, colligational, and seman-
tic relationships may explain linguistically the ef cacy of seman-
tic primings, though experimental evidence has not yet been
ofered in support of this claim. Te phenomenon of semantic
priming would appear to have no efect in giving rise to the lexi-
cal primings as described by Hoey, but the existence of seman-
tic priming is confrmation of the ef cacy of the prior lexical
primings .
Michael Hoey
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Colombo , Lucia . 1986. Activation and inhibition with orthographically
similar words . Journal of Experimental Psychology: Human Perception
and Performance 12 : 22634.
Den Heyer , Ken . 1985. On the nature of the proportion efect in semantic
priming . Acta Psychologica 60 : 25 38.
Forster , Kenneth I. 1976 . Accessing the mental lexicon. In New
Approaches to Language Mechanisms , ed. R. J. Wales and E. Walker ,
25787. Amsterdam : North-Holland .
Harley , Trevor . 2001 . Te Psychology of Language: From Data to Teory.
2d ed. Hove and New York: Psychology Press . Although the informa-
tion on priming is spread around the text, this is an ideal and accessible
introduction to the diferent kinds of priming.
Priming, Semantic Principles and Parameters Theory
667
the lexically specifed theta-roles of a predicate and its argu-
ment expressions in the syntactic representation. As for struc-
tures derived by transformations, movement rules are reduced
to a single and maximally general operation Move that can
move anything anywhere. A representational filter that lim-
its the application of Move is the empty category principle
(ECP), which demands that traces of movement be licensed
under a local structural relation called government . Apart from
the ECP and the bounding theory , which places an upper
bound on how far movement can take an element, various other
modules of UG, not narrowly geared to cut down the overgen-
eration of structures resulting from Move , act to flter the out-
put representations produced by movements. case theory
requires that (phonetically overt) noun phrases (NPs) occupy a
position at surface structure where they are assigned a case. Te
three principles of binding theory (which constrain the distri-
bution of anaphors, personal pronouns, and referential NPs,
respectively, relative to potential antecedents they can/cannot
be coreferential with) are sensitive to the binary [ anaphoric]
and [ pronominal] features of NP categories generally, includ-
ing phonetically empty NPs like various types of traces and null
pronouns.
Te modular organization itself, that is, the dissociation of
various aspects of syntactic phenomena for the purposes of the
grammar, is what makes it possible to keep principles of UG max-
imally simple. Te cohesion of each module is supplied by some
notion and/or formal relation on which its principles are cen-
tered. Te whole of the grammatical system is also characterized
by unifying concepts, most notably the notion of government,
which plays a key role in a variety of modules. Te components
interact in complex ways to restrict the massive overgeneration
of syntactic expressions that would otherwise result from the
fundamental freedom of possible basic phrase structures and
transformations applied to them, which ultimately yields the
actual set of well-formed expressions.
Te modularity of the diferent (sets of) principles is due
not only to the dissociation of the properties relevant to them
but also to the stipulation of distinctions with regard to where
in the grammar they apply. According to GB theory, each sen-
tence corresponds to a sequence of representations, starting
from D-structure (or deep structure , DS), proceeding through
S-structure (or surface structure , SS) to the fnal representation
called logical form (LF), where adjacent representations are
related by transformations. Te derivation from DS to SS feeds
phonetic realization, in particular the mapping from SS to pho-
netic form (PF) (it is overt ), whereas the derivation from SS to LF
does not (it is covert ). A principle can apply to transformations
(like bounding theory), or to one or more of the three syntac-
tic representational levels DS, SS, and LF (these constraints are
flters), though not to any intermediate representation. Figure 1
depicts this so-called Y- or T-model of GB, tagged to indicate
where the most prominent modules apply .
Parameters
UG, as a model of language competence, includes the principles
along with the locus of their application, as well as the primi-
tive syntactic objects (e.g., labels distinguishing full phrases,
heads of phrases, and intermediate-level categories), relations
Government Phonology (see Kaye 1989 ), and an instance of the
latter is a recently emerging principles and parametersbased
approach to moral psychology (see Hauser 2006 , and references
therein). In the domain of natural language syntax, the P&P
framework subsumes both government and binding (GB)
theory and its more recent development, called the minimal-
ist program , or linguistic minimalism , even though the term is
often used narrowly to refer to the former model only .
From Rules to Principles
Te P&P framework crystallized by the end of the 1970s as
a way to resolve the tension between two goals of generative
grammar. One objective was to construct descriptively ade-
quate grammars of individual languages (see descriptive,
observational, and explanatory adequacy ). Another
was to address the logical problem of language acquisition
(see innateness and innatism ) by working out a theory of
UG that constrains possible grammars to a suf ciently narrow
range, so that the determination of the grammar of the language
being acquired from the primary linguistic data can become
realistic (this is referred to as explanatory adequacy ). Te two
goals clearly pull in opposing directions: Te former seems
to call for allowing complex rules and a considerable degree
of variation across grammars (a liberal UG), while the latter
requires that possible grammatical rules be as constrained as
possible (a restrictive UG).
Te research program that culminated in P&P theory aimed
to approximate these twin goals by establishing in the ways in
which grammatical rules can and should be restricted, extract-
ing from them properties that seemed to be stable across con-
structions and languages, and formulating them as constraints
imposed by UG on the format of rules of individual grammars.
Uncovering, generalizing, and unifying such constraints elimi-
nated from rules the general conditions on their operation,
which made it possible for rules themselves to be considerably
simplifed. For instance, the transformational rule that forms
wh-interrogatives, the rule of relativization producing relative
clauses, the rule of topicalization , and several others, each
corresponding roughly to some construction recognized by
traditional grammars, share certain notable properties. Noam
Chomsky ( 1977 ) argued that instead of stating such properties
as part of each of these rules, some of them should be incorpo-
rated into UG, while others should be ascribed to the generalized
rule dubbed front-wh , of which each of the individual rules is an
instantiation. Te furthest such a factoring out and unifcation
strategy can potentially lead to is a model of language where
rules (as well as the corresponding constructions of traditional
grammar) are eliminated altogether from the theory as epiphe-
nomena deducible from the complex interaction of the general
principles of UG. Tis is precisely the approach that the P&P
framework has been pursuing .
Modularity
In the government and binding model of the P&P approach
(Chomsky 1981 ), principles of UG are organized into modules,
or subtheories. Such modules include x-bar theory , which
constrains possible phrase structure confgurations, and
theta theory , which determines a bi-unique mapping between
Principles and Parameters Theory
668
which parameters difer concerns the number of options, that
is, parameter settings that are allowed. Most parameters are
binary, but proposals have been made for parameters with more
options: for instance, the choice of the local domain in which
anaphors must fnd an appropriate antecedent. Binary param-
eters include the choice of the timing of a movement trans-
formation with respect to S-structure (either overt or covert; see
Figure 1). Finally, while some parameters are simply underspeci-
fed aspects of UG principles, others are grammatical properties
of (classes) of lexical items. Te head directionality parameter
(set as head-initial for English, where verbs, nouns, adjectives,
and adpositions precede their complements, and head-fnal for
Japanese, where they follow them) belongs to the frst of these
two types, while variation in terms of the lexical items that are
lexically [+anaphoric] exemplify the second .
The Shift to Minimalism
Te P&P framework inspired a vast amount of research on simi-
larities and diferences across languages, as well as on language
acquisition (see principles and parameters theory and
language acquisition , and syntax, acquisition of ),
which has produced an impressive array of novel discover-
ies and analyses that are both attractively elaborate in terms of
data coverage and at the same time genuinely illuminating as
regards the explanations they ofer. Tat said, in pursuit of the
twin objectives of descriptive and explanatory adequacy, some of
the basic notions and principles became increasingly non-nat-
ural and complex (like government and the ECP, or the notion
of local domain in binding theory). Tis gave cause for growing
concern in the feld, in no small part because the question of why
UG is the way it is became disappointingly elusive. Te ultimate
source of the emergent complexities, beyond the strive for ever-
improving empirical coverage, was the fact that GB lacked an
actual theory of possible principles or, for that matter, of possible
parameters. As for the latter, continued in-depth research on
cross-linguistic variation has shown many of the macro-param-
eters, among them the null subject parameter, to be unsustain-
able in the strong form they were originally proposed: Several of
the linguistic properties correlated by macro-parameters turned
out to be cross-linguistically dissociable. Even though the idea
of parametric linguistic variation was upheld, parameters them-
selves needed to be scaled down. In addition, as GB relied on
massive overgeneration resulting from the fundamental free-
dom of basic phrase structure and transformations, downsized
by declarative constraints imposed (mainly) on syntactic repre-
sentations, the computational viability of the model was often
called into question.
Te current minimalist research program (MP), initiated by
Chomsky in the early 1990s (see Chomsky 1995 ), while building
(e.g., c-command , dominance, government), and operations
(e.g., movement, deletion) that collectively defne the syntactic
expressions. Cross-linguistic variation, according to GB theory,
is rather limited. An obvious element of variation involves the
identity and properties of lexical items (referred to collectively
as the mental lexicon ). Apart from acquiring a lexicon, the
primary means of grammar acquisition and the key source of
cross-linguistic diferences is the inference of underspecifed
aspects of UG principles, that is, the setting of open parameters.
Parametric principles are an innovation to allow the model
to furnish descriptively adequate because suitably difer-
ent grammars for individual languages. To provide a realistic
account of language acquisition, a process that is fairly uniform
and remarkably efective both across speakers and across lan-
guages, the number of parameters to be fxed must be reason-
ably low, the parameter values permitted by UG must be limited
to relatively few, and the cues in the primary linguistic data that
can trigger their values must be suf ciently easy to detect. Due to
their rich deductive structure, a distinct advantage of parameter-
ized principles over language- and construction-specifc rules
is that the setting of a single parameter can potentially account
for a whole cluster of syntactic properties, thereby contributing
to a plausible explanation for the outstanding ef ciency of the
process of language acquisition itself. Such parameters are often
referred to as macro-parameters .
A canonical macro-parameter of GB theory is the so-called
pro -drop (or null subject ) parameter. Null subject languages
like Italian or Greek, in contrast to non- pro -drop languages like
English or Dutch, allow phonetically null pronominal subjects
(designated as pro ) in tensed clauses, have no overt pleonastic
element flling the subject position of weather verbs (cf. It rained ),
exhibit free subject inversion to the right of the verb, and permit
movement of a subject out of an embedded wh-clause and from
a position following a lexical complementizer (cf. * Who
i
do you
think that t
i
will win? ). Te classic account of this cluster of prop-
erties ascribes them to a single parameter, namely, whether or
not the fnite verbal agreement infection syntactically governs
the preverbal subject position, which in turn is related to the
morphological richness of the relevant conjugation paradigm. It
is due to the positive setting of this (ultimately lexical) parameter
that null subject languages license a phonetically null pronoun
or a trace in the canonical subject position of tensed clauses rel-
atively freely.
Parameters range from macro-parameters like the null sub-
ject parameter to micro-parameters whose scope is compara-
tively narrow, for instance, the parameter determining whether
or not the (fnite) main verb raises out of the verb phrase (VP)
before S-structure to a position above VP adverbs or clausal
negation (the verb raising parameter ). Another dimension along
DS SS LF
Bounding Theory
Case Theory
Theta Theory
PF
ECP
Binding Theory X- bar Theory
Lexicon
Theta Theory
Theta Theory
overt transformations covert transformations
Figure 1.
Principles and Parameters Theory
669
(applying them as soon as their respective trigger is merged in)
and that has multiple transfers. Derivational sequences between
two transfer points are called phases . Te basic architecture is
shown in Figure 2 .
Finally, grammatical components are reduced as well. First of
all, there are no distinct phrase structure and transformational
components, as both basic phrase structure and movements are
brought about by the operation merge: While basic structure
building involves merging two distinct elements, movement
involves (re)merging an element with a constituent that con-
tains it. In addition, the burden of description carried by mod-
ules of GB is partly reallocated to syntax-external components,
and is partly redistributed among the residual factors that can
enter syntactic explanation: the principal constraint imposed by
the interface components (full interpretation), the character of
the syntactic derivation (multiple transfers, principles of com-
putational economy, the nature of basic syntactic operations,
etc.), and the properties of lexical items. For instance, much of
the binding theory of UG is reduced to movement operations
and rules of interpretation, case theory is recast in more general
terms and is subsumed in a broader account of triggers for move-
ments (called checking theory ), and bounding theory is deduced
from the multiple transfers nature of the derivation .
Conclusion
Te fundamental question pursued by the P&P framework is
whether it is possible to construct an explanatorily adequate
theory of natural language grammar based on general prin-
ciples. Two further ambitions of P&P, gaining prominence with
the advent of its minimalist research program, are to fnd out
whether the primitive notions and principles of such a model are
characterized by a certain degree of naturalness, simplicity, and
nonredundancy, and concurrently, whether some properties of
the language faculty can be explained in terms of design con-
siderations pertaining to computational cognitive subsystems in
general, or even more broadly, in terms of laws of nature. Should
it turn out that these questions are answered in the af rmative (as
some initial results suggest), that would be a surprising empiri-
cal discovery about an apparently complex biological subsystem
(cf. biolinguistics ): in the case at hand, the human language
faculty. Te exploration of the ways in which general laws of
nature might enter linguistic explanation has barely begun.
Clearly, most of the work lies ahead .
Balzs Surnyi
on the achievements of GB theory, departs from it in various
important ways. It refocuses attention on the shape of UG itself
as a model of the innate faculty of language (FL), a computation-
al-representational module of human cognition, as well as on
the way it interfaces with (articulatory-phonetic and conceptual-
intentional) external systems. Te MP adopts the substantive
hypothesis (called full interpretation ) that representations that
the FL feeds to the external interface systems are fully interpre-
table by those components, with all uninterpretable aspects of
the representations eliminated internally to FL. As for the shape
of UG as a computational system, the MP puts forward the sub-
stantive hypothesis that FL is computationally ef cient: It incurs
minimal operational complexity in the construction of repre-
sentations fully interpretable by the interface systems. Syntactic
operations like movement apply only if they are triggered: only
if they must be carried out in order to satisfy full interpretation
by eliminating some uninterpretable property in the syntactic
expression under computation (a principle of computational
economy called last resort ). If there is more than one way that a
derivation can satisfy full interpretation, the least complex (set
of) operation(s) is selected by FL (the principle of least efort ).
On the methodological side, the MP proposes to apply
Ockhams razor considerations of theoretical parsimony to UG
as rigorously as possible. All syntax-internal principles con-
straining representations are disposed of, thereby eliminating
syntax-internal representational levels , including S-structure
and D-structure. Te incremental structure-building operation
merge starts out from lexical items, combining them recursively
into successively larger syntactic units. Empirical properties for-
merly captured at D-structure and S-structure are accounted for
by shifting the burden of explanation to full interpretation at the
interface levels of PF and LF, and to principles of economy of
derivation, the only principles operational in UG. Economy prin-
ciples have no built-in parameters: All parametric diferences
across languages are confned to the domain of lexical proper-
ties, an irreducible locus of variation, to which, accordingly, the
acquisition of syntax is reduced (cf. the lexical learning
hypothesis ). For instance, word order variation, previously
put down to the head directionality parameter, is typically attrib-
uted to movement operations: Movements can occur either in
overt or in covert syntax, and they can afect smaller or larger
units of structure, these choices being a function of uninterpre-
table lexical properties of participating elements.
Non-naturally complex notions and relations (including
government) are also eliminated from UG. A syntactic expres-
sion is taken to be a plain set (of sets of sets, etc.) of lexical items,
produced by recursive applications of merge: Nothing beyond
that is added in the course of the derivation. It follows from this
simplifying proposal (called inclusiveness ) that syntactic expres-
sions include no indices (to link a moved element to its trace, or
a binder to its bindee), no traces (but silent copies of the moved
elements themselves), no syntactic label for phrase or head
status, and perhaps no labels borne by complex syntactic units at
all. Two stipulative assumptions of the GB model that all overt
movements precede all covert movements and that transfer to
phonetic and conceptual interpretation can only take place at
a unique point in the derivation are also dropped. Tis yields
a model that has overt and covert movements intermingled
Figure 2.
Phonetic interpretation
Lexicon
Conceptual interpretation
principles of computational economy
Full Interpretation
Full Interpretation
Principles and Parameters Theory
670
P&P is that the options for grammatical variation are extremely
limited. On the P&P approach, the childs task during language
acquisition is akin to ordering food in a restaurant: One need
only make selections from a menu, not give the chef a recipe.
In other words, the information required for the child to
select an appropriate grammar from among the options is far
less, both in quantity and in quality, than would be required to
build a grammar from the ground up. First, grammars that can-
not be attained with the available parameter settings will never
be hypothesized by the child, even if they are compatible with
the childs linguistic input up to that point. Second, to the extent
that parameters are abstract, and thus have wide-spread con-
sequences, a variety of diferent sentence types in the linguistic
input can help the child select the correct option. Te challenge
of identifying the correct grammar is still considerable, but is far
more tractable than it would be if the child had to rely on general
learning strategies alone.
Investigating Language and Its Acquisition Within a P&P
Framework
Te P&P framework was frst clearly articulated for syntax, in the
context of government and binding theory (e.g., Chomsky
1981 , 1986 ). Yet the framework is considerably more general.
First, the same basic architecture has been applied to phonology,
notably in the framework of government phonology (e.g., Kaye,
Lowenstamm, and Vergnaud 1990 ), and also (in certain work) to
semantics and morphology. Second, recent syntactic and pho-
nological research in the minimalist program (Chomsky 1995 ,
2001 ; see minimalism ) and in optimality theory (Prince
and Smolensky 2004 ) still crucially assumes a P&P framework, in
the broad sense that it posits universal principles and narrowly
restricted options for cross-linguistic variation. (Tis point is dis-
cussed further in the next section.)
Within the P&P framework, research on childrens acquisi-
tion of language plays a number of important roles. First, such
research can clarify the logical problem of language acquisi-
tion , which any explanatorily adequate linguistic theory must
address: How in principle can the correct grammar be chosen
from among the proposed options, using only the types of lin-
guistic input that children actually need for successful lan-
guage acquisition? (See descriptive, observational, and
explanatory adequacy .) Acquisition research can help
determine which types of linguistic input are (and are not), in
fact, necessary for children to succeed at language acquisition.
For example, some of the most compelling evidence
for the irrelevance of corrective feedback comes from Eric
H. Lennebergs ( 1967 , 3059) study of a hypolingual child. Despite
the fact that the child had been mute since birth, and therefore
had had no possibility of producing any errors to be corrected, he
performed at an age-appropriate level on comprehension tests of
English grammar. Hence, receiving corrective feedback on ones
own utterances seems to be unnecessary. Hearing the linguistic
utterances of other speakers, produced in context, can suf ce. To
achieve explanatory adequacy, a linguistic theory must be able
to account for this.
A second role of acquisitional evidence within the P&P frame-
work lies in testing the acquisitional predictions of proposed
linguistic principles. All else being equal, if one proposes that
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baker , Mark C. 2001 . Te Atoms of Language: Te Minds Hidden Rules of
Grammar . New York : Basic Books .
Chomsky , Noam . 1977 . On wh-movement. In Formal Syntax , ed.
Peter Culicover , Tom Wasow , and Adrian Akmajian , 71132. New
York : Academic Press .
. 1981 . Lectures on Government and Binding . Dordrecht, the
Netherlands : Foris .
. 1995 . Te Minimalist Program . Cambridge, MA : MIT Press .
Hauser , Marc D . 2006 . Moral Minds: How Nature Designed Our Universal
Sense of Right and Wrong. New York : Ecco/Harper Collins .
Kaye , Jonathan . 1989 . Phonology: A Cognitive View . Hillsdale,
NJ : Lawrence Erlbaum .
Lasnik , Howard , and Juan Uriagereka , with Cedric Boeckx . 2005 . A Course
in Minimalist Syntax: Foundations and Prospects. Oxford : Blackwell.
PRINCIPLES AND PARAMETERS THEORY AND
LANGUAGE ACQUISITION
Nativism
Te basic idea in principles and parameters theory is to
distinguish the invariants of human language (the principles )
from the major points of cross-linguistic variation (the parame-
ters ). Both principles and parameters are taken to refect innately
determined, biological characteristics of the human brain (see
universal grammar ). In the course of normal child develop-
ment, however, the two diverge: Te principles come to operate
in much the same way in every child, with minimal sensitivity to
the childs environment, while the parameters take on distinct
values as a function of the childs linguistic input.
Te term parameter is normally reserved for points of nar-
rowly restricted variation. Te principles and parameters (P&P)
framework also acknowledges that languages vary in ways that
are relatively unconstrained by universal grammar, such as the
exact form of vocabulary items. Tese latter points of variation
are usually treated as arbitrary idiosyncrasies, to be listed in the
lexicon .
Te P&P framework has its origins in the two foundational
questions of modern linguistics (Chomsky 1981 ): What exactly
do you know, when you know your native language? And how
did you come to know it? A satisfactory answer to these questions
must address the poverty of the stimulus , including the fact that
children are not reliably corrected when they make a grammati-
cal error (Brown and Hanlon 1970 ; Marcus 1993 ) .
Despite the poverty of the stimulus, by the age of about fve
years we observe uniformity of success at language acquisi-
tion (Crain and Lillo-Martin 1999 ): Aside from cases of medical
abnormality, or isolation from natural-language input, every
child acquires a grammar that closely resembles the grammar of
his or her caregivers. Moreover, even when a child is younger,
and still engaged in the process of language acquisition, extraor-
dinarily few of the logically possible errors are actually observed
in the childs spontaneous speech (Snyder 2007 ). Clearly, chil-
dren do not acquire grammar through simple trial-and-error
learning.
Linguists working in the P&P tradition have concluded that
a great deal of grammatical information must already be pres-
ent in the childs brain at birth. Of course, diferent languages of
the world exhibit somewhat diferent grammars, but the claim in
Principles and Parameters Theory and Language Acquisition
671
a child acquire the verb-particle construction signifcantly ear-
lier than compounding. Tis prediction received strong support
from a longitudinal study of 10 children .
Tis example illustrates how the investigation of language
acquisition and the investigation of mature grammars can
be mutually reinforcing activities within the P&P framework.
Another example is provided by the work of Diane Lillo-Martin
and Ronice Mller de Quadros ( 2005 ), who considered the para-
metric prerequisites for the diferent types of wh-questions in
American Sign Language (ASL), according to two competing
syntactic analyses. Te two analyses yielded distinct predictions
about the time course of acquisition, which were then success-
fully tested against longitudinal data from children acquiring
ASL .
Areas of Debate
We mention here two areas of debate within the P&P approach
to child language acquisition, and of course there are others.
1) What types of parameters, exactly, is the child required to set?
2) What are the observable consequences of an unset or misset
parameter?
One point of disagreement in the P&P literature quite gen-
erally, including the acquisition literature, concerns the proper
conception of parameters. A classic conception, which Noam
Chomsky ( 1986 , 146) attributes to James Higginbotham, is the
switchbox metaphor: Each parameter is like an electrical switch,
with a small number of possible settings.
Yet this is only one of many possible ways that parameters
could work. A radically diferent conception is found in optimal-
ity theory, which posits a universal set of violable constraints.
Instead of choosing particular settings for switches in a switch-
box, the learner has to rank the constraints correctly. Te result
is a narrowly restricted set of options for the target grammar,
as required by the P&P framework. (Indeed, on the mathemat-
ical equivalence of a constraint ranking to a set of switchbox-
style dominance parameters, see Tesar and Smolensky 2005 ,
456.)
Still another approach to parameters is to connect them to the
lexicon. (See lexical learning hypothesis .) Tis is concep-
tually attractive because the lexicon is independently needed as
a repository of information that varies across languages. Exactly
what it means to connect parameters to the lexicon, however,
has been open to interpretation.
One idea is to connect points of abstract grammatical (e.g.,
syntactic) variation to the paradigms of infectional morphology.
Te idea is that paradigmatic morphology has to be stored in the
lexicon anyway, and might provide a way to encode parametric
choices. Tis approach can be found in Borer ( 1984 ) and Lillo-
Martin ( 1991 ), for example. A related idea is to encode parametric
choices in the morphology of closed-class lexical items. A good
example is Picas ( 1984 ) proposal to derive cross-linguistic varia-
tion in the binding domain of a refexive pronoun from the pro-
nouns morphological shape. A variant of Pierre Picas approach
is to encode parametric choices as abstract (rather than morpho-
logically overt) properties of individual lexical items. Tis is the
lexical parameterization hypothesis of Wexler and Rita Manzini
( 1987 ), who took this approach to cross-linguistic variation in the
binding domain for both refexives and pronominals.
a given property of language is an innate principle of universal
grammar, then one expects the principle to be operative in chil-
dren as early as we can test for it. (A notable exception is found
in the work of Hagit Borer and Ken Wexler 1992 , who propose
that several specifc linguistic principles undergo maturational
change during childhood.)
For example, Stephen Crain and Mineharu Nakayama ( 1987 )
conducted an acquisitional test of structure dependence , the
proposed principle that syntactic movement is always sensitive
to hierarchical structure. Teir study tested the prediction that
structure dependence, as an innate principle, should be opera-
tive very early. Te study was conducted with children acquiring
English who were three to fve years old (the youngest subjects
capable of performing the experimental task), and used prompts
such as the following: Ask Jabba if [the man who is beating a
donkey] is mean. Crucially, children never produced errors of
the form, Is [the man who __ beating a donkey] is mean? Such
errors might have been expected, however, if the children had
been at liberty to hypothesize structure-independent rules (such
as Move the frst auxiliary to the beginning of the sentence) .
Tird, by proposing a parameter of universal grammar, one
makes predictions about the time course of child language acqui-
sition. Tese predictions may involve concurrent acquisition or
ordered acquisition. To see this, suppose that two grammatical
constructions A and B are proposed to have identical prerequi-
sites, in terms of parameter-settings and lexical information. A
and B are then predicted to become grammatically available to
any given child concurrently, that is, at the same point during
language acquisition.
A prediction of ordered acquisition results when the proposed
linguistic prerequisites for one construction (A) are a proper
subset of the prerequisites for another construction (B). In this
case A might become available to a given child earlier than B,
if the child frst acquires the subset of Bs prerequisites that are
necessary for A. Alternatively, A and B might become available
to the child concurrently, if the last-acquired prerequisite for B
is also a prerequisite for A. In contrast, no child should acquire B
signifcantly earlier than A.
As a concrete example, consider William Snyders ( 2001 )
work on the compounding parameter (TCP). Teoretical research
had suggested a link (at least in Dutch and Afrikaans) between
the verb-particle construction (cf. Mary lifted the box up ) and
morphological compounding (cf. banana box , for a box where
bananas are kept). Snyder observed a one-way implication in
the data from a sizable number of languages: If a language per-
mits the verb-particle construction, then it also allows free cre-
ation of novel compounds like banana box . Te implication is
unidirectional, however: Tere do exist languages that allow this
type of compounding, yet lack the verb-particle construction.
Snyder therefore proposed that the grammatical prerequisite for
the English type of compounding (i.e., the positive setting of TCP)
is one of several prerequisites for the verb-particle construction.
A clear acquisitional prediction followed: Any given child
acquiring English will either acquire compounding frst (if [+TCP]
is acquired prior to the other prerequisites for the verb-particle
construction), or acquire compounding and the verb-particle
construction at the same time (if [+TCP] is the last-acquired
prerequisite for the verb-particle construction). In no case will
Principles and Parameters Theory and Language Acquisition
672
An alternative view is that the child reserves judgment on any
given parameter setting until he or she has enough information
to set it with confdence. Initially the parameter is in an unset
state, but this time the consequence is that none of the grammat-
ical options tied to a specifc setting of the parameter is actually
endorsed by the child. Snyder ( 2007 ) advances this view when he
argues that children who are speaking spontaneously , in a natu-
ral setting, make astonishingly few of the logically possible gram-
matical errors. Te vast majority of the errors that do occur either
are errors of omission or belong to a tiny subset of the logical
possibilities for comission errors (where the words are actually
pronounced in confgurations that are ungrammatical in the tar-
get language).
Most of the grammatical comission errors that are found
in studies of elicited production or comprehension are absent
from childrens spontaneous speech, even when the oppor-
tunities exist for the child to make them. Snyder concludes
that many of these errors result from the demands of the
experimental tasks. When left to their own devices, children
successfully avoid putting words together in ways that would
require them to make a premature commitment to a particular
parameter-setting .
Conclusion
Language acquisition is a rich source of evidence about both the
principles and the parameters of the human language faculty.
For this reason, research on language acquisition plays a central
role in the P&P framework .
William Snyder and Diane Lillo-Martin
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Borer , Hagit . 1984 . Parametric Syntax: Case Studies in Semitic and
Romance Languages . Dordrecht, the Netherlands : Foris .
Borer , Hagit , and Ken Wexler . 1992 . Bi-unique relations and the matu-
ration of grammatical principles . Natural Language and Linguistic
Teory 10 : 147 89.
Brown , Roger , and Camille Hanlon . 1970 . Derivational complexity and
order of acquisition in child speech. In Cognition and the Development
of Language , ed. John R. Hayes , 155207. New York : Wiley .
Chomsky , Noam . 1981 . Lectures on Government and Binding . Dordrecht,
the Netherlands : Foris .
. 1986 . Knowledge of Language: Its Nature, Origin, and Use . New
York : Praeger .
. 1995 . Te Minimalist Program . Cambridge, MA : MIT Press .
. 2001 . Derivation by phase. In Ken Hale: A Life in Language , ed.
Michael Kenstowicz , 152. Cambridge, MA : MIT Press .
Crain , Stephen , and Diane Lillo-Martin . 1999 . Linguistic Teory and
Language Acquisition . Cambridge, MA : Blackwell .
Crain , Stephen , and Mineharu Nakayama . 1987 . Structure dependency
in grammar formation . Language 63 : 522 43
Gibson , Edward , and Kenneth Wexler . 1994 . Triggers . Linguistic Inquiry
25 : 355407.
Hyams , Nina . 1986 . Language Acquisition and the Teory of Parameters .
Dordrecht, the Netherlands : Reidel.
Kaye , Jonathan D. , Jean Lowenstamm , and Jean-Roger Vergnaud . 1990 .
Constituent structure and government in phonology . Phonology
7 : 193 231.
Lenneberg , Eric H . 1967 . Biological Foundations of Language . New
York : Wiley .
Yet another idea is to encode cross-linguistic grammatical
variation in the abstract (often phonetically null) features of
functional heads. Chomsky ( 1995 , Chapter 2 ) takes this approach
to V-raising in French, for example, and its absence in English: In
French, the functional head Agr
0
is strong, and causes the
verb to move up and adjoin to Agr
0
before the sentence is pro-
nounced. Te result is the word order in Jean [
AgrP
voit [
VP
souvent
[
VP
V
t
Marie ]]], literally John [
AgrP
sees [
VP
often [
VP
V
t
Mary]]], in
place of English John [
AgrP
[
VP
often [
VP
sees Mary]]].
Chomskys approach is lexical in the sense that the mor-
phosyntactic features of functional heads like Agr
0
are taken to
be listed in the lexicon. Note, however, that the possible fea-
tures of a functional head are still assumed to be quite narrowly
restricted. Tus, where earlier work might have posited a switch-
like parameter of [ verb raising], for example, Chomsky instead
posits a choice between a strong feature versus a weak feature on
Agr
0
, and assumes that this particular lexical item will be present
above the verb phrase (VP) in most or all cases. For purposes of
language acquisition, the diference is extremely minor; the child
makes a binary choice, and it has consequences across a wide
range of sentence types. Terefore, Chomskys approach still
falls squarely within the P&P framework .
Te second and fnal point of disagreement that we mention
here concerns the consequences of unset or misset parameters.
For concreteness, we focus on the switchbox model: Can a switch
be placed in an intermediate, unset position? Alternatively, must
a child sometimes make temporary use of a setting that is not in
fact employed in the target language? If so, what are the conse-
quences for the functioning of the language faculty?
One school of thought is that there is no such thing as an
unset parameter: Every parameter is always in a determi-
nate setting, be it an arbitrary setting (cf. Gibson and Wexler
1994 ), or a prespecifed default setting (e.g., Hyams 1986 ).
On this view, temporary missettings may be routine during
the period when language acquisition is still underway. (Te
notion that certain parameter settings might be defaults, or
unmarked options, has its roots in the phonological concept
of markedness .)
A second school of thought maintains that parameters are
initially unset. Virginia Valian ( 1991 ) proposes that an unset
parameter permits everything that any of its potential values
would allow. Somewhat similarly, Charles D. Yang ( 2002 ) pro-
poses that the learner begins the language acquisition process
not with a single grammar but, rather, with a multitude of dif-
ferent grammars, all in competition against one another. Every
grammar corresponding to a permissible array of parameter-set-
tings is included. A consequence is that competing values of the
same parameter can be in play at the same time.
A cross-cutting view is that children may temporarily enter-
tain nonadult parameter settings (whether default or not; see,
e.g., Tornton and Crain 1994 ). Children may then produce
utterances that use a grammatical structure found in some of
the worlds languages, but not in the target. On this view, what
is crucial is simply that the learner must eventually arrive at the
target parameter-setting, regardless of what parameter-settings
have been temporarily adopted along the way. Tis is the learn-
ing problem that is addressed by Edward Gibson and Wexlers
( 1994 ) trigger learning algorithm , for example.
Principles and Parameters Theory and Language Acquisition
673
quantities by hand. Print culture is defned largely in contrast to
the scribal (or chirographic ) culture that had prevailed in the
West during previous millennia when handcopying was the sole
means of duplicating writing and drawing. Although handcopy-
ing persisted and indeed thrived after the introduction of print-
ing, it did so within a changed literary environment. Handwriting
itself was taught with reference to printed manuals; copyists imi-
tated the title pages, the punctuation, and pagination of printed
books .
Scribal culture had been characterized by an economy of
scarcity. Te large collections of texts gathered in the Alexandrian
library and in some later centers of learning were exceptional
and relatively short-lived. Te retrieval, copying, and recopying
of surviving texts took precedence over composing new ones.
Te acquisition of literacy was confned to restricted groups
of churchmen and lay professionals. Oral interchange predomi-
nated. (As noted in the following, this has led some authorities to
contrast print, not with scribal but with oral culture .)
Print culture introduced an economy of abundance. Te
continued output of handcopied books simply added to a grow-
ing supply. Wholesale production replaced a retail book trade.
Increased output was spurred by competition among printers
and booksellers, who curried favor with the authorities in order
to win the privilege of issuing primers, prayer books, of cial
edicts, and other works for which there was a steady demand.
Print culture gave rise to new laws governing copyright, patent-
ing, and intellectual property. Te literary diets of Latin-reading
professional groups were enriched by access to many more
books than had been available before. More abundantly stocked
bookshelves increased opportunities to compare ancient texts
with each other and with more recent work. Academic activities
were reoriented from preserving ancient wisdom to registering
new fndings and venturing into uncharted felds. Te expansive
character of print culture grew more pronounced over the course
of centuries. Multivolumed reference works required constant
updating; serial publication was introduced; bibliographies grew
thicker and more specialized. Concern about information over-
load was experienced by each generation in turn.
Te drive to tap new markets encouraged populariza-
tion, translations from Latin into the vernaculars, and a gen-
eral democratization of learning and letters. Te church was
divided over whether to support or counter these trends, espe-
cially over whether or not to authorize vernacular translations
of Bibles and prayer books. After the Lutheran revolt, lay Bible
reading was encouraged in Protestant regions and discouraged
in regions that remained loyal to Rome. Whereas a single Index
of Prohibited Books provided guidance to all Catholic of cials,
Protestant censorship was decentralized, taking diverse forms in
diferent realms.
In all regions, learning to read paved the way for learning
by reading. Autodidacts were urged to master various arts by
means of how-to texts (Cormack and Mazzio 2005 ). Authors,
artists, and craftsmen, in collaboration with printers and pub-
lishers, used self-portraits, title pages, and paratextual materials
to advertise their products and themselves. Individual initia-
tive was rewarded; the drive for fame went into high gear. But
the preservative powers of print made it increasingly dif cult
for successive generations to win notice from posterity. An ever
Lillo-Martin , Diane C . 1991 . Universal Grammar and American Sign
Language: Setting the Null Argument Parameters . Dordrecht, the
Netherlands : Kluwer .
Lillo-Martin , Diane , and Ronice Mller de Quadros . 2005 . Te acqui-
sition of focus constructions in American Sign Language and Lngua
de Sinais Brasileira. In BUCLD 29 Proceedings , ed. Alejna Brugos ,
Manuella R . Clark-Cotton , and Seungwan Ha , 36575. Somerville,
MA : Cascadilla Press.
Marcus , Gary F . 1993 . Negative evidence in language acquisition .
Cognition 46 : 53 85.
Pica , Pierre . 1984 . On the distinction between argumental and nonargu-
mental anaphors. In Sentential Complementation , ed. Wim de Geest
and Yvan Putseys , 18594. Dordrecht, the Netherlands : Foris .
Prince , Alan , and Paul Smolensky . 2004 . Optimality Teory: Constraint
Interaction in Generative Grammar. Malden, MA : Blackwell .
Snyder , William . 2001 . On the nature of syntactic variation: Evidence
from complex predicates and complex word-formation . Language
77 : 324 42.
. 2007 . Child Language: Te Parametric Approach . Oxford : Oxford
University Press .
Tesar , Bruce , and Paul Smolensky . 2000. Learnability in Optimality
Teory . Cambridge, MA : MIT Press .
Tornton , Rosalind , and Stephen Crain . 1994 . Successful cyclic move-
ment. In Language Acquisition Studies in Generative Grammar , ed.
Teun Hoekstra and Bonnie D . Schwartz , 21553. Amsterdam: John
Benjamins.
Valian , Virginia . 1991 . Syntactic subjects in the early speech of American
and Italian children . Cognition 40 : 21 81.
Wexler , Kenneth , and Rita Manzini . 1987 . Parameters and learnability in
binding theory. In Parameter Setting , ed. Tomas Roeper and Edwin
Williams , 4176. Dordrecht, the Netherlands : Reidel .
Yang , Charles D . 2002 . Knowledge and Learning in Natural Language .
Oxford : Oxford University Press .
PRINT CULTURE
Print (or typographic) culture designates all the activities entailed
in producing, distributing, collecting, and reading printed mate-
rials and engraved images. As a historical construct, it usually
refers to the literary environment that frst emerged in Western
Europe during the second half of the ffteenth century. Diverse
developments elsewhere (such as the use of xylography in China
and of movable type in Korea, prohibitions against Arabic print-
ing by Ottoman rulers, and the sluggish pace of Russian printing)
lend themselves to comparative study but cannot be covered
here.
In Western Europe (unlike other areas), the printing arts, once
introduced, spread with remarkable rapidity. Between the 1460s
and 1490s, printing shops were established in all of the major
political and commercial centers. New occupations (typefound-
ing and presswork) were introduced; bookmaking arts were
reorganized; trade networks were extended; book fairs inaugu-
rated. By 1500, the use of movable type had become the domi-
nant mode for duplicating texts, and xylography had replaced
hand illumination for replicating images.
Tough often classifed under the heading of book history ,
print culture encompasses a vast variety of nonbook materials,
such as advertisements, almanacs, calendars, horoscopes, proc-
lamations, tickets, and timetables. It also entails the provision
of visual aids (such as maps, charts, tables, graphs, and detailed
drawings) that are especially dif cult to duplicate in large
Print Culture
674
Legislators were less likely to be carried away by a treatise than
by a speaker, and were more likely to think calmly and carefully
before taking action.
Te same distancing efect of print that Condorcet regarded
as benefcial was found objectionable by others. Te roman-
tic movement was in part a reaction against the mathematical
reasoning and abstract thinking that were associated with print
culture. Political romanticism took the form of lamenting the
way the age of chivalry had succumbed to that of economists
and calculators (Burke 1790). Readers were urged by roman-
tic poets, such as William Wordsworth, to abandon dry-as-dust
books: close up those barren leaves! Objections to the pur-
ported distancing efects of print persisted among critics and
media analysts in the twentieth century. Trough the habit of
using print and paper, wrote Lewis Mumford ( 1934 , 1367),
thought lost something of its fowing organic character and
became abstract, categorical, stereotyped, content with purely
verbal formulations and solutions. A similar position was taken
by McLuhan in his depiction of Typographical Man (1962,
367).
Both the proponents and opponents of the ostensibly imper-
sonal character of print tended to overlook its coexistence with
a human presence and a human voice. One thinks immediately
of parents reading to children. But any text that appears in print
lends itself to being read aloud. During the early modern era,
printed broadsides and news reports were especially likely to
be transmitted by word of mouth to listeners gathered around a
few literate townsmen. Even now, public readings or lectures are
delivered to hearing publics by authors of printed best-sellers .
Print culture did not supersede oral culture but did have an
efect upon it. As was true of handwriting and handcopying, the
speech arts, far from languishing, fourished in a more regu-
lated form. Instruction in elocution and in holding debates fg-
ured among the many how-to books that printers kept in stock.
Tere were exceptional preachers (such as the Marian exiles or
the Huguenot refugees) who, when deprived of their pulpits and
sent into exile, turned to printing as their only recourse. But most
preachers (like Martin Luther himself) made full use of both pul-
pit and press.
Other considerations cast doubt on the distancing efect
of print. As noted previously, print culture encompasses images
as well as texts. Whatever the persuasive efects of printed car-
toons and caricatures, they cannot be described as distancing.
Similarly, the fgure of a distant ruler became less distant when
printed (or photographed) portraits could be cut out of news-
papers and enshrined in peasant huts. Even with regard to bare
texts, a skillful writer (whether distant or dead) can still move
unknown readers to tears or incite them to take action.
Before printing, powerful lungs were required for preachers
or orators who hoped to gain a popular following. But the later
political scene saw efective action taken by numerous fgures
who (like John Wilkes) were notably defcient in the speech arts.
Condorcet , for one, was remarkably blind to the political passions
that could be aroused by pamphleteers and journalists. During
the revolutionary era, readers of Tom Paine or Marat were not
distanced from political contestation but were drawn into it .
Te basic features of print culture remained more or less the
same after the industrialization of printing processes in the early
more strenuous efort was required to cope with the burden of
the past (Bate 1970 ) .
Tere are synchronic as well as diachronic aspects to print
culture (see synchrony and diachrony ). Unlike hand-
copied texts, printed copies get issued not consecutively but
simultaneously. Te distribution of printed copies was relatively
slow before the development of modern transport systems.
Nevertheless, the age of the wooden handpress saw a marked
improvement in the coordination of diverse activities, such as
checking the path of a comet against diverse predictions, incor-
porating new fndings and corrections in successive editions of
reference works, or mobilizing a protest movement in diferent
parts of a given realm. We made the thirteen clocks strike as
one, commented an American revolutionary.
Simultaneity went together with standardization, as is illus-
trated in an anecdote about Napoleons minister of education,
who looked at his watch and announced that at that moment all
French schoolboys of a certain age were turning the same page
of Caesars Gallic Wars. Te output of the handpress fell short
of achieving the degree of standardization that marks modern
editions. Yet early modern readers were able to argue, in both
scholarly tomes and polemical pamphlets, about identical pas-
sages on identically numbered pages.
Simultaneity is of special signifcance in conjunction with
journalism. Especially after the introduction of iron presses
harnessed to steam and wire services that made use of the tele-
graph, the newspaper press would restructure the way readers
experienced the fow of time. Simultaneity is nicely illustrated by
the front page layout of a modern newspaper, which has been
described by Marshall McLuhan as a mosaic of unrelated scraps
in a feld unifed only by a dateline (1964, 249). Given the jux-
tapositions and discontinuities presented by the front page, it is
a mistake to associate print only with linear sequential modes
of thought. Although books and newspapers are now fled sepa-
rately in libraries and archives, they are intertwined manifesta-
tions of print culture, beginning with early newsbooks and going
on to later serialized novels .
Even in the age of the handpress, newsprint altered the way
readers learned about afairs of state. It created a forum outside
parliaments and assembly halls and invited ordinary subjects
to participate in debates (by contributing letters to editors). It
provided ambitious journalists (from Jean-Paul Marat to Benito
Mussolini) with new pathways to political power. It served to
knit together the inhabitants of large cities for whom the daily
newspaper provided a kind of surrogate community. According
to Benedict Anderson ( 1983 , 3740), newsprint served a similar
function for millions of compatriots who lived within the bound-
aries of a given nation-state.
Te reception of news via print rather than voice points to a
facet of print culture that has given rise to much debate. It cen-
ters on a contrast not with scribal but with oral culture. To an
Enlightenment philosopher such as the Marquis de Condorcet
(see Baker 1982, 268), who was impressed by advances in math-
ematics, the use of print held the promise of introducing ratio-
nality into political afairs. Whereas speech was ephemeral,
Condorcet argued, a printed account lent itself to rereading and
careful consideration. By means of rhetorical devices, orators
could persuade their audiences to perform ill-considered acts.
Print Culture
675
Wittgensteins remarks need to be read closely; they are
designed to work on the reader rather than profer argu-
ments entailing conclusions, which might be summarized. For
Wittgenstein, philosophy was an activity, and its goal ought to be
to free us of problems, formulated through our misunderstand-
ing the logic of our language (1922, 3). Te interested readers frst
port of call, therefore, should be his central text, Philosophical
Investigations (hereafter PI ). Tis is the most complete of the
posthumously published works and the one that has had most
infuence on subsequent philosophical thought. I discuss
the ways in which interpreters have read the remarks so often
referred to as the private language argument and the conclu-
sions those interpreters have drawn. Wittgensteins writings are
designed to wean one away from certain alluring, though maybe
unconscious, commitments, pictures, analogies, and prejudices.
One cannot, therefore, merely summarize his argument(s) and
conclusion(s), for there is (are) none, in the traditional sense.
In PI , Wittgenstein asks, could we imagine a language in
which a person could write down or give vocal expression to
his inner experiences his feelings, moods, and the rest for
his private use? ([1953] 1958, 243). His imaginary interlocutor
responds by remarking that we do so in our ordinary language.
Wittgenstein rejoins: But that is not what I mean. Te individual
words of this language are to refer to what can be known only
to the person speaking; to his immediate, private, sensations. So
another person cannot understand the language (243) .
Wittgenstein is variously taken, in the 72 (or so) remarks that
follow this passage to be doing one of two things. Some inter-
preters take him to be providing a refutation of the claim that
the language described in 243 is possible through a reductio
ad absurdum , in the process advancing positive philosophi-
cal claims such as an expressive theory for the meaning of frst
person, present tense, psychological utterances, refuting certain
(alleged) Cartesian prejudices regarding the mindbody rela-
tionship, and availing us of a new answer to the problem of other
minds. On another reading, he is taken to bring readers to a posi-
tion whereby they freely acknowledge that such a logically pri-
vate language could have no signifcance for them, that in trying
to state what such a language could be, the philosopher fails to
make sense; furthermore, that their thought that such a language
could have signifcance, could be stated sensically, stemmed
from an unacknowledged, thought-constraining, attachment to
a particular nonobligatory picture of language, the mind or
privacy. Te debate, therefore, cashes out in the following way.
Tose who take Wittgenstein to provide an argument in these
72 or so remarks take that argument to be something along the
lines of the following: For something, say, a set of utterances
say, the signals of the builders in the opening remarks of PI , say,
a logically private language such as that which we are asked to
imagine in 243 to rightly be called a language, it must fulfll
a certain set of criteria. Te criteria give us the meaning of the
word language . Something failing to fulfll these criteria, there-
fore, cannot meaningfully be called language. Tere is a second-
order debate about the nature of the criteria: Are they formal/
logical or are they social? Tere are, thus, those who hold that
the existence of a language entails the existence of a linguistic
community, communitarians, and those who hold that it does
not do so. However, regardless of which position on this second-
nineteenth century and after the adoption of other new technol-
ogies (such as lithography, photography and the shift from hot
to cold type). Nineteenth-century observers believed that the
advent of newspapers signaled the end of the book. Late twen-
tieth-century commentators believed that radio, television, and
other electronic media were going to supersede print. At present,
the movement of texts onto screens has persuaded some observ-
ers that supersession is fnally at hand. In my view, continued
coexistence seems more likely, especially since the preservative
powers of print are still uncontested. An ever-growing short-
age of space on library shelves and an unending concern about
information overload suggest that print culture is still exerting a
cumulative efect and will continue to do so for the foreseeable
future .
Elizabeth L. Eisenstein
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , Benedict . 1983 . Imagined Communities . London : Verso .
Baker , Keith M . 1982 . Condorcet. Chicago : University of Chicago Press .
Bate , W. Jackson . 1970 . Te Burden of the Past and the English Poet .
Cambridge : Harvard University Press .
Burke , Edmund . 1790 . Refections on the Revolution in France.
Available online at: http://www.constitution.org/eb/rev_fran.htm.
Chartier , Roger . 1986 . Texts, printing, readings. In Te New Cultural
History , ed. L.Hunt , 15471. Berkeley : University of California Press .
. 1987 . Te Culture of Print . Princeton, NJ : Princeton University
Press .
Cormack , Bradin , and Carla Mazzio . 2005 . Book Use, Book Teory, 1500
1700 . Chicago : University of Chicago Press . See Part III on Te How-to
Book.
Eisenstein. Elizabeth L . 1997 . From the printed word to the moving
image. Social Research 64 (Fall): 104966.
. 2005 . Te Printing Revolution in Early Modern Europe . New
York : Cambridge University Press .
Martin , H. J ., and Lucien Febvre . 1976 . Te Coming of the Book . Trans.
D. Gerard . London : NLB . See pages 716 for section on Chinese
precedents.
McKenzie , D. F . 2002 . Speech manuscript print. In Making Meaning:
Printers of the Mind and Other Essays , ed. P. MacDonald and
M. Suarez , 237 59. Amherst : University of Massachusetts Press .
McLuhan , Marshall . 1962. Te Gutenberg Galaxy . Toronto : University of
Toronto Press .
. 1964 . Understanding Media . New York : McGraw-Hill [Signet
Books].
Mumford , Lewis . 1934 . Technics and Civilization . New York : Harcourt
Brace .
Ong , Walter . 1982 . Orality and Literacy . London : Routledge .
PRIVATE LANGUAGE
Ludwig Wittgenstein (18891951) is considered one of the most
infuential philosophers of the twentieth century. While his con-
tributions to philosophy are wide-ranging, one of his most widely
discussed and infuential contributions is taken to follow from the
sections of his Philosophical Investigations that explore the pos-
sibility of a logically private language ([1953] 1958, 243315).
Tese remarks have come to be called Wittgensteins private lan-
guage argument . Te label is not, however, without controversy;
we explore why in this entry.
Private Language
676
the expressive theory for the meaning of frst person, present
tense, psychological utterances, then one will be led to argue
that such utterances are not cannot be reports or descriptions
of inner states but are must be rather, expressions or avowals
of judgments or evaluations/appraisals: So, for example, many
cognitivist philosophers (e.g., Lyons 1980 ; Nash 1989 ) and
psychologists (e.g., Lazarus 1982 ) of emotion advance this view
(some, such as Kenny 1963 , even drawing on Wittgenstein as
chief infuence).
If one understands Wittgenstein as not having advanced such
views and, rather, takes his remarks to be designed to work on
one so as to facilitate the realization that there is nothing we
would wish to hold onto answering to the name private lan-
guage, we are not led to a philosophical commitment to any
view on psychological language. We might then, rather, engage
with those who claim or assume that frst person, present tense,
psychological utterances must be reports or descriptions of inner
states or sensations Jamesian accounts of emotion in general
and cognitive neuroscience in particular (e.g., Damasio 1994 )
and with those who claim or assume that they must be expres-
sions or avowals as both being driven by prejudice. Jamesian/
neoJamesian and cognitivist theories of emotion can be seen
to rest on prejudice about our use of psychological language (see
also emotion words ).
Wittgenstein, read aright, can provide much help in our
attempts to dissolve such prejudice in the human sciences .
Phil Hutchinson
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baker , Gordon . 2004 . Wittgensteins Method: Neglected Aspects .
Oxford : Blackwell . See Part I, Section B, Chapters 5, 6, and 7. Te book
contains a number of excellent chapters on the (so-called) private lan-
guage argument and its interpretation.
Damasio , A. R . 1994 . Descartes Error: Emotion, Reason, and the Human
Brain . New York : Grosset/Putnam.
Kenny , A . 1963 . Action, Emotion and Will . London : Routledge .
Lazarus , R. S . 1982 . Toughts on the relations between emotion and cog-
nition . American Psychologist 37 : 1019 24.
Lyons , W . 1980 . Emotion . Cambridge : Cambridge University Press .
Mulhall , Stephen . 2006 . Wittgensteins Private Language: Grammar,
Nonsense and Imagination in Philosophical Investigations, 243315 .
Oxford : Oxford University Press . Nuanced reading.
Nash , R. A . 1989 . Cognitive theories of emotion . Nos 23 : 482 504.
Wittgenstein , Ludwig . 1922 . Tractatus Logico-Philosophicus .
London : Routledge .
. [1953] 1958. Philosophical Investigations . Oxford : Blackwell .
PROJECTIBILITY OF PREDICATES
Te distinction between projectible and nonprojectible predi-
cates arises in analyses of inductive inference. David Hume
([1748] 2000 ) showed that it is not possible to justify the belief
that past empirical regularities will continue into the future.
Nelson Goodmans 1953 new riddle of induction raised a fur-
ther problem. Goodman showed that there are always an unlim-
ited number of hypotheses that encompass all of the evidence,
yet confict in their predictions. For example, if we introduce
the predicate grue (grue = examined before some future time
order debate such readers take, they all ( if they take Wittgenstein
to have been successful in his alleged aim) hold that a logically
private language, as described in the fnal paragraph of PI 243,
is shown by Wittgenstein to fail to fulfll the relevant criteria for
being a language. It is this, such readers claim, that he demon-
strates argues in the 72 or so remarks that follow it.
Te alternative way of reading these remarks is as fol-
lows: Wittgenstein, in PI , asks us to imagine such a language
([1953] 1958, 243), that is, to (try to) entertain the thought of the
possibility of a language which describes my inner experiences
and which only I can understand (256). Te remarks that follow
work on the reader to the extent that he or she sees that however
one tries to give sense to such a (putative) language, we never
arrive at a position where our desire to see it as such is satisfed.
Read aright, the remarks serve to dispel the desire to attempt to
give sense to the locution [a] language which describes my inner
experiences and which only I can understand or a logically pri-
vate language. On this reading, it is not that there is something
akin to a misuse according to the rules of grammar of the
concepts private and language, such that such locutions are
nonsense. It is not something that the philosopher wants to say
but cannot owing to the confguration of grammatical rules. It is,
rather, that when we try to imagine a private language, we real-
ize that there is no determinable thing a private language to
imagine. Te very notion of a private language dissipates as we
try to grasp it.
On the latter reading, therefore, Wittgenstein does not advance
a theory as to the nature of frst person, present tense, psycholog-
ical utterances but merely ofers suggestions as to how it might be
possible to understand these as learned replacements for (say) a
cry of pain, rather than as, for example, a description or report of
an inner state, such as a sensation. He ofers such suggestions, as
it were, as prophylactics. To accept such suggestions as possible
weans one of of the assumption that such utterances must be
descriptions of inner states (sensations, for example) and feeds
into weaning one away from the assumed need and the desire to
give sense to the locution private language.
Te debate between the two readings, therefore, hinges
on how one should understand Wittgensteins philosophical
method (or metaphilosophy). Tose who take him to ofer a refu-
tation of a logically private language and to be, in the course of
doing so, advancing positive claims as to (say) the nature of frst
person, present tense, psychological utterances, do so, as their
opponents suggest, by underplaying his remarks on philosophi-
cal method (especially PI 109 and 126 through 133), where he
lays out his therapeutic vision of philosophy. Here, the practice
of philosophy is undertaken as a therapeutic dialogue between
the Wittgensteinian philosopher and his or her interlocutor
indeed, the therapist and interlocutor might be conficting ten-
dencies in oneself. Te task of the philosopher-as-therapist is to
facilitate the interlocutors free realization that he or she is in the
grip of a particular picture of the way things must be that leads
him or her to be committed to certain nonobligatory philosophi-
cal positions.
What one takes to be done by Wittgenstein in these 72 (or
so) remarks has more than merely exegetical signifcance. If one
understands him to have refuted the possibility of a logically pri-
vate language and, in so doing, to have advanced, for example,
Private Language Projectibility of Predicates
677
PROJECTION (BLENDING THEORY)
metaphor, analogy, categorization , and many other cog-
nitive operations are customarily understood within cognitive
linguistics in terms of projected conceptual structures or in
terms that are compatible with conceptual projection. In the
conceptual blending framework, in particular, projection
describes the way that conceptual structures can be combined
or superimposed by copying conceptual content from one men-
tal space to another. Tis term is unrelated to the use of pro-
jection to describe anticipatory utterances and gestures in
conversation (e.g., Streeck 1995 ) or the projection principle
of government and binding theory.
Te notion of projection as a kind of mapping that serves as
a fundamental mechanism of thought has its roots in ideas from
conceptual metaphor theory and frame semantics , as
well as mental space theory. George Lakof and Mark Johnson
( 1980 ) explain individual instances of metaphorical language as
refections of systematic relationships between two conceptual
domains, in which language and structure from a source domain
is projected, or mapped onto , a situation in a target domain (see
source and target ). A sentence like Were drifting apart
depicts elements from the domain of emotional intimacy in
terms of structures mapped from the domain of physical prox-
imity. Charles J. Fillmore ( 1982 ) observes that words like lend
make sense only in light of certain schematic representations
of situation types, or frames . A word is said to evoke an associated
frame or frames, prompting a language user to project the evoked
frames structure onto an unframed assembly of elements. Tis
projection allows the language user to understand those elements
in terms of the relationships and roles belonging to the frame .
Projection in the conceptual blending framework difers from its
counterparts in conceptual metaphor theory and frame semantics
in two ways. First, projection in conceptual blending is a process
that takes place between mental spaces, not domains or schemas.
Second, where the former theories describe projections involving
exactly two conceptual structures, the conceptual blending frame-
work calls for four at a minimum. Material is projected from at least
two input spaces into at least two middle spaces (Fauconnier and
Turner 1994 ): a generic space that refects the roles, frames, and
schemas that the inputs have in common, and a blended space
where projected elements are integrated and develop emergent
structure through processes of composition , completion , and
elaboration (Fauconnier and Turner 1998 , 2002 ).
Backward projection refers to circumstances whereby struc-
ture is projected from the blended space back to its inputs. Tis
can be a desirable outcome in which inferences developed in
the blend enhance understanding of the input material. Gilles
Fauconnier and Mark Turner ( 2002 ) illustrate this kind of desir-
able backward projection with the Buddhist Monk riddle: A
monk walks up a mountain one day and walks down the same
path over the same time period on another day. A solution to
the puzzle of whether there is a place on the path that the monk
occupies at the same time of day on both journeys is to imagine
that the monk takes both walks on the same day in this blended
scenario, at some moment he will meet himself. Only project-
ing the location of the meeting place from the blended space
back to the input spaces yields the conclusion that the original
t and green, or not so examined and blue), the hypotheses [H1]
All emeralds are green and [H2] All emeralds are grue both
conform to the observed data, though making divergent predic-
tions. So, Goodman asks, what justifes adopting H1 rather than
H2 or countless other conficting hypotheses? Why, that is, on
the basis of the very same evidence do we project some predi-
cates (e.g., green) and not others (e.g., grue)? A predicate,
then, is considered projectible if it is suitable for use in inductive
inference.
Te projectible/nonprojectible distinction, however, has sig-
nifcance for other important issues. In particular, it is appealed
to in distinguishing laws of nature (eg. H1) from generalizations
that while true, seem true only by accident (e.g., [H3] All the
objects on table D are green). Te distinction is also central in
evaluating the truth or falsity of counterfactual conditionals and
in accounts of the basis of similarity judgments. Tis leads some
to identify projectible predicates with so-called natural kind
predicates. Green and emerald pick out natural kinds; grue
and on table D do not.
Providing criteria for distinguishing projectible from non-
projectible predicates (or natural kinds from other kinds) remains
a challenge (Schef er 1981 ; Stalker 1994 ; Schwartz 2005 ). Te dif-
ference between projectible and nonprojectible predicates does
not hinge on explicit temporal reference. If grue and bleen
(bleen = blue and examined before time t , or not so examined and
green) are taken as primitives and green and blue defned,
the latter predicates, not the former, will mention time. In fact,
temporal considerations need not play any role in setting the new
riddle. A curve drawn through data points represents a hypothesis
that projects new values from those observed. But an unbounded
number of curves that make conficting predictions about unob-
served values can be drawn through these same points .
Continuing eforts to explicate the concept projectible pred-
icate in purely syntactic or semantic terms have not been
successful. Attempts to draw a projectible/nonprojectible dis-
tinction along epistemological and metaphysical lines have
also been unsatisfactory, tending either to be parochial or to
presuppose dichotomies tantamount to the distinction itself.
Solutions appealing to simplicity, similarity, and innate quality
spaces have run into like dif culties. Goodman proposes draw-
ing the distinction along pragmatic lines. Projectible predicates
are those with a history of successful predictive use. As a result,
they become entrenched in our vocabulary and inductive
practices. Pragmatic entrenchment underlies the confdence
we have in projecting them and is a primary reason for their felt
naturalness .
Robert Schwartz
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Goodman , N. 1953 . Fact, Fiction and Forecast . Cambridge : Harvard
University Press.
Hume , D . [1748] 2000. An Enquiry Concerning Human Understanding .
Oxford: Oxford University Press.
Schef er , I . 1981 . Anatomy of Inquiry . Indianapolis : Hackett .
Schwartz. R . 2005 . A note on Goodmans problem . Journal of Philosophy
102: 3759.
Stalker , D ., ed. 1994. Grue! Chicago : Open Court .
Projectibility of Predicates Projection (Blending Theory)
678
properties of lexical items (that is, both their features and selec-
tional requirements) must be represented at every syntac-
tic level of representation. Its key efect is to allow structure at
every level of representation to be directly determined by lexical
properties, thereby largely eliminating the need for language-
particular rules. For this reason, the projection principle has
played a key role in the transition from early rule-based trans-
formational grammar to principles and parameters theory.
Let us briefy review two major consequences of the projec-
tion principle. Te frst concerns the elimination of language-
particular base rules (the rules that generate deep structures; see
underlying structure and surface structure ) in trans-
formational grammar. Such rules would state, for example, that
in English a verb phrase may consist of a verb followed by either
a noun phrase or a prepositional phrase or a noun phrase and a
prepositional phrase, and so on. In addition, the lexical entries of
verbs specify the syntactic environment in which they can occur.
Tus, the entry for the verb put states that it must be followed by a
noun phrase and a prepositional phrase. Tis arrangement gives
rise to redundancy: In the case at hand, the fact that there are
English verbs that can be followed by a noun phrase and a prepo-
sitional phrase is expressed twice, namely, in the base rules and
in the lexical entry of put (and other verbs). Since lexical proper-
ties are idiosyncratic, elimination of this redundancy requires a
simplifcation of the base component, and this is precisely what
the projection principle makes possible. As it requires syntactic
representations to be projections of lexical properties, it allows
the base component to be reduced to a small universal skeleton,
known as x-bar theory , and relegates language-specifc prop-
erties of deep structures to the lexicon and to a set of word-order
parameters .
Te projection principle also implies that syntactic represen-
tations that are the result of movement must contain traces
that function as place holders for categories that have undergone
this operation. It is easy to see why this should be so. As stated
earlier, the lexical entry for the verb put must state that this verb is
followed by a noun phrase and a prepositional phrase. However,
in the question in (1a), the selectional requirements of this verb
appear not to be met, since the noun phrase that usually follows
it has undergone movement. Satisfaction of the projection prin-
ciple, therefore, requires that movement leave behind a trace
shown as t
NP
in (1b) so that the selectional properties of put are
also expressed in the structure that results from movement.
(1) a. I wonder [[
NP
what] Jack has [
VP
[
V
put] [
PP
in the oven]]]
b. I wonder [[
NP
what] Jack has [
VP
[
V
put] t
NP
[
PP
in the oven]]]
Hans van de Koot
WORK CITED AND SUGGESTION FOR FURTHER READING
Chomsky , Noam . 1981 . Lectures on Government and Binding . Dordrecht,
the Netherlands : Foris .
. 1986 . Knowledge of Language: Its Nature, Origin, and Use . New
York : Praeger.
PROPOSITION
As used in the philosophy of language and logic, a proposition
is what is believed and/or asserted. Te prima facie case for the
single monk, walking on two diferent days, will indeed cross
some spot on the path at the same time on each day .
An important principle of projection in conceptual blending
theory is that it is always selective . Not all elements and relations in
the input spaces are projected into the blended space. Sometimes
only one of a pair of counterpart elements is projected, some-
times both, and sometimes none. Te organizing structure of one
or both inputs can also be projected in part, or not at all. In some
integration networks, called single-scope networks , only one input
projects its organizing frame to the blended space, while the other
input contributes other elements but little or no organizing struc-
ture. (Conventional sourcetarget metaphors are considered
prototype examples of this kind of integration network.) Te
projection involved in other kinds of networks is also selective: If I
were you, Id take the job , for example, prompts the listener to take
some aspects of the interlocutor into consideration and some
aspects of himself or herself, but by no means all of either .
Unconstrained projection, by contrast, would undermine the
usefulness of these inferences, and projecting too much struc-
ture in either direction constitutes an error. Excessive projection
from the inputs in a blend leads to mistakes, such as assuming
that because my computer interface includes something called
a menu, I will need to fnd some waiter who can accept my
order. Inappropriate projection of emergent structure from the
blended space back to the inputs leads to other mistakes, such as
confusing actors with the characters they play on TV. Te pos-
sible competing pressures governing these constraints on pro-
jection are examined in detail in Fauconnier and Turner ( 1998
and 2002 , 30952) .
Vera Tobin
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Fauconnier , Gilles , and Mark Turner . 1994 . Conceptual projection and
middle spaces. Department of Cognitive Science Technical Report
9401, University of California, San Diego.
. 1998 . Principles of conceptual integration. In Discourse
and Cognition , ed. Jean-Pierre Koenig, 26983. Stanford, CA : CSLI
Publications .
. 2002 . Te Way We Tink: Conceptual Blending and the Minds
Hidden Complexities . New York : Basic Books .
Fillmore , Charles J . 1982 . Frame semantics. In Linguistics in the Morning
Calm , ed. Te Linguistic Society of Korea, 11137. Seoul : Hanshin
Publishing .
Grady , Joseph , Todd Oakley , and Seanna Coulson . 1999 . Blending
and metaphor. In Metaphor in Cognitive Linguistics , ed. Raymond
Gibbs and Gerard Steen, 10024. Amsterdam and Philadelphia : John
Benjamins . A useful comparison of related concepts in conceptual
metaphor and conceptual blending theory.
Lakof , George , and Mark Johnson . 1980 . Metaphors We Live By .
Chicago : University of Chicago Press .
Streeck , Jrgen . 1995 . On projection. In Social Intelligence and
Interaction: Expressions and Implications of the Social Bias in Human
Intelligence , ed. Esther N. Goody, 87110. Cambridge : Cambridge
University Press.
PROJECTION PRINCIPLE
Te projection principle (Chomsky 1981 ) is a cornerstone of
principles and parameters theory . It states that the
Projection Principle Proposition
679
equivalent propositions would then have to be identical. Te
necessarily true proposition that 1 + 1 = 2 is identical to the prop-
osition that 60,375 is divisible by 3 (both identical to the set of all
possible worlds). But intuitively, one can believe the frst propos-
ition without believing the second.
Many take this to show that a proposition must be more fne-
grained than simply a set of worlds. On the Russellian approach,
a proposition expressed by a that-clause is constructed from the
normal referents of the expressions contained in the clause. Te
number 2 is a constituent of the proposition that 1 + 1 = 2, but it is
not part of the proposition that 60,375 is divisible by 3. Since the
propositions are distinct, we can believe one without believing
the other.
Unfortunately, the Russellian approach has dif culties
accounting for an observation by Gottlob Frege ([1892] 1948 ).
Frege famously pointed out that identity statements can difer
in cognitive value. Intuitively, Amie believes that Mark Twain is
Mark Twain can be true while Amie believes that Mark Twain
is Samuel Clemens is false. But on the Russellian approach, this
is impossible. Since Mark Twain is Samuel Clemens, the two
that-clauses refer to the same proposition. In response, Nathan
Salmon ( 1986 ) insists that the two belief reports are indeed
equivalent. Our contrary intuition refects only a diference in
pragmatics .
A Fregean alternative is to postulate that the proper names
Mark Twain and Samuel Clemens have diferent senses (see
sense and reference ) for Amie, that she represents the same
person in two diferent ways. In addition, these names do not have
their customary referents within the context of believe. Instead
of referring to the same famous author, the two names actually
refer to their associated senses within propositional attitude
contexts. Consequently, the that-clauses in the two belief reports
refer to distinct propositions, which is why the two reports are
not equivalent. However, one objection to the Fregean approach
is that proper names do not change their referents in proposi-
tional attitude contexts (Davidson 1968 ). Another objection is
that people can believe the same proposition even if they asso-
ciate diferent senses with the same name. Tese children all
believe that Santa exists can be true even if the children repre-
sent Santa in diferent ways.
Rudolf Carnap ( 1958 ), W. V. O. Quine ( 1956 ), and Donald
Davidson ( 1968 ) analyzed propositional attitudes as relations
to sentences, rather than language-independent propositions.
Michelle believes that 1 + 1 = 2 is true if and only if Michelle
stands in a certain relationship to the sentence 1 + 1 = 2.
Tis avoids the need to postulate propositions, but there are
new problems to resolve. For example, the sentence Delman
believes something profound. can be true, even if Delman does
not speak English, or there is no English sentence that captures
the content of his profound belief.
Other authors, such as Mark Richard ( 1990 ) and Richard
Larson and P. Ludlow ( 1993 ), have sought to develop hybrid
approaches, where a proposition combines language-
independent entities with linguistic items or mental represen-
tations. A more radical approach is to deny that propositional
attitudes are binary relations. Mark Crimmins ( 1992 ) argues that
although a propositional attitude verb like believe is syntactically
a two-place predicate, it actually expresses a three-place relation
existence of propositions is that they make sense of the fact that
from
Terry believes that cats are cute.
we can infer that
Tere is something that Terry believes.
Propositions, while often thought to be what is expressed by a
sentence, are said to be distinct from sentences. One sentence
(e.g., Fred put the money in the bank) can express more than
one proposition (in this case, depending on the meaning of
bank); and one proposition can be expressed by two diferent
sentences (e.g., snow is white and neige est blanc ).
Proponents of propositions typically treat them as the pri-
mary bearers of truth and falsity, but they disagree over their
nature and structure .
Michael P. Lynch
PROPOSITIONAL ATTITUDES
Propositional attitudes are mental relations believing, desiring,
hoping, and so on between an individual and a proposition.
Te sentence Amie believes that 1 + 1 = 2 is a propositional
attitude report, where Amie stands in the belief relation to the
proposition that 1 + 1 = 2. Te proposition specifes the content of
the attitude (see intentionality ). Propositional attitudes play
a central role in ordinary psychological explanations. We explain
why Elia presses the doorbell by saying that he hopes someone
will answer the door, and he believes that this will happen if he
presses the bell.
The Nature of Propositional Attitudes
One controversy about propositional attitudes concerns their
relationship to the mental representations investigated in
psychology and cognitive science. Te language of thought
theory says that propositional attitudes correspond to language-
like mental representations their combinatorial syntax allows
them to form propositional attitudes with new contents .
A second controversy about propositional attitudes con-
cerns their purported social or environmental dependency. For
internalists , whether one has a certain propositional attitude
depends only on ones internal physical brain state (Searle 1983 ).
However, externalists (McGinn 1977 ; Burge 1979 ) argue that
mental relations to propositions are mediated through linguistic
conventions and the external environment. Two physically iden-
tical individuals can possess diferent propositional attitudes if
they are embedded in diferent environments (see meaning
externalism and internalism ) .
The Semantics of Propositional Attitudes Reports
Te analysis of propositional attitude reports is a particularly
thorny problem for semantics . Tere is no consensus as to the
right approach, and the resolution of the problem depends on
many other issues in semantics, such as the nature of a propos-
ition. Under possible worlds semantics , a proposition is a
set of possible worlds. Te proposition that pigs fy is the set of
worlds where pigs fy. But a serious problem is that necessarily
Propositional Attitudes
680
need not be defnite (sometimes called fuzzy boundaries), and
category membership is graded with respect to how good a
member is judged to be.
Te prototype view was frst proposed by Eleanor Rosch
( 1973 ) as a general framework encompassing her cross-cultural
work on color and form categories, and then elaborated into a
programmatic body of empirical research challenging the classi-
cal view (Rosch 1978 , 1999 ). Te theory was that categories form
around perceptually, imaginally, or conceptually salient stimuli,
then, by stimulus generalization, spread to other similar stimuli;
such prototype stimuli serve as the conceptual and imaginal ref-
erence points by which the category as a whole is represented
and understood. Empirically, all categories do show gradients
of membership; that is, subjects easily, rapidly, and meaning-
fully rate how well a particular item fts their idea or image of
the category to which the item belongs. Note that these are not
probability judgments but judgments of degree of membership.
Gradient of membership judgments apply to diverse kinds of cat-
egories: perceptual categories such as red , semantic categories
such as furniture , biological categories such as woman , social
categories such as occupations , formal categories that have clas-
sical defnitions such as odd number , and ad hoc , goal-derived
categories such as things to take out of the house in a fre. In con-
trast, subjects cannot list criterial attributes for most categories
(Rosch and Mervis 1975 ) .
Gradients of membership must be considered psychologi-
cally important because such measures have been shown to
afect virtually every major method of study and measurement
used in psychological research (Rosch 1975 , 1978 , 1999 ):
1. Association : When asked to list members of the category,
subjects produce better examples earlier and more frequently
than poorer examples. Association is taken as the key to men-
tal structure in many systems; for example, in approaches as
diverse as British empiricist philosophy, connectionism , and
psychoanalysis .
2. Speed of Processing : Te better an example is of its cat-
egory, the more rapidly subjects can judge whether or not that
item belongs to the category. Reaction time has been consid-
ered a royal road to the study of mental processes in cognitive
psychology.
3. Learning : Good examples of categories are learned by sub-
jects in experiments and acquired naturalistically by children
earlier than are poor examples, and categories can be learned
more easily when better examples are presented frst fndings
with implications for education (see Markman 1989 ; Mervis
1980 ).
4. Expectation : When subjects are presented a category name
in advance of making rapid judgment about the category, per-
formance is helped (i.e., reaction time is faster) for good and
hindered for poor members of the category. Called priming or
set in psychology, this fnding has been used to argue that the
mental representation of the category is in some ways more
like the better than the poorer exemplars .
5. Inference : Subjects infer from more to less representa-
tive members of categories more readily than the reverse, and
the representativeness of items infuences judgments in formal
logic tasks, such as syllogisms (see also Smith and Medin 1981 ;
cf. verbal reasoning ).
among a subject, a proposition, and a way of believing. A way
of believing is supposed to be similar to a Fregean sense, but it
is contextually specifed and does not correspond to any explicit
referring item in the report .
Joe Lau
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Burge , Tyler . 1979. Individualism and the mental. In Midwest Studies in
Philosophy . Vol. 4. Ed. P. French , T. Uehling , and H. Wettstein , 73121.
Minneapolis : University of Minnesota Press .
Carnap , Rudolf . 1958 . Meaning and Necessity: A Study in Semantics and
Modal Logic . 2d ed. Chicago : University of Chicago Press .
Crimmins , Mark . 1992 . Talk about Beliefs . Cambridge, MA : MIT Press .
Davidson , Donald . 1968 . On saying that . Synthese 19 : 130 46.
Frege , Gottlob . [1892] 1948. Sense and reference . Philosophical Review
57 : 209 30.
Larson , Richard , and P. Ludlow . 1993 . Interpreted logical forms .
Synthese 95 : 305 55.
McGinn , Colin . 1977 . Charity, interpretation, and belief . Journal of
Philosophy 74 : 521 35.
McKay , Tomas , and Michael Nelson . 2008 . Propositional attitude
reports. In Te Stanford Encyclopedia of Philosophy (Fall), ed. Edward
N. Zalta. Available online at: http://plato.stanford.edu/archives/
fall2008/entries/prop-attitude-reports/. Tis is a comprehensive sur-
vey on the semantics of propositional attitude reports.
Quine , Willard Van Orman . 1956 . Quantifers and propositional atti-
tudes . Journal of Philosophy 53 : 177 87.
Richard , Mark. 1990 . Propositional Attitudes: An Essay on Toughts and
How We Ascribe Tem . Cambridge : Cambridge University Press .
Salmon , Nathan . 1986 . Freges Puzzle . Cambridge, MA : MIT Press .
Searle , John . 1983 . Intentionality. An Essay in the Philosophy of Mind .
Cambridge : Cambridge University Press .
PROTOTYPES
Prototypes refers to one of the ways in which psychology has
attempted to account for the nature of concepts and concep-
tual categories (see categorization) . In the prototypes view,
conceptual categories form around and/or are represented in the
mind by salient, information-rich ideas and images that become
prototypes for the category. Other members of the conceptual
category are judged in relation to these prototypes, thus forming
gradients of category membership.
Te importance of prototype theory needs to be seen in its
historical context. Prior to this work, concepts and conceptual
categories were assumed from philosophy to be arbitrary logical
sets with defning features and clear-cut boundaries. All mem-
bers of the conceptual category were considered equivalent with
respect to membership. Tis is now called the classical view .
Psychological research on concept learning used artifcial sets
of stimuli, structured into microworlds in which these assumed
characteristics of categories were already built in. Mainstream
linguistics was constructed on similar assumptions; phonol-
ogy, semantics , and syntax all sought to decompose the
subject matter of their domains (speech sounds, word mean-
ing , and grammar) into sets of abstract binary defning features,
a procedure called componential analysis . In contrast, in the pro-
totype view, there need be no defning attributes that all concep-
tual category members have in common; category boundaries
Prototypes
681
as an account of concepts and categorization. Te main objec-
tions fall into two camps: In the frst, prototypes and graded
structure violate the classical requirement that the real meaning
of a concept (that to which it refers) must be the identifable nec-
essary attributes of a classical defnition. One argument for this
view is that prototype and graded structure efects can be found
for conceptual categories that have a formal classical defni-
tion, such as odd number (Armstrong, Gleitman, and Gleitman
1983 ), another that prototypes do not form componential com-
binations as do the elements of classical defnitions (e.g., a good
example of pet fsh is neither a prototypical pet nor a prototypical
fsh; Osherson and Smith 1981 ). Both fndings are taken to indi-
cate that prototypes are something other than and irrelevant to a
concepts meaning. One solution is a dual model in which proto-
types are assigned the function of rapid recognition of conceptual
referents, whereas the true meaning is provided by a classically
defned core (Osherson and Smith 1981 ; Smith, Shoben, and Rips
1974 ). In the second camp, prototypes change with context for
example, prototypical animals in the context of a zoo difer from
those in the context of a farm (Barsalou 1987 ). Such fndings are
taken to indicate that it is theories that determine concept mean-
ing (Medin 1989 ). (For a review of more philosophical objections
to prototypes and all other accounts of concepts see Fodor
1998 and Laurence and Margolis 1999 .)
In conclusion, prototype and graded structure efects are well
established; the debate concerns what they mean for our under-
standing of concepts, categorization, thinking, decision making,
the meaning of words and word combinations, and innumerable
other aspects of human functioning. Te prototype view is not
necessarily incompatible with other approaches if the diferent
views are seen at a level deep enough that their complementar-
ity can be appreciated. Prototype efects do not deny that under
some circumstances we explicitly think in terms of necessary
and sufficient conditions , nor does the essentialist intu-
ition that there are nonobvious realities behind outer appear-
ances (Gelman and Wellman 1991 ) exclusively require the form
of classical defnitions for its implementation . Furthermore, pro-
totypes, like any other mental or cultural process, only function
within the ever-shifting contexts of partially organized forms of
encyclopedic knowledge and belief the sort of understanding
toward which the theories view points. What prototype research
uniquely points out is that there is a level of organization of the
mind in which concepts appear to be represented by information-
rich, imagistic, sensory-based, often emotion-linked wholes that
are used in thinking and communicating without reference to
defnitions, category boundaries, or even truth conditions .
Eleanor Rosch
WORKS CITED AND SUGGESTIONS FOR FURTHER REFERENCES
Armstrong , Sharon , Lila Gleitman , and Henry Gleitman . 1983 . What
some concepts might not be . Cognition 13 : 263 308.
Barsalou , Lawrence . 1987 . Te instability of graded struc-
ture: Implications for the nature of concepts. In Concepts and
Conceptual Development: Ecological and Intellectual Factors in
Categorization , ed. Ulric Neisser , 10140. Cambridge : Cambridge
University Press .
Fodor , Jerry . 1998 . Concepts: Where Cognitive Science Went Wrong . New
York : Oxford University Press .
6. Probability Judgments : Representativeness strongly infu-
ences probability judgments (Kahneman, Slovic, and Tversky
1982 ), which is important because probability is thought by
many philosophers to be the basis of inductive inference and,
thus, of the way in which we learn about the world.
7. Natural Language Indicators of Graded Structure : Natural
languages themselves contain various devices that acknowledge
and point to graded structure, such as hedge words like techni-
cally and really (see also Lakof 1987 ; Taylor 2003 ).
8. Judgment of Similarity : Less good examples of categories
are judged more similar to good examples than vice versa. Tis
violates the way similarity is treated in logic, where similarity
relations are symmetrical and reversible (Tversky 1977 ) .
What determines the items that will be prototypical of catego-
ries in the frst place? Some are based on statistical frequencies,
such as the means or modes (or family resemblance struc-
tures) for various attributes; others appear to be ideals made
salient by such factors as physiology (good colors, good forms),
social structure (president, teacher), culture (saints), goals
(ideal foods to eat on a diet), formal structure (multiples of 10
in the decimal system), causal theories (sequences that look
random), and individual experiences (the frst learned or most
recently encountered items or items made particularly salient
because they are emotionally charged, vivid, concrete, meaning-
ful, or interesting). Note that particular exemplars can be pro-
totypes if they serve as reference points for the category; thus,
prototype and exemplar theories are not necessarily contradic-
tory. Note also that it is a misapprehension to take prototypes to
mean only one kind of prototype and to critique prototype theory
on that basis.
Te prototype view has spread beyond psychology to many
felds, including linguistics and narratology . Gradients of
exemplariness are ubiquitous in linguistic phenomena, even
in phonology where actual speech is less clear-cut than would
appear in an abstract componential analysis. In semantic and
syntactic analyses (particularly in cognitive grammar and
the understanding of metaphor ), prototype efects, as well as
providing specifc case studies are often used as evidence that
formal analysis is insuf cient of itself and that world knowledge
must be part of ones theory (Lakof 1987 ; Langacker 1990 ; Taylor
2003 ) .
Tere are societal implications of prototype theory. For exam-
ple, social stereotypes are a type of prototype; it is called a
stereotype when it applies to a group of people and has social
consequences. Another example: Anglo-American case law is
based on prototypes; precedent cases provide the reference
points in arguing present cases (see legal interpretation ).
Further examples: Political arguments are often about conficting
prototypes (e.g., diferent images of welfare mothers) even when
goals (care for children) may be quite similar, while scholarly
debates are frequently about attempts to draw clear-cut bound-
aries where there are none when the participants actually agree
on central prototypes (e.g., agreement on clear cases of what we
mean by language or religion but disagreement about borderline
phenomena and about criteria). Many such issues could be clar-
ifed by understanding the principles of prototyping .
Although prototype efects are now acknowledged as empiri-
cally established, there have been many criticisms of prototypes
Prototypes
682
and the journal Proverbium , the latter in Honeck ( 1997 ), and
both views in Mieder ( 2003 ) .
The Cognitive Science of Proverbs
Proverbs are best considered not in isolation but as members of
a larger family or category that includes the proverb, pictorial
renderings of the proverbs literal meaning, verbal and picto-
rial interpretations of the proverbs fgurative meaning, and ver-
bal and pictorial instances of the proverbs fgurative meaning.
Teoretically, the proverbs fgurative meaning serves as the con-
ceptual glue that connects the family members.
For example, the family for the proverb Great weights hang
on small wires, might include a picture of barbells hanging on
a thin wire, an interpretation such as Te outcome of important
events often depends on seemingly minor details, and verbal
instances such as Te shortstop tripped on a pebble and the
game was lost and Te nurse accidentally bumped the sur-
geons hand and the patient died. Crucially, and except for the
proverbliteral picture connection, the family members share no
literal similarity, either of a linguistic or imagistic sort.
Psychological research with proverbs has contributed to
issues in several areas of cognitive science (see Honeck 1997 ),
as follows:
MENTAL REPRESENTATION. Because research shows that people
can reliably connect the family members, a case can be made for
the role of an abstract, amodal, nonlinguistic, nonimagistic men-
tal representation in cognitive processes such as remember-
ing and categorization .
For example, proverbs are remembered better if they are frst
presented along with related as opposed to unrelated, interpreta-
tions. And related interpretations, and even verbal instances, serve
as efective prompts for recall of a proverb. Indeed, people judge a
good interpretation to be the best way to represent the meaning of
a proverb family. Moreover, if people fxate on the literal imagery
evoked by a proverb, they are unable to recognize verbal instances
of its fgurative meaning. Tese several results strongly suggest
that the knowledge that guides memory and categorization can be
quite abstract and theory-like, and that imagery is best construed
as the outgrowth of a particular level of understanding .
ON-LINE PROCESSING. Must the literal meaning of a trope be
deciphered before its fgurative meaning can be constructed?
Te answer from reaction-time research on proverbs is in the
af rmative, although the literal primacy efect reduces with prov-
erb familiarity. Providing advance markers (e.g., proverbially
speaking) reduces processing time for unfamiliar proverbs, but
event-related potential (brain wave) measures indicate that the
words in proverbs used fguratively, versus literally, are harder to
integrate with the discourse context (Schwint, Ferretti, and Katz
2006 ). Moreover, since proverbs almost always function as indi-
rect speech-acts, these results also indicate that such acts can be
dependent on prior literal access .
CEREBRAL ASYMMETRY. Research on brain damage indicates
that the right hemisphere is better than the left hemi-
sphere at processing nonliteral meanings, the kind involved in
metaphors , connotation, verbal humor , sarcasm, the point
Gelman , Susan , and Henry Wellman , 1991. Insides and essences: Early
understanding of the non-obvious . Cognition 38 : 213 44.
Kahneman , Daniel , Paul Slovic , and Amos Tversky , eds. 1982. Judgment
under Uncertainty: Heuristics and Biases. New York : Cambridge
University Press .
Lakof , George . 1987 . Women, Fire, and Dangerous Tings: What
Categories Reveal about the Mind . Chicago : University of Chicago
Press .
Langacker , Ronald . 1990 . Concept, Image, and Symbol: Te Cognitive
Basis of Grammar . Berlin : Mouton de Gruyter .
Laurence , Stephen , and Eric Margolis . 1999 . Concepts and cognitive
science. In Concepts: Core Readings , ed. Eric Margolis and Stephen
Laurence , 381. Cambridge, MA : MIT Press .
Markman , Ellen . 1989 . Categorization and Naming in Children .
Cambridge, MA : MIT Press .
Medin , Douglas . 1989 . Concepts and conceptual structure . American
Psychologist 44 : 1469 81.
Mervis , Carolyn . 1980 . Category structure and the development of
categorization. In Teoretical Issues in Reading Comprehension , ed.
William Brewer , Bertram Bruce , and Rand Spiro , 279308. Hillsdale,
NJ : Erlbaum.
Osherson , Daniel , and Edward Smith . 1981 . On the adequacy of proto-
type theory as a theory of concepts . Cognition 9 : 35 58.
Rosch , Eleanor . 1973 . Natural categories . Cognitive Psychology
4 : 328 50.
. 1975 . Cognitive representations of semantic categories . Journal
of Experimental Psychology: General 104 : 192233.
. 1978 . Principles of categorization. In Cognition and
Categorization , ed. Eleanor Rosch and Barbara Lloyd , 2748. Hillsdale,
NJ : Lawrence Erlbaum .
. 1999 . Reclaiming concepts . Journal of Consciousness Studies
6.11/12 : 61 77.
Rosch , Eleanor , and Carolyn Mervis . 1975 . Family resemblances: Studies
in the internal structure of categories . Cognitive Psychology
7 : 573 605.
Rosch , Eleanor , Carolyn Mervis , Wayne Gray , David Johnson , and
Penelope Boyes-Braem . 1976 . Basic objects in natural categories .
Cognitive Psychology 8 : 382 439.
Smith , Edward , and Douglas Medin . 1981 . Categories and Concepts .
Cambridge : Harvard University Press .
Smith , Edward , Edward Shoben , and Lance Rips . 1974 . Structure and
process in semantic memory: A featural model for semantic deci-
sions . Psychological Review 81 : 214 41.
Taylor , John . 2003 . Linguistic Categorization . Oxford : Oxford University
Press.
Tversky , Amos . 1977 . Features of similarity . Psychological Review
84 : 327 52.
PROVERBS
Proverbs in the Humanities and Cognitive Science
Te humanities treat proverbs as repositories of wisdom about
everyday life. Tis premise motivates religious, literary, practical,
and culturalfolklore approaches that capitalize on the fact that
proverbs can be pithy, express a moral or precept, sound author-
itative, perform various pragmatic functions (e.g., exhortation),
and serve as indirect speech-acts , in which what is intended
encompasses more than what is actually said.
Te cognitive science perspective examines how proverbs
both illuminate and are illuminated by knowledge about the
mind. Ultimately, the humanities and cognitive science views
are complementary. Te former is represented in Mieder ( 1994 )
Proverbs
683
the unconscious, symptoms, dreams, etc. Some later writers,
such as Marshall Edelson ( 1975 ) and Bonnie Litowitz ( 1978 ),
drew on transformational grammar to rethink psycho-
analytic ideas. (See also Ricoeur 1978 on both approaches.)
Function and Architecture
As the preceding cases may already begin to suggest, it is valu-
able to distinguish functional accounts of mental phenomena in
psychoanalysis from specifc, explanatory architectures. Put very
simply, functional accounts treat what mental operations accom-
plish (e.g., fear causes us to fee predators); a particular architec-
ture defnes the causal organization that enables the functions
(e.g., fear is produced when certain perceptual stimuli produce
convergent arousal in the amygdala ). In psychoanalysis, the
functional idea of repression is separable from any particular
account of how repression occurs and where it is located in a
set of mental structures. When Lacan drew on structuralism to
theorize the unconscious, he was preserving the basic functional
idea of Freud while altering the precise way in which the mental
structures thus, the implementation of those functions were
specifed. Te same point holds for Litowitz and the Freudian
idea of overdetermination (roughly, the multiple interpretable
meanings and causal sources of a dream image or symptom).
Unfortunately, these uses of specifc linguistic theories seem
highly problematic. First, they often presuppose the fundamen-
tal principles of Freudian architecture and try to graft a very dif-
ferent theory onto them. It is not clear that the theories being
combined are compatible. Moreover, the initial Freudian archi-
tecture often seems ill-defned and not particularly well sup-
ported by more recent research. Second, the linguistic theories
may not apply literally to the psychoanalytic functions anyway
(e.g., the relation between underlying structure and the
Freudian unconscious is, at best, only analogical). Finally, when
not used in a loose, metaphorical way, the linguistic theories
have sometimes been taken up in specifc forms that were subse-
quently bypassed in linguistics, leaving the psychoanalytic rede-
velopments outdated.
Tere may be a more productive way of revising Freudian
architecture, however, while preserving basic functional prin-
ciples of psychoanalysis, including those that bear on language.
Specifcally, we might begin with the relatively well-estab-
lished architecture that has emerged from cognitive neurosci-
ence in recent years. From here, we might seek to reformulate
some central functional principles of psychoanalysis in that
architecture. (Tis is roughly the program of neuropsychoanaly-
sis .) In the remainder of this entry, I set out one possible account
of this sort.
Before going on to language, however, we should briefy con-
sider the general relation between some key functional divisions
in psychoanalysis primarily the conscious/unconscious, or
repression and neurocognitive architecture .
Repression and the Brain
Diferent readers of Freud and diferent psychoanalysts will
characterize psychoanalysis diferently. For our purposes, we
may consider a few elements of psychoanalytic theory to be cen-
tral and distinctive. I would frst of all list the dynamic uncon-
scious, thus, mental contents that are not available to conscious
of a story, gist, and proverbs. Te right brain appears to process
inputs in a more fexible, open, contextualist, and holistic way,
propensities that would facilitate proverb comprehension .
DEVELOPMENTAL TRENDS. Proverb comprehension improves
slowly over time, beginning at about age seven. Comprehension
is facilitated by supporting context (e.g., pictures), proverb con-
creteness and familiarity, and reduced information-processing
load, as well as by the child having a better vocabulary, meta-
cognitive awareness of nonliteral meanings, a formal education,
and a good social knowledge base. Te cognitive processes sub-
serving improvement are unclear, although analogy has been
implicated .
Theories of Proverb Comprehension
Te extended conceptual base theory (Honeck 1997 ) is a cognitive
psychological, laboratory-derived approach. It claims that prov-
erbs are distinguished by form and function, are variably famil-
iar, and are understood via several phases of problem solving
that are nonautomatic, heavily inferential, and error prone . Te
great chain metaphor theory (Lakof and Turner 1989 ), a cogni-
tive linguistics cultural approach, makes antithetical claims.
To date, almost all of the empirical research has been done in
conjunction with, and is supportive of, the conceptual base the-
ory (Honeck and Temple 1996 ) .
Richard P. Honeck
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Honeck , Richard . 1997 . A Proverb in Mind: Te Cognitive Science of
Proverbial Wit and Wisdom . Mahwah, NJ : Erlbaum .
Honeck , Richard , and Jon Temple . 1996 . Proverbs and the complete
mind . Metaphor and Symbolic Activity 11 : 217 22.
Lakof , George , and Mark Turner . 1989 . More Tan Cool Reason: A Field
Guide for Poetic Metaphor . Chicago : University of Chicago Press .
Mieder , Wolfgang . 1994 . Wise Words: Essays on the Proverb . New
York : Garland .
Mieder , Wolfgang , ed. 2003. Cognition, Comprehension, and
Communication: A Decade of North American Proverb Studies .
Baltmannsweiler, Germany : Schneider Verlag .
Schwint , Christopher A ., Todd R. Ferretti , and Albert N. Katz . 2006 . Te
infuence of explicit markers on slow cortical potentials during fgura-
tive language processing. In Proceedings of the 28th annual Conference
of the Cognitive Science Society , 268. Mahwah, NJ : Erlbaum .
PSYCHOANALYSIS AND LANGUAGE
Psychoanalysis is, frst, a talking cure, a verbal treatment for
mental illness. As such, it has been bound up with speech, mean-
ing, and interpretation from the outset. In keeping with this
connection, psychoanalytic theorists and scholars examining
psychoanalysis have repeatedly addressed the intertwining of
psychoanalysis and language. Some scholars have explored the
place of language and/or language science in Freudian thought
and practice (see, for example, Forrester 1980 and Loewald 1978 ).
Others have drawn on specifc theories of language in order to
alter, reorient, enhance, or reunderstand Freudian principles.
Te most famous case of the latter was Jacques Lacan, who drew
on structuralist ideas to revise psychoanalytic accounts of
Psychoanalysis and Language
684
Language and the Unconscious
Since language in psychoanalysis necessarily involves compo-
nents, I organize my discussion (somewhat loosely) by reference
to Roman Jakobsons infuential sixfold analysis of the speech
situation into speaker, hearer, context, medium of contact, mes-
sage, and code.
SPEAKER. Te psychoanalytically crucial idea here, developed
most famously by Lacan, is the diference between the subject
as speaker and the subject as object of speech. When I describe
myself, there is necessarily a diference between the I that does
the describing and the myself that is described. Te latter is
linguistic in two senses. First, my account of myself sometimes
called my self-concept is largely given in verbal descrip-
tions. Tese include not only descriptions that I have formu-
lated on my own but also those that I have taken up from others
(e.g., from student evaluations of my teaching). More generally,
language provides me with the categories by which I identify
myself as a teacher or a scholar, as boring or interesting, clear
or obscure (see meaning and belief ). According to Lacan
and others, it is this necessary division between the (largely
verbal) self-object and the (speaking) subject that allows there
to be an unconscious. Put very simply, the unconscious occurs
in that motivationally consequential part of me that is excluded
from my self-concept. Te very idea of such a division may seem
counterintuitive, as we ordinarily imagine our self-knowledge to
result directly from introspection. However, it is well established
outside of psychoanalysis that this is not the case. For example,
drawing on an extensive body of research, Richard Nisbett and
Lee Ross explain that [k]nowledge of ones own emotions and
attitudes, though commonly believed to be direct and cer-
tain, has been shown to be indirect and prone to serious error.
Such knowledge is based in large part on inferences about causes
of behavior (1980, 227) .
HEARER. Te crucial feature of the hearer in psychoanalysis
concerns not the hearer per se but, rather, ones implicit imagi-
nation of the hearer. In psychoanalysis, the unconscious has
both emotional and cognitive consequences. Among the most
crucial are those that defne the transference . Transference is,
roughly, unconsciously basing ones response to someone in
ones current environment on repressed memories of, attitudes
toward, or ideas about some earlier fgure (e.g., ones father).
It is well established in pragmatics that speakers engage
cooperatively with addressees, not only in following general
principles but also in making specifc adjustments for the likely
knowledge, interests, attitudes, and so on of their addressees.
In all conversations, these adjustments will be partially guided
by cognitive structures, such as prototypes and exemplars .
For example, I may avoid making a particular sort of joke with
a colleague because I have a strong memory of such a joke
fopping in the past. Tat memory would be a salient exem-
plar, guiding my expectations about my addressee. One might
understand transference as the unconscious reliance on a par-
ticular exemplar for ones addressee. Moreover, this reliance
bears on motivationally crucial aspects of the interaction (e.g.,
trust). Finally, it operates even when incompatible with indi-
vidual or situational information (e.g., about the addressees
knowledge even though they are the type of contents that are ordi-
narily available to consciousness. In other words, some aspects
of the mind for example, grammatical principles are the
sorts of things we can only infer; we cannot know them directly.
However, other aspects of the mind for example, memories,
beliefs, and desires are the sorts of things that we ordinarily
seem to know directly. When we do not know memories, beliefs,
and desires, they are unconscious, as are grammatical principles.
But the reason for their unconsciousness is diferent. It is a mat-
ter of repression (hence, the dynamic part of dynamic uncon-
scious ). Repression begins with mental confict that gives rise to
emotional pain. To take a simple example, one might experience
confict between ones desires and ones moral aspirations, lead-
ing to feelings of guilt. In certain circumstances, we respond to
this pain by making one element of the confict inaccessible to
conscious thought in this case, presumably, the desires. Tis
can occur at any age. However, classical psychoanalysis posits
that primary repression occurs in early childhood. In adulthood,
secondary repression takes place when a confict is assimilated to
primary repressed material.
Tis simplifed picture of repression is at least broadly consis-
tent with current neurocognitive architecture. We often experi-
ence emotional confict due to the diferent imperatives of our
various (largely subcortical) emotion systems. Research suggests
that the anterior cingulate cortex (ACC) monitors conficts in task
performance (see, for example, MacDonald et al. 2000 , 1835 ).
When the ACC recognizes painful confict, it activates the dor-
solateral prefrontal cortex, which engages inhibitory processes
(see Ito et al. 2006 ; Preston and de Waal 2002 ; Lieberman and
Eisenberger 2006 ). More generally, our brains operate to reduce
emotional conficts, producing valenced outcomes from ambiva-
lent inputs (Ito and Cacioppo 2001 , 69). Psychoanalysis adds two
things to this model. First, it isolates a class of feelings and ideas
so thoroughly inhibited that they do not, then or subsequently,
enter working memory . Second, it gives special importance to
early childhood experiences though this is broadly in keeping
with the cognitive neuroscientifc idea of critical periods in
childhood when certain aspects of cognition (e.g., knowledge of
language) are shaped in crucial ways .
Te problem with repression is that the motivational force of
the unconscious elements continues to have consequences in
ones behavior. One part of this fts in a general way with current
neurocognitive accounts of memory, which distinguish explicit
episodic and semantic/factual memories from implicit emotional
memories (see LeDoux 1996 , 182). It is well established that we
are not consciously aware of emotional memories, though they
afect our behavior. Emotional memories are commonly asso-
ciated with episodic and semantic/factual memories that allow
us to explain our feelings. One way in which the two may be
dissociated is through brain damage. Psychoanalysis suggests
that another way they may be separated is through repression
of the relevant episodic and semantic/factual memories. Such
repression would presumably allow the emotional memories to
operate motivationally without our awareness, awareness nor-
mally guided by the (now repressed) explicit memories. In any
case, the behavior afected by consciously unavailable motives
prominently includes ones verbal behavior which brings us to
language .
Psychoanalysis and Language
685
impulses. But the generation of ordinary speech is no less deter-
minate, and it is not determined solely by ones self-conscious
purposes. At any given moment, there are many ways in which
I might phrase what I wish to say. However, I say things in one
way only. For example, in speaking of a particular object, I might
refer to it as a p.c., a computer, a laptop, and so on. But I choose
one. In a connectionist model , we would say that a range of
factors has given the greatest activation to a particular phrasing.
In some cases, this is just the basic level term. In other cases,
there is some contextual or other reason for the choice. It might
be a matter of repetition (I just saw a copy of PC Computing ,
so I say p.c.); alternatively, it might involve a more complex
lexical relation (someone just mentioned desktops, so I
say laptop). Psychoanalysis adds a series of circuit activations
from dynamically unconscious motivational contents. For exam-
ple, Ernst Lanzer (Freuds Rat Man) tells Freud about some-
one who fattered and deceived him. He identifes the person as
a medical student. Tere are many ways in which Lanzer could
have introduced this person. What caused the specifcally med-
ical detail to be activated? One possibility is that Lanzer had a
transferential relation to Freud in which he feared that Freud
would fatter and deceive him in the treatment. Circuits con-
necting Freud with this student (and presumably with parental
fgures from the period of primary repression; see Hogan 1996 ,
14850) would then have shaped this speech specifcally, the
choice of medicine as an identifying characteristic separately
from Lanzers self-conscious intentions .
CODE. Various features of a linguistic code particularities of the
lexicon, morphology, grammar may allow the expression of
unconscious ideas and impulses beyond the straightforward
semantics of lexical selection. Tis may occur most obviously
through ambiguity or puns, a point stressed particularly by
Lacan. For example, James Gorney ( 1990 ) reports a case in which
someone reported conscious anxiety over a bill and thereby
expressed unconscious anxiety over someone named Bill. Tis
is most readily understood in terms of left hemisphere versus
right hemisphere processes. Neurocognitive research shows
us that the right hemisphere generates multiple meanings, some
of which are inhibited by the left hemisphere, which selects the
contextually relevant meanings (see Chiarello 1998 , 145; Faust
1998 , 180). For example, in the semantic context of references to
a restaurant, and the linguistic context of following a determiner,
left hemisphere interpretive processes will limit the meanings of
bill to a list of charges. However, right hemisphere interpre-
tive processes will briefy generate beak and William also.
Psychoanalysis once again adds to this well-established archi-
tecture a dynamic unconscious component. Specifcally, it adds
the idea that in some cases, a particular word (e.g., bill , rather
than check ) may be produced due to unconscious activations
(see spreading activation ) that are motivational, though
inaccessible to the working memory of the speaker .
Conclusion
Te relation between psychoanalysis and language has been
explored by many writers using many approaches. In the pre-
ceding sections, I have tried to show that central functional
principles of psychoanalysis as they relate to language may be
trustworthiness, as when my unconscious exemplar fosters
unwarranted trust) .
CONTEXT. Te context of a given speech event might be under-
stood to encompass not only such matters as the social and
material situation but also the array of ideas and feelings that
accompany speech on the part of both the speaker and the
hearer. Tis includes what psychoanalysts refer to as fantasy ,
which is bound up with transference. Specifcally, my thoughts
about an addressee are embedded in a series of often feeting
ideas and imaginations that are connected with that addressee,
the larger situation, and other matters. Tis fantasy structure is
likely to give rise to ephemeral emotional ambivalences in any
circumstances. When transference is involved, it is likely to gen-
erate more sustained confict that repeatedly infects my moti-
vational and cognitive responses to the addressee. Such fantasy
may be understood, frst, as the complex of images, memories,
feelings, and, of course, words and meanings that are continually
produced in association with speech and action through prim-
ing . Tese primed contents are partially activated but have not
reached a threshold where they draw attentional focus in work-
ing memory. In some cases, psychoanalysts would add, they can-
not draw attentional focus. Nonetheless, they undoubtedly afect
our action, including our speech production guiding choices
of, for example, lexical alternatives .
MEDIUM OF CONTACT. Many aspects of the physical channel
(Jakobson 1987 , 66) of communication are psychoanalytically
consequential, including visual and aural aspects of both lin-
guistic and paralinguistic action (see paralanguage ). One
important aspect of psychoanalytic interpretation is that lin-
guistic and paralinguistic features may contradict one another
such that the paralinguistic features may partially express the
unconscious attitudes and ideas that are inhibited in the verbal
statement. Tis is connected with our relatively limited control
over the expressive component of emotions. Specifcally, our
prefrontal control of our expressions usually cannot convinc-
ingly replicate the expressions produced by the emotion systems
themselves a point familiar to anyone who has tried to smile
for a photograph. Tis division allows us to recognize many sorts
of insincere expression in ordinary life. Psychoanalysis adds that
paralinguistic features may reveal inhibited ideas or feelings with
strong motivational force, even when we are unaware of those
ideas or feelings. Moreover, certain symptomatic acts may have
the same function. In one case, one of Freuds patients repeat-
edly moved her fnger in and out of a small purse as she spoke.
Freud took this to be a masturbatory action ( [1905] 1981, 75).
Whether or not he was right in this case, in principle it is pos-
sible for this sort of thing to occur. One form of implicit memory
involves procedural schemas, prominently motor routines for
various habitual actions. It is possible to activate a motor routine
without self-conscious monitoring. If there are repressed moti-
vations, it may be possible for those to manifest themselves in at
least certain schemas of this sort .
MESSAGE. One aspect of expression stressed by Freud is its
determinacy. Tis is most famously (or notoriously) the case
in errors, which Freud explained by reference to unconscious
Psychoanalysis and Language
686
Ito , Tifany , Geofrey Urland , Eve Willadsen-Jensen , and Joshua Correll .
2006 . Te social neuroscience of stereotyping and prejudice: Using
event-related brain potentials to study social percpetion. In Cacioppo,
Visser, and Pickett 2006, 189208.
Jakobson , Roman . 1987 . Language in Literature . Ed. Krystyna Pomorska
and Stephen Rudy. Cambridge, MA : Belknap.
Lacan , Jacques . 1966 . crits. Paris : Editions de Seuil .
LeDoux , Joseph . 1996 . Te Emotional Brain: Te Mysterious Underpinnings
of Emotional Life. New York : Touchstone .
Lieberman , Matthew , and Naomi Eisenberger . 2006 . A pain by any
other name (rejection, exclusion, ostracism) still hurts the same: Te
role of dorsal anterior cingulate cortex in social and physical pain. In
Cacioppo, Visser, and Pickett 2006, 16787.
Litowitz , Bonnie . 1978 . On overdetermination. In Smith 1978, 35594.
Loewald , Hans . 1978 . Primary process, secondary process, and lan-
guage. In Smith 1978, 23570.
MacDonald , Angus , Jonathan Cohen , Andrew Stenger , and Cameron
Carter . 2000 . Dissociating the role of the dorsolateral prefrontal and
anterior cingulate cortex in cognitive control . Science 288 : 1835 8.
Nisbett , Richard , and Lee Ross . 1980 . Human Inference: Strategies and
Shortcomings of Social Judgment . Englewood Clifs, NJ : Prentice-Hall .
Preston , Stephanie , and Frans de Waal . 2002 Empathy: Its ultimate and
proximate bases . Behavioral and Brain Sciences 25 : 1 72.
Ricoeur , Paul . 1978 . Image and language in psychoanalysis. In Smith
1978, 293324.
Smith , Joseph , ed. 1978. Psychoanalysis and Language . New Haven,
CT : Yale University Press .
PSYCHOLINGUISTICS
Psycholinguistics studies the relationship between the mind
and language. Its concern is with the cognitive processes that
underlie the acquisition, storage, and use of language and their
correlates in observable neurological processes in the brain. Te
feld is thus heavily reliant upon principles and research meth-
ods adopted from cognitive psychology. However, it is essentially
multidisciplinary, drawing also upon linguistics, speech science,
phonetics , computer modeling, neurolinguistics, discourse
analysis (linguistic) and semantics .
Psycholinguistic inquiry is driven by the premise that lan-
guage has developed in ways that refect the structure of the
human mind. Certain shared cognitive routines are assumed
to underpin the production and reception of language, despite
variations between individuals in terms of level of vocabu-
lary and powers of self-expression. Research thus tends to be
normative in direction, though due allowance is made for the
characteristics of the language being examined, the task being
undertaken, and the population under investigation. A favored
approach is to investigate small-scale components of a process-
ing system in order to gradually build up a picture of the system
as a whole.
An interest in the psychology of adult language emerged
during the nineteenth century, with initiatives such as Paul
Pierre Brocas work (1863) on the location of language in the
brain and Francis Galtons (1883) on word association. A paral-
lel interest in child language acquisition had developed earlier
from the Enlightenment debate between rationalist followers
of Descartes, who believed that much human knowledge was
innate, and empiricists such as Hume and Locke, who asserted
that it was entirely acquired.
integrated with current cognitive neuroscientifc architecture.
Some psychoanalytic principles seem plausible in this architec-
ture, and they enrich our understanding of human psychology
and language use in that context, though they do not alter the
architecture per se. Te question that remains open is whether
there a function of repression and, if so, precisely how it operates
in neurocognitive architecture. For example, as the Nobel lau-
reate and prominent neuroscientist Gerald Edelman explained,
referring to the Freudian unconscious and the notion of repres-
sion, it is conceivable that the modulation of value systems
in the brain could provide a basis for the selective inhibition of
pathways related to particular memories (2004, 95; Edelman
suggests that subcortical involvement, particularly in relation to
focal attention, is crucial to this inhibition). If Freud was right to
posit such a functional principle, then psychoanalysis may alter
our understanding of cognitive neuroscientifc architecture by
introducing new processes or structures or at least new sys-
temic consequences resulting from previously unrecognized
interactions of already posited processes and structures. If he
was not right about this functional principle, it remains the case
that many principles of psychoanalysis may be accommodated,
in altered form, in cognitive neuroscience, including many prin-
ciples of psychoanalysis and language .
Patrick Colm Hogan
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Beeman , Mark , and Christine Chiarello , eds. 1998. Right Hemisphere
Language Comprehension: Perspectives from Cognitive Neuroscience .
Mahwah, NJ : Lawrence Erlbaum .
Cacioppo , John , Penny Visser , and Cynthia Pickett , eds. 2006 . Social
Neuroscience: People Tinking About Tinking People . Cambridge,
MA : MIT Press .
Chiarello , Christine . 1998 . On codes of meaning and the meaning
of codes: Semantic access and retrieval within and between hemi-
spheres. In Beeman and Chiarello 1998, 14160.
Edelman , Gerald . 2004 . Wider Tan the Sky: Te Phenomenal Gift of
Consciousness . New Haven, CT : Yale University Press .
Edelson , Marshall . 1975 . Language and Interpretation in Psychoanalysis .
New Haven, CT : Yale University Press .
Faust , Miriam . 1998 . Obtaining evidence of language comprehension
from sentence priming. In Beeman and Chiarello 1998, 16186.
Fiumara , Gemma . 1992 . Te Symbolic Function: Psychoanalysis and the
Philosophy of Language . Oxford : Blackwell .
Forrester , John . 1980 . Language and the Origins of Psychoanalysis . New
York : Columbia University Press .
Freud , Sigmund . [1905] 1981. Bruchstck einer Hysterie-Analyse . Frankfurt
am Main : Fischer Taschenbuch .
. [1909] 1982. Bemerkungen ber einen Fall von Zwangsneurose . In
Zwei Falldarstellungen . Frankfurt : Fischer Taschenbuch.
Gorney , James . 1990 . Refections on impasse. In Criticism and
Lacan: Essays and Dialogue on Language, Structure, and the
Unconscious, ed. Patrick Colm Hogan and Lalita Pandit, 14751.
Athens : University of Georgia Press .
Hogan , Patrick Colm . 1993 . Dora: Desire and ambiguity in the fragment
of a psychoanalysis . American Journal of Psychoanalysis 53.3 : 205 18.
. 1996 . On Interpretation: Meaning and Inference in Law,
Psychoanalysis, and Literature . Athens : University of Georgia Press .
Ito , Tifany , and John Cacioppo . 2001 . Afect and attitudes: A social neu-
roscience approach. In Handbook of Afect and Social Cognition , ed.
Joseph Forgas, 5074. Mahwah, NJ : Lawrence Erlbaum .
Psychoanalysis and Language Psycholinguistics
687
evidence but is also infuenced by the expectations created by
context. Groups of words are parsed into syntactic structures
(see parsing, human ), and a proposition is extracted from
the utterance. Te second phase, meaning construction, requires
the user to elaborate on the literal meaning of the utterance by
inferring details that have been left unexpressed by the speaker/
writer. Te user then adds the new information to the meaning
representation or mental model built up so far in the discourse
and checks for consistency .
Te productive skills of speaking and writing proceed in the
opposite direction: from idea to language. An abstract represen-
tation of the planned sentence is created, and linguistic form is
then conferred upon it. While the sentence is being produced, it
has to be stored in a mental bufer , in the form of a set of instruc-
tions to the articulators or the fngers. Te user monitors perfor-
mance and self-corrects if necessary. An important diference
between most writing events and most speech events lies in the
time available for planning, self-monitoring, and review.
Accounts of language processing rely upon constructs from
cognitive psychology, such as attention, memory , and automa-
ticity . An important principle is that human working memory
is limited in capacity. Te language user minimizes the demands
made upon memory by transferring information into a more per-
manent store ( long-term memory ) or by establishing highly auto-
matic routines for speech assembly or word recognition that do
not require focused attention. In addition, schema theory helps
to explain how a listener/reader enriches a message by the add-
ition of external knowledge and how a speaker/writer is able to
abridge a message, relying upon the recipient to fll in missing
details .
Te methods used to study language processing include obser-
vation and verbal report, but an experimental approach tends to
predominate. Te tasks employed are often small in scale, tap-
ping into on-line processes ; those that enable the researcher to
measure reaction time (e.g., time taken to recognize a word) are
especially favored. Increasingly, researchers also draw upon
fndings from brain imaging (see neuroimaging ) that map the
neurological correlates of diferent types of language activity .
Language Representation
A closely allied area of enquiry considers how linguistic knowl-
edge is stored in the mind in a way that enables rapid matches to
be made. Words are said to be stored as lexical entries , containing
information about the items orthographical form and pronun-
ciation, word class , infections, and combinatorial possibilities.
Tere has been discussion of whether there are separate entries
for af xes such as un- in the word unhappy or whether words of
this type have their own entries. Also much discussed has been the
question of category membership: how a language user manages
to classify an item of crockery as a cup rather than a mug or bowl .
Te entries in a users mental lexicon are massively interconnected,
enabling a process known as spreading activation , in which it
becomes easier to locate (e.g.) the words nurse, hospital , or patient
after recently hearing or seeing the word doctor . Te connections
vary in strength, favoring words of high frequency and words with
a high probability of occurring together .
Traditional linguistic accounts assume that syntax is repre-
sented in the mind as a set of rules, but recent psycholinguistic
Te term psycholinguistics probably dates from the 1930s,
but progress during the frst half of the twentieth century was
discouraged by the dominant behaviorist view that the human
mind is unknowable. Te feld did not emerge as a discipline
in its own right until the mid-1950s. An impetus was given by
a series of essays in which George Miller mapped out possible
areas of inquiry. A major landmark was then Noam Chomskys
1959 rebuttal of the behaviorist assumptions expressed by B.
F. Skinner in his Verbal Behavior . Citing the speed with which
an infant masters a linguistic system and the poverty of the evi-
dence available, Chomsky concluded that language is a geneti-
cally acquired faculty. His nativist stance (see innateness and
innatism ) stimulated modern studies of child language and
began a controversy that continues to the present day.
Much early psycholinguistic inquiry was closely allied to lin-
guistic theory. It explored aspects of Chomksys early genera-
tive grammar on the assumption that the rules represented
psychological reality : that is, that they provided a model of the
operations of the mind, as well as a linguistic account of gram-
matical structure. Te fndings were mainly negative or incon-
clusive, and the paths of linguistics and psycholinguistics began
to diverge, although a body of psychological linguists contin-
ues today to work within a framework of Chomskyan theory .
In its current form, psycholinguistics falls into a number of
distinct areas, but considerable overlaps between them give the
discipline coherence.
Language Processing
Processing research seeks to characterize the four language
skills of reading , writing, speaking ( speech production )
and listening ( auditory processing , speech perception )
in terms of the cognitive operations that underlie them. It also
investigates how vocabulary is retrieved from the mind ( word
recognition ) and how syntactic structures are assembled
or interpreted. Its fndings have important applications for the
teaching of literacy (see teaching reading and teaching
writing ), for language therapy, and for second language learn-
ing (see second language acquisition ).
Drawing upon an information-processing precedent, accounts
often depict language users as taking linguistic material through
a series of levels of representation . For example, a listener
might be seen as building sounds into syllables , syllables into
words , words into clauses, and clauses into abstract meanings.
In fact, the process is not necessarily a sequential one, as the
language user is capable of operating at all these levels simul-
taneously. What is more, there is evidence that higher-level
knowledge infuences processing at a lower level (knowledge of
the existence of a particular word might assist the recognition of
a string of sounds in a top-down way). Opinion continues to be
divided as to whether the relationship between the levels is fully
interactive or whether each level operates with a degree of inde-
pendence (in a modular way). Te beneft of the former is that
all sources of information become available at once; the beneft
of the latter is that rapid decision making is enabled.
Te receptive skills of listening and reading entail two dis-
tinct operations. In the frst, visual or auditory information is
decoded, a process that entails mapping from strings of sounds
or letters to known words. Te user relies partly upon perceptual
Psycholinguistics
688
wild or confned to their rooms for long periods. Tere are also
studies of the course of language development in twins, blind
children, the children of deaf parents, and children acquiring a
pidgin language .
Brain and Language
Interest in the location of language in the brain was stimulated
by the discoveries of Broca and Carl Wernicke, which showed
that damage to two small areas of the left hemisphere was
associated with the loss of speech. Later research supported the
view that the left hemisphere was the dominant one for language
in nearly all right-handers and the majority of left-handers. A
number of children who sufered damage or surgical interven-
tion to the left hemisphere before the age of fve appeared to
relocate their language operations in the right one (see right
hemisphere ), an efect that was less evident in adults. From
these fndings grew a theory that the brain was plastic in the
early years of life, and that language became lateralized to the left
hemisphere around the age of fve. Te conclusion was that brain
lateralization corresponded to a critical period for acquiring
ones frst language.
Views have now changed in the face of evidence that the brain
continues to manifest fexibility into adulthood. Research sug-
gests that some areas are more heavily involved than others in
particular aspects of language use but that language processing
as a whole cannot be localized. It is widely distributed across the
brain and supported by massive neural interconnections that
enable information to be transmitted extremely rapidly. It has
also become apparent that the right hemisphere of the brain has
an important role in language, dealing broadly with larger units,
such as prosody, discourse structure, and so on.
Neuroscience contributes increasingly to psycholinguistics,
with brain-imaging technology enabling researchers to locate
and track the electrical impulses or blood fows associated with
diferent types of linguistic behavior .
Language as a Human Faculty
Here, two central concerns dominate. Te frst is the extent to
which language can be regarded as restricted to the human race.
Current evidence suggests that it is, though pigmy chimpanzees
have shown themselves capable of acquiring large sets of symbols
and using them in quite elaborate ways (see animal commu-
nication and human language and primate vocaliza-
tions ). Te second issue concerns how and when language frst
evolved (see origins of language ). It is suggested that its
emergence was dependent both upon the evolution of an appro-
priate vocal apparatus (see speech anatomy, evolution
of ) and upon changes in the human brain. A problematic issue
for nativists is to explain how language frst entered the human
gene pool; a catastrophic account has been proposed, in which a
sudden change in brain infrastructure coincided with increased
communicative demands. A weakness of the theory is that the
brain evolves very slowly, whereas language changes quickly.
A number of leading commentators have, therefore, concluded
that language initially developed in a way that accorded with the
processing capabilities of the brain, rather than vice versa, and
that the functions of the brain have since adapted to accommo-
date it through coevolution .
thinking has swung toward the notion that speakers can only pro-
duce language as rapidly and accurately as they do if they make
extensive use of stored chunks of language already assembled
syntactically. Te human mind appears better adapted for the
storage of enormous amounts of information than for rapid pro-
cessing. exemplar theory suggests that linguistic knowledge
derives from multiple memory traces of individual language epi-
sodes. Te attraction of the theory is that it accounts not only for
the speakers ability to retrieve chunks but also for the listeners
ability to deal with a range of voices and accents and to recog-
nize the many forms that a single word can take in connected
speech. Instead of mapping back to one idealized template for
a phoneme or word, the listener draws upon vast numbers of
traces laid down by earlier encounters .
Language Acquisition (Developmental Psycholinguistics/
Child Language Development)
Tose who adopt a Chomskyan line examine the productions
of infants for evidence of universal linguistic principles or of
the setting of parameters to incorporate features specifc to the
target language (see principles and parameters theory
and language acquisition ). Others prefer a data-driven
approach that analyzes the speech of the developing child
with no pretheoretical assumptions. Some 30 years of research
has produced evidence that input to infants (child-directed
speech) is more informative than Chomsky once asserted.
Developmental patterns have been traced in the childs phonol-
ogy ( phonology, acquisition of ) and vocabulary ( lexical
acquisition ). Tere has been research into how an infant
builds up lexical categories such as dog or fower, overex-
tending them in some instances and underextending them in
others. Issues in morphology and syntax include whether
there is a fxed order for the acquisition of infections and how
infants manage to derive semantic relationships from cues such
as word order .
It has long been recognized that the timing and rate of lan-
guage acquisition varies enormously from one child to another.
It is customary, therefore, to mark development not by the childs
age but by the average number of morphemes in a childs utter-
ances, a measure known as mean length of utterance (MLU) .
Gradual increases in MLU refect changes in the childs cognitive
capacity and in its ability to articulate groups of words.
Alongside longitudinal observation in the form of recordings
and diaries, language acquisition is also studied experimentally.
Children are asked to perform linguistic or perceptual tasks
designed, for example, to elicit plural forms or to demonstrate
phonological awareness. Methods have also been devised that
provide insights into the mental processes of preverbal children
by tracking shifts in the direction of their attention (see commu-
nication, prelinguistic ). Computer simulation also plays a
part. A recent challenge to nativism comes from connection-
ist modeling, which has shown, on a limited scale, that a com-
puter program is capable of acquiring language rules by exposure
to repeated examples.
Further insights into language acquisition are achieved by
studying children who grow up in exceptional circumstances .
Te most extreme cases are those of children who have been
deprived of human company through being abandoned in the
Psycholinguistics
689
by a single semantic system, or completely integrated. Other
issues concern the extent to which one language infuences the
use of the other and whether there are cognitive costs and ben-
efts in growing up bilingual .
Language and Thought
Te relationship between language and thought was much
discussed in the early days of psycholinguistics. Major issues
were the extent to which one is dependent upon the other and
the extent to which language shapes our perceptions of reality.
Developmental studies include accounts by Jean Piaget of the
impact of cognitive development upon language acquisition and
by Lev Vygotsky of how egocentric speech becomes internalized
as thought. Tere have been attempts by psycholinguists to put
the Sapir-Whorf hypothesis to the test, especially in the area of
color terms .
Conclusion
Research is active in all the areas of psycholinguistics that have
been profled. Te methodologies employed are diverse and (as
noted) include experimentation, observation, elicited response,
verbal report, and grammaticality judgments. However, the
future direction of research will be greatly infuenced by tech-
nological advances. Te most important recent innovation has
been the use of brain imaging to validate psycholinguistic theo-
ries about the operations underlying language performance.
Researchers are able to identify diferent types of brain event,
distinguishing, for example, one associated with syntactic func-
tions from one associated with semantic. A second develop-
ment lies in the increased use of eye-tracking equipment not
only to study reading but also to record the way in which a lan-
guage users attention is attracted to an object by semantic cues.
Computer science also makes a contribution. Psycholinguistics
employs computer modeling of linguistic performance, includ-
ing neural networks that simulate spreading activation, con-
nectionist word-recognition models based upon competition
between items, and connectionist learning models. It also draws
upon ongoing research into artifcial intelligence for example,
studies of expertise, artifcial speech recognition, or automatic
translation .
John Field
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aitchison , Jean . 2008 . Te Articulate Mammal . 5th ed. London : Routledge .
Readable general introduction to many areas.
. 2004 . Words in the Mind . 3d ed. Oxford : Blackwell . Accessible
overview of lexical storage and retrieval.
Clark , Eve V . 2003 . First Language Acquisition . Cambridge : Cambridge
University Press . Extensive account of the feld.
Deacon , Terrence . 1997 . Te Symbolic Species . New York : W. W. Norton .
In-depth account of neurolinguistics and language evolution.
Field , John . 2004 . Psycholinguistics: Te Key Concepts . London : Routledge .
Comprehensive reference guide for non-psycholinguists.
Gathercole , Susan E ., and A. D. Baddeley . 1993 . Working Memory and
Language . Hove, UK : Erlbaum . Key account of the part played by work-
ing memory.
Gleason , Jean Berko , and N. B. Ratner , eds. 1998. Psycholinguistics . Fort
Worth, TX : Harcourt Brace . Overviews of nine major areas.
Language Impairment
Psycholinguistics also studies the psychological factors that con-
tribute to language disorders . One type of dif culty investigated
is developmental : It includes problems of speech production,
such as stammering, as well as disorders of reading and
writing, such as dyslexia and dysgraphia . Another type is
acquired as a result of accident, stroke, or surgery and includes
aphasia , an impairment of the ability to produce and/or under-
stand speech. Also investigated are the efects of aging upon
language. In all of these areas, an important distinction has to be
made between loss of language from the mental store and dis-
ruption of the processes by which language is accessed.
Besides supporting the work of clinicians and therapists in
treating disorders, the study of language impairment feeds back
into other areas of psycholinguistics. Firstly, it sheds light on the
skills and cognitive processes that are essential to normal lan-
guage operations but are absent or obstructed in the patients
studied. Secondly, it afords possible insights into the process of
language acquisition. Research into cases of specific language
impairment (SLI) has sought evidence as to whether language is
or is not genetically transmitted. Research into the linguistic com-
petence of suferers from syndromes such as Downs, Williams,
and autism has sought evidence as to whether language forms
part of general cognition or is independent of it .
Second Language Acquisition (SLA)
As with frst language acquisition, there are two distinct branches
of inquiry. One group of researchers takes linguistic theory as a
point of departure, while the other draws its guiding principles
from cognitive psychology. Te linguistic group employs mod-
els of language (predominantly Chomskyan) on the premise that
they correspond to internalized rules in the mind of the user.
Teir research seeks evidence of the resetting of Chomskyan
parameters or compares the components of the second lan-
guage (L2) grammar with those of a native speaker. Research
in this tradition concerns itself with competence, rather than
performance; thus, there is considerable reliance upon gram-
maticality judgments as a source of evidence.
A more cognitive approach views the ability to perform in a
second language as an acquired skill and draws upon cognitive
theory relating to problem solving and to the development of
expertise . On this analysis, language learners begin with declara-
tive rule-based knowledge, which gradually becomes procedur-
alized . Separate steps in the process become combined, and
access to the rules becomes less and less subject to conscious
control and increasingly automatic. Recent thinking in SLA has
also been infuenced by emergentism , leading to a theory that
L2 profciency derives not from internalized rules but from mul-
tiple stored exemplars of linguistic encounters .
Researchers in SLA draw increasingly upon psycholinguistic
constructs such as memory, attention, and automaticity when
considering the cognitive demands imposed upon the L2 user.
Another approach employs psycholinguistic models of language
processing as descriptors of skilled L1 behavior and, thus, as tar-
gets for those teaching language skills.
Tere has been extensive psycholinguistic study of bilin-
gualism , focusing especially upon how a bilinguals two lan-
guages are stored: whether they are kept entirely apart, subserved
Psycholinguistics
690
questions, that research has typically not been informed by the
body of scholarship in literary studies on narrative structure and
form. Tus, psychonarratology is an attempt to bring the empir-
ical methods and analytical style of cognitive psychology to the
problems and domain of narratology.
Psychonarratology builds on previous research in the empiri-
cal study of literature, a feld frst designated by Sigfried Schmidt
( 1981 ), who developed some of its theoretical and philosophical
foundations. Pioneering empirical work in this efort included
research by A. C. Graesser et al. ( 1999 ), Willie van Peer and Max
Louwerse ( 2002 ), D. S. Miall and D. Kuiken ( 1995 ), R. J. Gerrig
( 1992 ), and D. Vipond and R. A. Hunt (1984). Current research
in the general area includes not only psychonarratology but also
research on a wide range of other topics, such as genre, emotion,
aesthetics and aesthetic appreciation, and flm and media, as
well as evolutionary and intercultural approaches .
Main Tenets of Psychonarratology
As developed in Psychonarratology by Bortolussi and Dixon, four
cascading proposals are made for the empirical study of reader
response to narratives. Te frst is a conceptual distinction
between textual features and reader constructions that allows a
suitable framing of empirical questions. Te second is the appli-
cation of statistical models for conceptualizing the nature of the
reader. Te third is a preference for the methodology of the tex-
tual experiment . And the fourth is a hypothesis concerning read-
ers representation of a conversational narrator.
FEATURE/CONSTRUCTION DISTINCTION. Psychonarratology makes
a crucial distinction between textual features , that is, those prop-
erties of the text that can be objectively identifed, and reader
constructions , or the mental representations of the reader.
Features are objective and can, in principle, be reliably identifed
by trained observers. Ideally, a textual feature could be specifed
by an explicit algorithm operating on the text. As a minimum,
there should be some consensus about an explicit defnition of
a feature and clear examples. In contrast, constructions are sub-
jective and can vary across readers as a function of specifc goals,
reading context, expectations, literary experience, and individual
reader characteristics. Logical and conceptual analysis is suf -
cient for understanding the former, but in order to draw infer-
ences about the nature of constructions, one requires empirical
evidence on real readers reading. According to the analysis of
the authors, many of the dif cult controversies in literary stud-
ies arise from a failure to carefully distinguish features and con-
structions and to apply appropriate methodologies to each .
THE NATURE OF THE READER. One of the greatest challenges fac-
ing scholars interested in literary reception, regardless of their
particular orientation, is the reconciliation of the need to make
general claims about readers (such as those entailed by the
concept of an ideal reader) with the fact that reader response is
seemingly variable and idiosyncratic. In psychonarratology, this
resolution is accomplished by applying the statistical concepts
of population and measurement distributions, an approach that
the authors refer to as the statistical reader .
A population is a particular group of people about which
one hopes to draw interesting conclusions, for example,
Levelt , Willem J. M . 1989 . Speaking . Cambridge, MA : MIT Press . Standard
comprehensive account.
Obler , Lorraine K ., and K. Gjerlow . 1999. Language and the Brain.
Cambridge : Cambridge University Press . Short but wide-ranging intro-
duction for non-specialists.
Perfetti , C . 1985 . Reading Ability . New York : Oxford University Press .
Infuential account of skilled reading.
PSYCHONARRATOLOGY
Defnition
Tis term was frst coined by P. Dixon and M. Bortolussi in 2001
in their chapter Prolegomena for a science of psychonarratol-
ogy and developed further by Bortolussi and Dixon in their
book Psychonarratology: Foundations for the Empirical Study
of Literary Response (2003) . It designates an interdisciplinary,
empirical approach to the study of literary response (see lit-
erature, empirical study of ) and the processing of narra-
tive. Generally, as the term suggests, it brings together two very
diverse felds: psychology and literary studies. However, while in
literary studies psychology typically refers to psychoanalysis ,
psychonarratology draws on cognitive psychology and discourse
processing for its methodology. Within literary studies, it looks to
narratology for its conceptual and theoretical insights about
narrative prose but also draws on related felds, such as recep-
tion and reader response theories .
Theoretical Background
Te conjoining of cognitive psychology and narratology has two
theoretical motivations. On the one hand, literary studies in gen-
eral, and narratology in particular, seemed to have reached an
impasse from which the traditional stock of methods provided no
means of escape. Numerous scholars had understood that texts
are just collections of letters, words, sentences, and paragraphs,
that meanings emerge in the interaction between texts and read-
ers, and that these interactions could be varied and complex. For
example, from within the phenomenological tradition, Roman
Ingarden conceived of literary texts as schematic structures that
needed to be concretized or completed by the reader, and W.
Iser, following in his footsteps, regarded literary texts as inde-
terminate objects that elicit gap-flling strategies on the part of
the reader. However, without appropriate methodologies for
assessing the cognitive processes of real readers, literary schol-
ars were restricted to making intuitive and purely speculative
assumptions about the efects of particular aspects of literature
on its readers. Tis gave rise to notions of an ideal reader , a time-
less, homogeneous entity possessing the ideal competence
required to process texts in some ideal fashion. Variants of this
ideal entity were expressed through a host of related terms: the
model reader, the super reader, and so on. Various branches of
literary studies, regardless of their specifc goals or scope, inevi-
tably came up against the same roadblock, namely, the need to
account for the efect(s) of texts on readers, without the required
tools for advancing beyond the purely hypothetical.
On the other hand, empirical research in discourse processing
generally neglected to address the processing of complex, real-
world narratives and literary narratives in particular. Moreover,
to the extent that cognitive psychology has investigated such
Psychonarratology
691
the representation of a conversational participant. In particular,
readers may draw inferences about the nature and mental state
of the narrator that are licensed by the assumption that the nar-
rator is rational and cooperative. Such inferences are analogous
to the conversational implicatures of H. Paul Grice and
may be referred to as narratorial implicatures by extension .
Research Findings of Psychonarratology
Although psychonarratology is intended primarily as a frame-
work in which to develop a scientifc understanding of liter-
ary response, the extant work provides some evidence for the
hypothesis of the conversational narrator. Two examples are
briefy described here.
In one study, Dixon and Bortolussi manipulated the narrative
technique used in the story Rope by Katherine Ann Porter . In
this story, virtually all of the narrative consists of an argument
between a husband and wife related in free indirect speech.
Diferent versions of the story were created in which the speech
of the man or the wife was changed to tagged direct speech. For
example, the following passage: Had he brought the cofee?
She had been waiting all day long for cofee. Tey had forgot it
when they ordered at the store the frst day was changed to: She
asked him, Did you bring the cofee? Ive been waiting all day
long for cofee. We forgot it when we ordered at the store the frst
day. When groups of readers read these diferent versions, they
tended to assume that the gender of the (absent) narrator was
the same as the character whose speech was related in the free
indirect style. Moreover, that character was seen as more rea-
sonable and rational in his or her arguments. Te results were
interpreted as demonstrating that the free indirect speech style
closely associates a character with the narrator and that the nar-
rator is, by default, assumed to be rational and cooperative, just
as a conversational participant would be.
In another study, diferent versions of the story Te Of ce
by Alice Monroe were constructed. Tis story begins with several
paragraphs describing the frst person narrators attitudes and
sensibilities, prior to any description of the story plot events. In
Dixon and Bortolussis analysis, this material appeared to have
a number of narratorial implicatures that might lead readers to
make assumptions or inferences regarding the nature of the nar-
rator. Moreover, because readers make such inferences by draw-
ing on their own experiences and knowledge, they may come
to see the narrator as more like themselves, in efect identifying
with the narrator. When material was added to the story in order
to make such inferences unnecessary, readers were less likely
to identify with the narrator and less likely to see the narrator as
reasonable and justifed in his or her actions. Tese results were
seen as supporting the view that readers treat narrators much as
they would a conversational participant .
Marisa Bortolussi and Peter Dixon
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bortolussi , M . and Dixon , P. 2003. Psychonarratology: Foundations
for the Empirical Study of Literary Response . Cambridge and New
York : Cambridge University Press .
Bruce , B . 1981 . A social interaction model of reading. Discourse
Processes 4 : 273311. A description of the levels of communication in
narrative discourse.
undergraduate students in Alberta and California or all gradu-
ates of a community college who have taken no more than one
literature course. One of the main tenets of psychonarratology is
that scientifc claims can only be made about a clearly delimited
population of readers. However, any given group of readers need
not be homogeneous and uniform; rather, each group may con-
sist of any number of overlapping subgroups that may or may not
difer with respect to interesting aspects of narrative response.
Because populations are potentially heterogeneous, there
is no necessity that general statements about reader response
apply equally to all individuals. Nevertheless, this does not mean
that it is impossible to describe reader response. Te solution
is to describe measurement distributions. With respect to any
given empirical measurement, one can describe the central ten-
dency that is, what the measurements have in common and
the variability that is, how the measurements vary over indi-
viduals. Tere are standard statistical and analytical techniques
for developing these kinds of descriptions, not only for numeri-
cal measurements (such as reading time or rating scales) but also
for more open-ended responses. Te scientifc goal is, thus, to
generate theories that relate the description of measurement dis-
tributions to the observed characteristics of readers and texts .
THE TEXTUAL EXPERIMENT. Te textual experiment provides a
methodology for understanding the relationship between tex-
tual features and reader constructions by systematically varying
those textual features. Te essence of this approach is to mea-
sure reader constructions using, for example, questionnaires
and rating scales, and to assess how those constructions vary as
a function of the features of the text. When a particular reader
construction (e.g., the perceived justifcation of a characters
actions) covaries with a particular textual feature (e.g., the use
of free indirect speech style, a mode of representing some of
the form of a characters enunciated speech or thought with-
out direct quotation), then one can conclude that the feature
causally contributes to the reader construction. An alternative
to textual experiments (in which the nature of the text is manip-
ulated) is to assess reader constructions of diferent texts sam-
pled from extant materials. However, any two sampled texts will
difer in a wide range of characteristics. As a consequence, it
is dif cult to draw causal inferences about which features are
related to particular reader constructions. Such inferences are
much more sound in a textual experiment in which features are
manipulated. In order to apply this method properly, though, it
is essential that only a single feature be varied, and that other,
potentially causal, features are not inadvertently changed at the
same time .
THE CONVERSATIONAL NARRATOR. A theoretical hypothesis that
unifes many of the authors research results is the idea that read-
ers represent the narrator as a conversational participant. Tis
follows from the theoretical advances of B. Bruce and others in
narratology in which the narrative is conceptualized as a com-
municative transaction between a narrator and a narratee. In the
psychonarratological version of this idea, readers are hypoth-
esized to develop a mental representation of an individual that
could have produced the words of the narrative and, in many
circumstances, treat that representation much as they would
Psychonarratology
692
of speech can be characterized this way. However, the term is
perhaps an unfortunate moniker for describing this approach
because auditory memory, attention, object recognition, cross-
modal processing, learning, plasticity, and development all play
important roles in processing complex auditory signals, and these
processes relate to speech processing as well. Te psychophysics
of speech might be more broadly described as an auditory cogni-
tive neuroscience approach to speech perception that considers
the richness of the acoustic (and, in fact, cross-modal) percep-
tual environment, the infuence of long-term experience, and the
efects of higher-order knowledge and processing.
Adherence to a general cognitive/perceptual account of
speech perception has meant that the psychophysical approach
to speech perception has played a central role in the theoreti-
cal debate about whether speech is perceived in a mode dis-
tinct from other acoustic signals. A major contribution of the
approach, apart from this theoretical debate, has been its insis-
tence on attention to the precise spectrotemporal characteristics
of the speech signal and to the neural mechanisms of auditory
processing involved in representing these signals.
An application of the psychophysical approach is observed
in the study of phonetic context efects. A great deal of early
research in acoustic phonetics documented the consid-
erable variability inherent in the acoustics of speech. To sum-
marize this broad literature, there do not appear to be acoustic
signatures that uniquely specify phonetic categories. Tus, lis-
teners are faced with the perceptual challenge of mapping highly
variable acoustic signals onto speech in a many-to-one manner.
Behavioral studies demonstrate that listeners appear to meet this
variability in speech acoustics by perceiving speech in a wholly
context-dependent manner. Many studies have reported phon-
etic context efects in which physically identical acoustic signals
are judged by listeners to be diferent speech sounds as a func-
tion of the phonetic context in which they are presented .
Phonetic context efects are ubiquitous in speech percep-
tion and have been documented for many speech segments.
Of interest in understanding the mechanisms that give rise to
such efects, Japanese quail ( Coturnix coturnix japonica ) that
were trained to peck a lighted key in response to presentation
of /g/ endpoints of a /ga/ to /da/ stimulus series pecked more
vigorously to novel ambiguous midseries speech stimuli when
they were preceded by /l/. A second set of birds trained to peck
in response to /d/ responded more robustly to the same novel
stimuli when they were preceded by /r/. Te directionality of the
efect is the same as for human listeners. Te extension of pho-
netic context efects to a nonhuman species suggests that general
auditory processing may assist in accommodating the complex
variability present in speech.
To pose the results in a general way, context sounds with
higher-frequency acoustic energy (like /l/) shift perception
of the following syllable toward the category alternative with
greater low-frequency energy, /g/. Tus, contrastive processes
by which the auditory system exaggerates change in the acoustic
signal may be suf cient to explain phonetic context efects. Tis
conclusion is supported by research demonstrating that adult
human listeners shift phonetic categorization responses not
only as a function of neighboring speech contexts but also as a
function of non-speech tones, chirps, and noises that precede or
Dixon , P , and Bortolussi , M . 2001 . Prolegomena for a science of psycho-
narratology. In New Perspectives on Narrative Perspective , ed. Will van
Peer and Seymour Chatman , 27588. Albany : State University of New
York Press . A brief introduction to the psychonarratology framework.
Gerrig , R. J . 1992 . Experiencing Narrative Worlds . New Haven, CT : Yale
University Press . An interdisciplinary attempt to examine what it means
to be transported mentally and emotionally by literary narratives.
Graesser , A. C , C. A. Bowers , B. Olde , K. White , and N. K. Person . 1999 .
Who knows what? Propagation of knowledge among agents in a liter-
ary storyworld. Poetics 26 : 143 75. A good example of the application
of discourse-processing methodology to a problem in literary reading.
Holub , R. C . 1989 . Reception Teory: A Critical Introduction .
London : Routledge . A very good, lucid, accessible introduction to the-
ories of literary reception.
Iser , W . 1974 . Te Implied Reader: Patterns of Communication in Prose
Fiction from Bunyan to Beckett . Baltimore : John Hopkins University
Press . A seminal work describing the mechanics of readertext inter-
actions; includes some key concepts such as the ideal reader, and
gaps.
Miall , D. S , and Kuiken , D . 1995 . Aspects of literary response: A new
questionnaire . Research in the Teaching of English 29 : 37 58. A semi-
nal efort to develop quantifable indices of literary expertise.
Peer , Willie van , and Max Louwerse , eds. 2002 . Interdisciplinary Studies
in Tematics . Amsterdam and Philadelphia : John Benjamins . Brings
together research on themes from a variety of felds in an attempt to
defne the concept of theme.
Rimmon-Kenan , S . 1983 . Narrative Fiction: Contemporary Poetics . New
York : Routledge . An accessible, solid introduction to narratological
basics.
Schmidt , Sigfried . 1981 . Empirical studies in literature: Introductory
remarks . Poetics 10 : 317 36. A pioneering study that lays the theoret-
ical groundwork for the empirical study of literature.
Singer , M . 1990 . Psychology of Language: An Introduction to Sentence and
Discourse Processes . Hillsdale, NJ : Lawrence Erlbaum . An introduction
to discourse processing and cognitive psychology.
Vipond , D ., and R. A. Hunt . 1984 . Point-driven understanding: Pragmatic
and cognitive dimensions of literary reading . Poetics 13 : 261 77. A
pioneering investigation into how literary reading varies with reading
strategy.
PSYCHOPHYSICS OF SPEECH
Te psychophysics of speech describes an interdisciplinary
approach to the understanding of speech perception . Te
approach considers speech as a complex acoustic signal sharing
much in common with other complex perceptual events and pos-
its that, as such, speech may be studied in the broader context
of general perceptual, cognitive, and sensorineural systems. Tis
approach is distinguished from those that consider speech to be
a special signal processed in a manner distinct from non-speech
sounds. Te essence of a psychophysical approach is to determine
the extent to which speech perception makes use of general cog-
nitive and perceptual processes before postulating mechanisms
specialized to the speech signal. Tus, understanding the psy-
chophysics of speech may include the utilization of animal mod-
els of auditory behavior and physiology to examine how much of
speech perception may be accounted for by general rather than
specialized mechanisms, and the relation of speech perception to
neural coding at peripheral and central levels of processing .
Psychophysics often connotes bottom-up or peripheral pro-
cessing, and, in fact, a great deal of research of the psychophysics
Psychophysics of Speech
693
natural language. On this view, there are four computational
resources available to a lexical item as part of its linguistic encod-
ing: lexical typing structure , argument structure ; event structure ;
and qualia structure . Tere are four possible qualia roles associ-
ated with a word:
(a) Formal : the basic category distinguishing the meaning of
word within a larger domain;
(b) Constitutive : the relation between an object and its con-
stituent parts;
(c) Agentive : the factors involved in the objects origins or
coming into being;
(d) Telic : the purpose or function of the object, if there is
one .
Tere are two general points that should be made concerning
qualia roles: 1) Every category expresses a qualia structure, and
2) not all lexical items carry a value for each qualia role. Te frst
point is important for the way a generative lexicon provides a
uniform semantic representation compositionally from all ele-
ments of a phrase. Te second point allows us to view qualia as
applicable or specifable relative to particular semantic classes.
In efect, the qualia structure of a noun determines its mean-
ing in much the same way as the typing of arguments to a verb
determines its meaning. Te elements that make up a qualia
structure include such familiar notions as container, space, sur-
face, fgure, or artifact. One way to model the qualia structure is
as a set of constraints on types (cf. Copestake and Briscoe 1992 ;
Pustejovsky and Boguraev 1993 ). Te operations in the compo-
sitional semantics make reference to the types within this sys-
tem. Te qualia structure, along with the other representational
devices (event structure and argument structure), can be seen as
providing the building blocks for possible object types.
Consider, for example, the qualia structure for the nouns beer
and sandwich , with formal (F), agentive (A), telic (T), and con-
stitutive (C):
a. beer . x:[F = liquid A = brew T = drink]
b. sandwich . x:[F = physical A = make T = eat C = bread,]
From qualia structures such as these, it now becomes clear how
a sentence such as Mary enjoyed her sandwich receives the
default interpretation it does, namely, that of Mary enjoying eat-
ing the sandwich. Similarly, for Mary fnished her beer, the
composition of the event-selecting aspectual verb fnish and its
object involves a rule that retrieves a possible event interpreta-
tion of drinking the beer. Tese are examples of type coercion,
where the compositional rules in the grammar make reference to
values such as qualia structure, if such interpretations are to be
constructed on-line and dynamically.
Te qualia structure of verbs characterizes the general role of
the subpredicates of a verbs event structure (as in Dowty 1979 ).
It also interacts with the aspectual category of the predicate. For
example, run and bake are process verbs, where the process
predicate is assigned to the agentive role, as in John ran and
Mary baked the potato.
a. run(x) P:[A = run_act(x)]
b. bake(x) P:[A = bake_act(x)]
follow speech. In the case of human and non-human perception
of speech and non-speech contexts, speech perception appears
to be relative to and contrastive with the acoustics of context
sounds, whether speech or non-speech.
Tis portfolio of research fndings is indicative of a psycho-
physical approach to speech perception in that it pays careful
attention to the spectrotemporal information available to listen-
ers, it makes use of nonhuman animals as a means of examining
the generality of the mechanisms available to speech process-
ing, and it examines the extent to which complex non-speech
signals may give rise to some of the same patterns of perception
as speech. Research relating the context-dependent coding of
acoustic signals to neural response (see phonetics and phon-
ology, neurobiology of ) adds to the understanding of how
phonetic context efects may arise from general characteristics of
the perceptual system. Te constellation of available results sug-
gests that general perceptual mechanisms play a role in phonetic
context efects .
In other domains, the psychophysical approach has contrib-
uted to the understanding of auditory representation, auditory
learning, and cross-modal processing as they relate to speech
processing. Tere remains much potential for understanding the
perceptual, cognitive, and neural underpinnings of speech com-
munication from a general perceptual/cognitive perspective .
Lori L. Holt
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Diehl , R. L. , and K. R. Kluender . 1989 . On the objects of speech per-
ception . Ecological Psychology 1 : 121 44. Discussion of a general
approach to speech perception.
Diehl , R. L ., A. J. Lotto , and L. L. Holt . 2004 . Speech perception . Annual
Review of Psychology 55 : 149 79. Tis review contrasts theoretical
approaches to speech perception.
Liberman , A. M . 1996 . Speech: A special code . Cambridge, MA : MIT Press .
Discussion of the motor theory of speech perception, an alternative
account to a general approach to speech perception.
Lotto , A. J ., and L. L. Holt . 2006 . Putting phonetic context efects into con-
text: A commentary on Fowler (2006) . Perception and Psychophysics
68 : 178 83. Briefy reviews phonetic context efects in literature.
Schouten , M. E. H ., ed. 2003. Te nature of speech perception . Speech
Communication 41 : 1 271. A collection of 20 papers on the psycho-
physics of speech perception.
Q
QUALIA ROLES
Qualia structure is a system of relations that characterizes the
semantics of a lexical item or phrase. Te notion of qualia struc-
ture is derived in part from the Aristotelian theory of explanation
(Moravcsik 1975 ). An important semantic concept within gen-
erative lexicon theory (GL), qualia roles are the major building
blocks for constructing word and phrasal meaning in a language
compositionally .
GL (Pustejovsky 1995 ) is a theory of linguistic semantics,
which focuses on the distributed nature of compositionality in
Qualia Roles
694
(4) Everyone wanted () to win: For every person x, x wanted
[x win]
Te same phenomenon appears in (5a) and (5b).
(5) a. John rescued himself = John rescued John
b. Someone rescued himself = For some x, x rescued x
Someone rescued someone
Tus, whereas anaphora with referential antecedents may
involve coreference , anaphora with quantifcational antecedents
involves binding ; anaphora and binding remain a major topic in
syntax and semantics .
The Semantics of Quantifcation
Te rise of formal semantics brought investigations into
the model-theoretic semantics of NPs and determiners. In
montague grammar , all English NPs, even proper names,
are generalized quantifers (Montague 1973), denoting sets of
properties of individuals. Tis uniform treatment launched
the study of the semantic properties of NPs and determiners
(Barwise and Cooper 1981; Keenan and Stavi 1986), leading to
progress on semantic universals of determiner meanings (see
semantics, universals of ), the semantics of existential
sentences and weak NPs (those that can occur in existential
sentences: a, some, three, no, many , but not the, every, both ,
most ), the semantics of determiners like any that can occur in
negative and certain other contexts but not in simple af rma-
tive sentences (the negative polarity phenomenon), and other
topics in quantifcation.
In the early 1980s, Irene Heim ([1982] 1989) and Hans Kamp
(1981) independently argued against Richard Montagues uniform
treatment of NPs, distinguishing defnite and indefnite NPs (with
determiners such as a, the, three, the three, some, several ) from essen-
tially quantifcational NPs ( every, all, most ). On their approaches,
an indefnite introduces a discourse referent into the context, bring-
ing context into semantics proper (see semantics-pragmatics
interaction ); only the essentially quantifcational NPs are
treated as generalized quantifers. Barbara H. Partee (1986) recon-
ciled Montagues uniform semantics with the Kamp-Heim theory
through type-shifting mechanisms such that all NPs can have
generalized quantifer-type meanings, but many NPs have refer-
ential and/or predicative meanings as well. Te king , for instance,
may have a quantifcational meaning (roughly, whoever is the
one and only king, with no presuppositions), a referential mean-
ing (referring to the unique king if there is one, failing to refer if
existence and uniqueness presuppositions are not satisfed), or a
predicative meaning in is the king , asserting of its subject that he is
the one and only king.
Other topics in the semantics of quantifcation include the
semantics of distributive, collective, and cumulative quantifca-
tion; the semantics of the mass-count distinction; event quantif-
cation and tense logic; generic sentences; implicit quantifcation;
and the binding of implicit variables. Tere is also active research
on childrens acquisition and adult processing of the syntax and
semantics of quantifcation.
Logicians have continued to make progress on the logic of
quantifcation, including work in game-theoretical semantics
(Hintikka and Sandu 1997; Clark 2007), where the foundations of
Tey can both, however, be coerced to accomplishments
(transitions) by specifying a termination predicate, assigned to
the formal role (cf. Pustejovsky 1995 ), for example, John ran to
the store, Mary baked a cake.
Recently, researchers in computational linguistics and lexi-
cography have adopted the notion of qualia roles as one orga-
nizing principle in the process of building resources for lexical
knowledge bases .
James Pustejovsky
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Copestake , A. and T. Briscoe . 1992. Lexical operations in a unifcation-
based framework. In Lexical Semantics and Knowledge Representation ,
ed. J. Pustejovsky and S. Bergler . Berlin : Springer-Verlag .
Dowty , D. 1979. Word Meaning and Montague Grammar . Dordrecht, the
Netherlands : Kluwer .
Moravcsik , J. M. 1975. Aitia as generative factor in Aristotles philoso-
phy . Dialogue 14 : 62236.
Pustejovsky , J. 1995. Te Generative Lexicon . Cambridge, MA : MIT Press .
Pustejovsky , J. , and B. Boguraev. 1993. Lexical knowledge representation
and natural language processing . Artifcial Intelligence 63 :193223.
QUANTIFICATION
Quantifcation has been a central concern in Logic and
Language at least since Aristotle, who systematized all valid and
invalid syllogisms involving the forms All/ some/ no/ not all As are
Bs (see Kneale and Kneale 1962). In linguistics, quantifcational
phenomena played a role in upsetting the architecture of standard
transformational grammar (Chomsky 1965) in which deep
structure determines semantic interpretation. Many transforma-
tions that were meaning-preserving on sentences involving refer-
ential terms were not so when applied to quantifers:
(1) John wanted [John win] John wanted to win
(2) Everyone wanted [everyone win] (?) Everyone wanted to
win
Te semantic inappropriateness of derivations such as (2) helped
to ignite the so-called linguistic wars (Newmeyer 1980) .
Quantifcation and the Syntax-Semantics Interface
Quantifcation raises issues for the syntax-semantics interface
concerning scope ambiguity , binding , and anaphora .
Teories difer sharply in the treatment of scope-ambiguous sen-
tences like (3), which challenge the otherwise plausible assump-
tion that every ambiguity involves a lexical ambiguity or an
ambiguity of syntactic structure; (3) on the face of it has neither.
(3) At least two students read every book.
i. Wider scope for at least two : Tere are at least two who read
the whole lot.
ii. Wider scope for every : Every book got at least two readings.
Te problem illustrated in transformation (2) is a problem
of binding and anaphora. It was soon recognized that pronouns
and null anaphors whose antecedent is a quantifer behave
like logical variables, as in (4), and not like repetitions of the
antecedent noun phrase (NP).
Quantifcation
695
Montague , Richard. 1973. Te proper treatment of quantifcation in ordi-
nary English. In Approaches to Natural Language , ed. K. J. J. Hintikka
et al., 22142. Dordrecht, the Netherlands : Reidel .
Newmeyer , Frederick . 1980. Linguistic Teory in America: Te First
Quarter-Century of Transformational Generative Grammar . New
York : Academic Press .
Partee , Barbara H. 1986. Noun phrase interpretation and type-shifting
principles. In Studies in Discourse Representation Teory and the
Teory of Generalized Quantifers , ed. J. Groenendijk et al., 11543.
Dordrecht, the Netherlands : Foris .
. 1995. Quantifcational structures and compositionality. In
Quantifcation in Natural Language , ed. Emmon Bach et al., 541602.
Dordrecht, the Netherlands : Kluwer . A formal semanticists approach
to semantic typology of quantifcational expressions.
Szabolcsi , Anna , ed. 1997. Ways of Scope Taking. Dordrecht, the
Netherlands : Kluwer .
QUANTITATIVE LINGUISTICS
While the formal branches of linguistics use the qualitative
mathematical means (algebra, set theory) and logics to model
structural properties of language, quantitative linguistics (QL)
studies the multitude of quantitative properties as the essential
basis for the description and understanding of the development
and functioning of linguistic systems and their components. Te
objects of QL research do not difer from those of other linguistic
disciplines. Te diference lies, rather, in the ontological points
of view (do we consider a language as a set of sentences with
their structures assigned to them, or do we see it as a system
that is subject to evolutionary processes in analogy to biological
organisms, etc.) and, consequently, in the concepts that form the
basis of the disciplines.
Diferences of this kind enable researchers to perceive new
phenomena in their area of study. A linguist accustomed to
thinking in terms of set theoretical constructs is not likely to fnd
the study of such properties as length, frequency, age, degree
of polysemy, and so on interesting or even necessary. Zipfs
Law is the only quantitative relation that almost every linguist
has heard about, but for those who are not familiar with QL,
it appears to be a curiosity more than a central linguistic law,
which is connected with a large number of properties and pro-
cesses in language. From a quantitative point of view, however,
it is quite natural to detect features and interrelations that can
be expressed only by numbers. Tere are, for example, depen-
dences of length (or complexity) of syntactic constructions
on their frequency and on their ambiguity ; of homonymy of
grammatical morphemes on their dispersion in their para-
digm; and of the length of expressions on their age, the dynam-
ics of the fow of information in a text on its size, the probability
of change of a sound on its articulatory dif culty, and so on.
In short, in every feld and on each level of linguistic analysis,
phenomena of this kind are signifcant. Tey are observed in
every language in the world and at all times. Moreover, it can
be shown that these properties of linguistic elements and their
interrelations abide by universal laws of language , which
can be formulated in a strict mathematical way in analogy to
the laws of the natural sciences. Emphasis has to be put on the
fact that these laws are stochastic; they do not capture single
cases (this would be neither expected nor possible) but, rather,
notions of scope, variable binding, and variable dependence are
being reexamined .
Typology of Quantifcation
Jon Barwise and Robin Cooper (1981) hypothesized that all
languages use NPs interpreted as generalized quantifers.
Research reported in Bach et al. (1995) identifed several lan-
guages that falsify that hypothesis. Other ways of expressing
quantifcation include adverbs of quantifcation like usually,
mostly (Lewis 1975), adjectives ( numerous, Russian mnogie
many), nouns ( majority, lot, dozen ), and verbal prefxes
(Evans 1995) .
Cross-linguistic studies show that languages difer with
respect to syntactic positions and strategies for expressing difer-
ent quantifcational notions (Szabolcsi 1997), with respect to the
degree to which surface structure constrains semantic quantifer
scope, with respect to the range of meanings a bare NP like
horses can have, in the variety and interpretation of indefnites
(Haspelmath 1997 ; Chung and Ladusaw 2003 ), in interactions
between nominal quantifcation and verbal aspect, and in other
ways that are still being explored .
Barbara H. Partee
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bach , Emmon, et al., eds. 1995. Quantifcation in Natural Languages .
Dordrecht, the Netherlands : Kluwer Academic .
Barwise , Jon , and Robin Cooper . 1981. Generalized quantifers and nat-
ural language . Linguistics and Philosophy 4 : 159219.
Chomsky , Noam. 1965. Aspects of the Teory of Syntax . Cambridge,
MA : MIT Press .
Chung , Sandra , and William A. Ladusaw. 2003. Restriction and Saturation .
Cambridge, MA : MIT Press .
Clark , Robin . 2007. Games, quantifers, and pronouns. In Game Teory
and Linguistic Meaning , ed. A.-V. Pietarinen , 20728 . Amsterdam :
Elsevier .
Evans , Nick. 1995. A-quantifers and scope in Mayali. In Quantifcation
in Natural Language , ed. E. Bach et al., 20770. Dordrecht, the
Netherlands : Kluwer .
Haspelmath , Martin. 1997. Indefnite Pronouns . Oxford : Oxford University
Press .
Heim , Irene. [1982] 1989. Te Semantics of Defnite and Indefnite Noun
Phrases . New York : Garland .
Hintikka , J. , and G. Sandu. 1997. Game-theoretical semantics. In
Handbook of Logic and Language , ed. J. van Benthem and A. ter
Meulen , 361410. Cambridge, MA : MIT Press .
Kamp , Hans. [1981] 1984. A theory of truth and semantic representa-
tion. In Truth, Interpretation, Information , ed. J. Groenendijk , T.
Janssen , and M. Stokhof , 141. Dordrecht, the Netherlands : Foris .
Keenan , Edward. 1996. Te semantics of determiners. In Te
Handbook of Contemporary Semantic Teory , ed. Shalom Lappin .
Oxford : Blackwell . A survey of the semantic properties of quantifer
expressions by a leader in the feld.
Keenan , Edward , and Jonathan Stavi . 1986. A semantic characteriza-
tion of natural language determiners. Linguistics and Philosophy
9 : 253326.
Kneale , William , and Martha Kneale . 1962. Te Development of Logic .
Oxford : Oxford University Press .
Lewis , David . 1975. Adverbs of quantifcation. In Formal Semantics
of Natural Language , ed. E. L. Keenan , 315. Cambridge : Cambridge
University Press .
Quantifcation Quantitative Linguistics
696
linguistics was conducted in particular by Michail V. Arapov
(Arapov 1988 ; Arapov and Cherc 1974), who based his models
of text dynamics and of language development on the analysis of
rank order. In Georgia, a group around Jurij K. Orlov established
a tradition of studies into the statistical structure of texts based
on the Zipf-Mandelbrot Law . Te Estonian researcher Juhan
Tuldava ( 1995 , 1998 ) is famous for his mathematical methods of
analysis of numerous text phenomena .
Objectives and Methods
Te fact that language cannot adequately be analyzed without
quantitative methods follows from a number of principal con-
siderations (cf. also Altmann and Lehfeldt 1980 , 1 f).
EPISTEMOLOGICAL REASONS. Te possibilities of deriving state-
ments about language(s) are seriously limited. Direct observa-
tion of language is impossible. Introspection cannot provide
more than heuristic contributions and does not possess the sta-
tus of empirical evidence (even if the contrary is often claimed in
linguistics). As a source of scientifc data, only linguistic behavior
is available in the form of oral or written text , in the form of
psycholinguistic experiments, and so on.
Furthermore, the situation is aggravated by the fact that we
never dispose of complete information on the object under
study. On the one hand, only a limited part or aspect of the object
is accessible (because it is principally infnite, such as the set of
all texts or all sentences, or because it cannot be described in full
for practical reasons). On the other hand, very often we lack the
complete information about the number and kinds of all factors
that might be relevant for a given problem and are, therefore,
unable to give a full description.
Only mathematical statistics enables us to fnd valid conclu-
sions in spite of incomplete information and, indeed, with objec-
tive, arbitrary reliability .
HEURISTIC REASONS. One of the most elementary tasks of any
science is to create some order within the mass of manifold,
diverse, and unmanageable data. Classifcation and correlation
methods can give indications to phenomena and interrelations
not known before. A typical example of a domain where such
inductive methods are very common is corpus linguistics ,
where huge amounts of linguistic data are collected and could
not even be inspected with the bare eye .
METHODOLOGICAL REASONS. Any science begins with categori-
cal, qualitative concepts, which divide the feld of interest into
delimited classes as clearly as possible. A linguistic example of this
kind of concept is the classical category parts of speech ( see word
classes ) . It is possible to decide whether a word should be consid-
ered as, for example, a noun or not. Every statement based on cat-
egories can be reduced to dichotomies (having exactly two values,
such as {true, false}, {1, 0}, {yes, no}). Tis kind of concept is funda-
mental and indispensable but insuf cient for a deeper insight.
Te possibility of gradual statements is provided by compara-
tive (ordinal-scale) concepts. Tey allow us to determine that an
object possesses more or less of a given property than another
one or the same amount of it. A linguistic example is the gram-
matical acceptability of sentences.
predict the probabilities of certain events or certain conditions
in a whole. It is easy to fnd counterexamples with respect to any
of the examples cited above. However, this does not mean that
they contradict the corresponding laws. Divergences from a sta-
tistical average are not only admissible but even lawful they
are themselves determined with quantitative exactness. Tis
situation is, in principle, not diferent from that in the natural
sciences, where the old deterministic ideas have been replaced
by modern statistical/probabilistic models .
Te role of QL is to unveil corresponding phenomena, to sys-
tematically describe them, and to fnd and formulate the laws
that explain the observed and described facts. Quantitative inter-
relations have an enormous value for fundamental research and
can also be used and applied in many felds, such as computa-
tional linguistics and natural language processing, teach-
ing language , optimization of texts, and so on .
Historical Background
Te frst scientifc counts of units of language or text were pub-
lished in the nineteenth century. Te frst theoretical insight, after
many years of merely descriptive counts of various kinds, was
ofered by the Russian mathematician A. A. Markov , who created
the base of the theory of Markov chains in 1913. Tis mathemat-
ical model of the sequential (syntagmatic) dependence among
units in linear concatenation in the form of transition probabil-
ities was despite its mathematical signifcance of only little
use for linguistics. In modern natural language processing, how-
ever, (hidden) Markov models are a central component of many
methods in language technology .
Later, quantitative studies of linguistic material were, in
the frst place, a consequence of practical demands: Eforts to
improve second language training (see bilingual education )
and optimization of stenographic systems are examples. Te
unveiled interrelations between frequency of words and the
ranks of the frequency class (alternatively: between frequency
and the number of words in the given frequency class) were sys-
tematically investigated by the aforementioned George Kingsley
Zipf . He was the frst to set up a model in order to explain the
observations and to fnd a mathematical formula for the cor-
responding function the famous Zipfs Law . Among his pub-
lications, his books ([1935] 1968 and 1949 ) are considered the
most important. Zipf formulated (in diferent terms) innovative
thoughts on self-organization, the principle of language econ-
omy, and fundamental properties of linguistic laws long before
modern systems theory arose. His ideas, such as the principle
of least efort and the forces of unifcation and diversifcation,
are still important today (even if they sufer from certain short-
comings). Later, his model was conceptually and mathematically
improved by Benot Mandelbrot (cf. Rapoport 1982 ), the origina-
tor of fractal geometry. Zipfs body of thought inspired various
scientifc disciplines and enjoys increasing exposure again.
C. E. Shannon and W. Weaver ( 1949 ) applied information the-
ory to linguistics without much success. Physicist Wilhelm Fucks
( 1955 ) was responsible for a turn toward theoretical consider-
ations in German QL. He studied, among others, word-length
distributions and various phenomena of language, literature,
and music. In France, Charles Muller ( 1973 , 1979 ) created a novel
approach for studying the vocabulary of a text. In Russia, Zipfan
Quantitative Linguistics
697
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Altmann , Gabriel , and Werner Lehfeldt. 1980. Einfhrung in die quanti-
tative Phonologie . Bochum, Germany : Brockmeyer .
Arapov , Michail V. 1988. Kvantitativnaja lingvistika . Moscow : Nauka .
Arapov , Michail, V. , and Maja M. Cherc. 1974. Matematieskie metody v
istorieskoj lingvistike . Moscow : Nauka .
Fucks , Wilhelm. 1955. Mathematische Analyse von Sprachelementen,
Sprachstil und Sprachen . Cologne : Westdeutscher Verlag .
Hebek , Ludek. 1997. Lectures on Text Teory . Prague : Oriental
Institute .
Khler , Reinhard , Gabriel Altmann , and Rajmund G. Piotrowski. 2005.
Quantitative Linguistik. Ein internationales Handbuch. Berlin : Walter
de Gruyter . Quantitative Linguistics: An International Handbook . New
York: Walter de Gruyter.
Muller , Charles. 1973. Elments de statistique linguistique. In
Linguistica, Matematica e Calcolatori. Atti del convegno e della prima
scuola internazionale Pisa 1970, 34978 . Florence : Zampolli .
. 1979. Langue franais et linguistique quantitative . Geneva : Slatkine .
Rapoport , A. , 1982. Zipfs Law revisited. In Studies on Zipfs Law ,
ed. Henri Guiter and Michail V. Arapov , 128. Bochum, Germany :
Brockmeyer .
Shannon , C. E. , and W. Weaver. 1949. Te Mathematical Teory of
Communication . Urbana : University of Illinois Press .
Tuldava , Juhan. 1995. Methods in Quantitative Linguistics . Trier,
Germany : WVT
. 1998. Probleme und Methoden der quantitative-systemischen
Lexikologie . Trier, Germany : WVT .
Zipf , George Kingsley . [1935] 1968. Te Psycho-Biology of Language: An
Introduction to Dynamic Philology . Cambridge, MA : MIT Press .
. 1949. Human Behavior and the Principle of Least Efort . Reading,
MA : Addison-Wesley .
R
RADICAL INTERPRETATION
Radical interpretation is one of the central concepts in the work
of the American philosopher Donald Davidson (19172003). For
Davidson, to interpret speakers means to understand their lin-
guistic utterances (cf. [1975] 1984, 157), to assign meanings to
their words and, in a slightly extended usage, assign contents to
their propositional attitudes. Radical interpretation takes place
in a specifc scenario in which a person encounters the speakers
of a completely unknown language L . Te radical interpreter has
the task of devising a formal semantic theory (see semantics )
for L on the basis of data of a very specifc kind: His or her evi-
dence consists entirely of (all available) data about the linguistic
and nonlinguistic behavior of the speakers of L in its observable
circumstances. According to Davidsons method, radically inter-
preting a language automatically includes systematic ascriptions
of belief to its speakers.
Background
Te basic scenario of a feld linguist trying to understand a
radically foreign language L on the basis of purely behavioral
data is introduced in W. V. O. Quines seminal work Word and
Object ( 1960 ). Here, the task is to construe a translation manual
for L, and Quine uses radical translation to consider how much
Te highest degree of order is achieved with the help of met-
rical concepts, which are needed if the diference between the
amounts of a given property possessed by two objects plays a
role. In this case, the values of the property are mapped to the ele-
ments of a set of numbers in which the relations between these
numbers correspond to the relations between the values of the
properties of the objects. In this way, specifc operations, such
as subtraction, correspond to specifc diferences or distances in
the properties between the objects. Tis enables the researcher to
establish an arbitrarily fne conceptual grid within his or her feld
of study. Concepts of this kind are called interval-scale concepts.
If a fxed point of reference is added (e.g., zero), ratio-scaled con-
cepts are obtained that allow the operation of multiplication and
division. Only the latter scale enables a formulation of how many
times object A has more than B of a property .
The Objectives of Quantitative Linguistics
In contrast to other branches of linguistics, QL emphasizes the
introduction and application of additional, advanced scien-
tifc tools. Principally, linguistics tries, in the same way as other
empirical sciences do in their felds, to fnd explanations for the
properties, mechanisms, functions, development, and so on of
language(s). Due to the stochastic properties of language, quan-
tifcation and probabilistic models play a crucial role in this pro-
cess. In the framework of this general aim, QL has a special status
only because it makes special eforts to care for the methods
necessary for this purpose. We can characterize this endeavor by
two complementary aspects:
1. On the one hand, the development and application of
quantitative models and methods are indispensable in all cases
where purely formal (algebraic, set-theoretical, and logical)
methods fail, that is, where the variability and vagueness of
natural languages (see language, natural and symbolic )
cannot be neglected, where gradual changes debar the applica-
tion of static/structural models. Briefy, quantitative approaches
must be applied whenever the dramatic simplifcation caused
by the qualitative yes/no scale is inappropriate for a given
investigation.
2. On the other hand, quantitative concepts and methods
are superior to the qualitative ones on principled grounds. Te
quantitative ones enable a more adequate description of real-
ity by providing an arbitrarily fne resolution. Between the two
extreme poles yes/no, true/false, 1/0 , as many grades as are
needed can be distinguished up to the infnitely many grades
of the continuum.
Generally speaking, the development of quantitative meth-
ods aims at improving the exactness and precision of the possible
statements on the properties of linguistic and textual objects. Tey
help us derive new insights that would not be possible without
them: Subjective criteria can be made objective and operational-
ized (e.g., in stylistics ); interrelations between units and prop-
erties can be detected, remaining invisible to qualitative methods;
and workable methods for technical and other felds of application
can be found where traditional linguistic methods fail or produce
inappropriate results due to the stochastic properties of the data or
to the sheer mass of them (e.g., in natural language processing) .
Reinhard Khler and Gabriel Altmann
Quantitative Linguistics Radical Interpretation
698
Task, Method, and Procedure
Davidsons answer to the frst question, the question of what
we could know that would allow us to interpret a language, is a
Tarskistyle theory of truth ( T-theory ). Such a theory, Davidson
argues, can be used as a formal semantic theory for a natural lan-
guage L ([ 1967 ] 1984). He is one of the main advocates of truth
conditional semantics , and a correct T-theory T for L com-
positionally specifes the meanings of the sentences of L : For
every sentence S of L , a T-sentence specifying its truth conditions
can be derived from the axioms and rules of T . For instance, a
correct T-theory for German would allow derivation of the fol-
lowing T-sentence from the axioms for the simple expressions
Schnee and ist weiss :
(S) Schnee ist weiss is true-in-German if and only if snow is white.
T-theories can be constructed for signifcant parts of natu-
ral language (for a list of problems such as belief sentences or
counterfactuals, see Davidson [ 1973 ] 1984, 132). Tey are sup-
posed to theoretically model semantic competence, but need not
be objects of knowledge for the speakers in any psychologically
realistic sense.
According to Davidson, the radical interpreter can devise a
T-theory for an unknown language L in two steps. On the basis
of their behavior, the interpreter can determine the sentences
of L that the speakers hold true in particular circumstances. Tis
amounts to detecting a propositional attitude: Holding a sen-
tence true is having a belief , but so long as the sentence held
true remains uninterpreted, no meaning-theoretical question
is begged. In the second step, the radical interpreter uses data
about the circumstances under which speakers hold sentences
true to determine their truth conditions.
Holding a sentence true, however, is a product of two fac-
tors: what the sentence means and what the speaker believes to
be the case (cf. Davidson [ 1973 ] 1984, 134). Assigning a mean-
ing to a sentence held true is ascribing a belief to the speaker.
Because of this interdependence of belief and meaning (p.
134), the ascription of belief needs to be restricted in relevant
ways if there is to be any evidence relation between holding true
and T-theory; so long as there is no such restriction, so long,
that is, as beliefs can be as absurd as the interpreter pleases, any
meaning can be assigned to any sentence. To establish such an
evidence relation is one of the main functions of the principle of
charity (see charity, principle of) (cf. Gler 2006 , 340). It
tells the radical interpreter to assign truth conditions to the sen-
tences of L such that the speakers of L have true and coherent
beliefs so far as that is plausibly possible. Since this can be done
only according to the interpreters own view of what is true,
application of charity amounts to agreement maximiza-
tion or, as Davidson prefers, agreement optimization between
speaker and interpreter. Te idea here is that some mistakes are
more destructive for understanding than others; an interpreta-
tion that avoids ascribing fagrant logical errors or very basic,
inexplicable perceptual mistakes is prima facie better than one
that does not. Ultimately, charity tells the interpreter to pick that
T-theory that stands in the relation of best ft (Davidson [ 1973 ]
1984, 136) to the totality of his or her data. In this way, the prin-
ciple of charity fulflls its second main function: It allows for a
ranking of T-theories in terms of how well they ft the totality
of language can be made sense of in terms of its stimulus condi-
tions, and what scope this leaves for empirically unconditioned
variation in ones conceptual scheme (Quine 1960 , 26). Among
other things, he uses radical translation to argue for the inde-
terminacy of translation . Tis is the claim that on the
basis of the evidence available in radical translation, diferent
but equally correct translation manuals can be set up between
two languages L
1
and L
2
, manuals diverging in a number of
places by translating a sentence of L
1
into sentences of L
2
which
stand to each other in no plausible sort of equivalence however
loose (p. 27). Davidson subscribes to indeterminacy for analo-
gous reasons, even though in a more limited form (cf. Davidson
[ 1973 ] 1984, [ 1974 ] 1984) .
Te radical interpretation scenario derives its signifcance
for the theory of meaning, both in Quine and in Davidson, from
the foundational claim that meaning is entirely determined by
observable behavior (cf. Quine 1960 , ix; Davidson 2005 , 56).
According to this weak semantic behaviorism, the data avail-
able in radical interpretation are the data ultimately and entirely
determining the meanings of the expressions of the language
interpreted. In contrast to Quine, this determination remains
nonreductive in Davidson, but it is nevertheless both epistemic
and metaphysical in nature: Te data available in radical inter-
pretation are not only the ultimate evidence on the basis of
which meanings can be known; they are what (metaphysically)
determines or constitutes linguistic meaning. According to
Davidson, this is an individualistic afair not essentially involv-
ing social convention or a shared language; even though there
cannot be such a thing as a solitary speaker, one who never had
contact with other speakers, what a speaker means by his or her
words on an occasion of utterance is determined solely by his or
per own (dispositions to) behavior (cf. [ 1975 ] 1984, [ 1982 ] 1984,
[ 1992 ] 2001).
According to semantic behaviorism, meaning is determined
on a nonsemantic basis by data that can be described without
using semantic concepts, such as meaning or propositional
content. Davidson motivates his particular version of semantic
behaviorism partly by appeal to the essential publicness of lan-
guage: Te semantic features of language are public features.
What no one can, in the nature of the case, fgure out from the
totality of the relevant evidence cannot be part of meaning
([1973] 1984, 135). According to Davidson, the relevant evidence
is evidence plainly available to an observer unaided by instru-
ments (1994, 127). Tis restriction on the evidence stems from
the claim that terms like meaning and language are theoretical
terms deriving their signifcance from occasions of successful
linguistic communication (which do not, typically, involve the
use of any instruments) .
Davidson thus transforms the basic question of the philo-
sophical theory of meaning, the question What is it for words to
mean what they do? ( 1984 , xiii), into two others: Given that we
can interpret the linguistic utterances of a speaker, what could
we know that would enable us to do this? How could we come
to know it? ([ 1973 ] 1984, 125). Radical interpretation addresses
the second of these questions and is supposed to show that there
is a method by which we can know meaning on the basis of the
nonsemantic evidence that, according to Davidson, determines
it (cf. 1994 , 127) .
Radical Interpretation
699
externalim currently dominating the theory of meaning and
content. However, foundational issues such as these remain
insuf ciently explored; because of the role the shared environ-
ment plays in radical interpretation, Davidson thought of him-
self as a social and physical externalist, though clearly not of the
mainstream kind (cf. 2001 ) . So long as a systematic compari-
son of these competing accounts of meaning determination
is lacking, it remains premature to simply write of semantic
behaviorism; prima facie, it is not even clear that Davidsonian
semantic behaviorism and mainstream (physical) externalism
are incompatible .
Kathrin Gler
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Davidson , Donald. [1967] 1984. Truth and meaning. In Davidson 1984,
1736.
. [1973] 1984. Radical interpretation. In Davidson 1984, 12539.
. [1974] 1984. Belief and the basis of meaning. In Davidson 1984,
14154.
. [1975] 1984. Tought and talk. In Davidson 1984, 15570.
. [1982] 1984. Communication and convention. In Davidson
1984, 26580.
. 1984. Inquiries into Truth and Interpretation . Oxford : Clarendon
Press .
. [1992] 2001. Te second person. In Subjective, Intersubjective,
Objective , 10722. Oxford : Clarendon .
. 1994. Radical interpretation interpreted . Philosophical
Perspectives 8 : 121 8.
. 2001. Externalisms. In Interpreting Davidson , ed. P. Kotatko,
P. Pagin, and G. Segal, 116. Stanford, CA : CSLI .
. 2005. Truth and Predication . Cambridge, MA : Belknap .
Fodor , Jerry , and E. Lepore. 1994. Is radical interpretation possible?
Philosophical Perspectives 8 : 101 19.
Gler , Kathrin. 2006. Te status of charity: I. Conceptual truth or apos-
teriori necessity? International Journal of Philosophical Studies
14 : 337 60.
Lepore , Ernest , and K. Ludwig. 2005. Donald Davidson: Meaning, Truth,
Language, and Reality . Oxford : Clarendon .
Lewis , David. 1974. Radical interpretation . Synthese 23 : 331 44.
Pagin , Peter. 1997. Is compositionality compatible with holism? Mind
and Language 12 : 11 33.
Quine , Willard V. O. 1960. Word and Object . Cambridge, MA : MIT Press .
Ramberg , Bjrn. 1989. Donald Davidsons Philosophy of Language ,
Oxford : Blackwell .
Rawlings , Pierce. 2003. Radical interpretation. In Donald Davidson , ed.
K. Ludwig, 85112. Cambridge : Cambridge University Press .
READING
Reading is the process of decoding and comprehending written
language. Decoding, the conversion of written forms into lin-
guistic messages, is central to this defnition to the extent that
the comprehension of written language shares its features with
the comprehension of spoken language.
Reading connects printed information conveyed in a writing
system with the readers knowledge of the language encoded by
that system. Writing systems vary in their mapping principles, in
their implementation in a particular language (the orthography),
and in their visual appearance (the script). Alphabetic writing
systems map graphic units to phonemes . Syllabary systems,
of the data, a ranking such that the best theory is the correct
one (cf. Gler 2006 , 342). Tis amounts to a form of semantic
holism; the principle of charity determines all the meanings of
the expressions of L together, and on the basis of the totality of
the evidence (cf. Pagin 1997 , 13, 18). indeterminacy, then, is
the claim that there can be more than one best T-theory for
any given natural language L .
Davidson provides what he calls a crude outline ([ 1973 ]
1984, 136) for the process of devising a T-theory on the basis of
data about holding true attitudes. It has three steps: First, the
logical form of the sentences of L is determined. Use of a
T-theory as a formal semantic theory requires paraphrasing the
expressions of L in the language of frst-order quantifed logic
(plus identity). Te relevant evidence for this frst step consists
of sentences that are held true (or false) under all circumstances
(candidates for logical truth or falsity) and of patterns of infer-
ence, that is, sentences held true on the basis of other sentences
held true.
Te second step focuses mainly on sentences containing
indexicals , expressions whose interpretation depends on fea-
tures of the context, such as I or here. Take the sentence It
is raining or its German equivalent Es regnet . Teir truth value
varies with easily observable circumstances in the environment
of the speaker. Te idea (according to Davidson [ 1973 ] 1984, 135)
is to take data of the form
(E) Kurt belongs to the German speech community and Kurt
holds true Es regnet on Saturday at noon and it is raining near
Kurt on Saturday at noon as evidence for a T-sentence of the
form
(R) Es regnet is true-in-German when spoken by x at time t if
and only if it is raining near x at t .
Together, these two steps signifcantly limit the possibilities for
interpreting the predicates of L . Te third step deals with the
remaining sentences of L .
Questions and Criticism
Over the years, there has been extensive discussion of radi-
cal interpretation and the underlying Davidsonian philoso-
phy of language. Davidsons views on meaning determination
have been criticized as verifcationist (see also verifiability
criterion ) or idealist, charges he was keen on refuting. His
semantic individualism and holism have been issues of debate.
With respect to the principle of charity, such questions as
whether it overrationalizes empirical speakers or illegitimately
imposes our logic on alien speakers have been raised. Te most
important philosophical issues here concern the epistemic
and metaphysical status of the principle, and the questions
of whether and how it can be justifed. Whether radical inter-
pretation is possible, what kind of argument is required for its
possibility, and the precise role it plays in Davidsons philos-
ophy of language, as well as its wider signifcance, are topics
on which there is no general consensus among the commen-
tators. Today, many philosophers of mind and language are
of the opinion that the basic semantic behaviorism charac-
terizing both Davidsons and Quines philosophy of language
has been superseded by the (social and physical) meaning
Radical Interpretation Reading
700
wrong for an individual who also has a problem in some other
area (Stanovich and Siegel 1994 ) .
Dual route models allow two diferent sources of word-read-
ing dif culties: Either the direct route or the indirect phonologi-
cal route can be impaired (Coltheart et al. 1993 ). Surface dyslexics
have trouble with exception words, explained as selective dam-
age to the direct route. Phonological dyslexics have trouble with
regular words and pseudowords, explained by selective damage
to the indirect phonological route. A diferent view from single
mechanism models is that only phonological dyslexia is the
result of a processing defect. Surface, or orthographic, dyslexia is
a delay in the acquisition of word-specifc knowledge (Harm and
Seidenberg 1999 ), which comes through experience .
Reading Comprehension
Reading comprehension shares linguistic and cognitive pro-
cesses with spoken language and correlates highly with it among
adults (Gernsbacher 1990 ). Because this correlation is based on
the use of equivalent texts across listening and reading, it may
miss the diferences between ordinary spoken language and typ-
ical written texts that arise from their divergent syntactic struc-
tures, lexicons, and other aspect of their diferent registers. While
reading comprehension strongly depends on listening compre-
hension, reading and speech each place specifc demands on
comprehension processes .
Reading comprehension processes begin with word identif-
cation and include context-relevant selection of word mean-
ings and parsing (see parsing [human]) , the basic process of
extracting grammatical relations among words in a sentence .
Beyond these word- and sentence-level basics, higher-level com-
prehension involves readers constructions of mental models
of text information. One mental model is based closely on the
language of the text, and another, the situation model, refects
what the text is about (van Dijk and Kintsch 1983 ). Te reader
builds a situation model from the linguistically based model (the
text base) by combining knowledge sources through additional
inference processes. A situation model may contain nonlinguis-
tic representations, including spatial imagery (Glenberg, Kruley,
and Langston 1994 ) and the temporal organization of events
(Zwaan 1996 ), among others. Reading multiple texts that refer to
the same situation challenges the construction of a single situ-
ation model (Perfetti, Rouet, and Britt 1999 ) and requires addi-
tional comprehension skills in document use and evaluation
(Rouet 2006 ) .
Because texts are never fully explicit, comprehension
research has had an enduring interest in inferences. Inferences
that link anaphora (e.g., pronouns) with their antecedents to
establish coreference are a routine part of comprehension. Te
extent of elaborative and predictive inferences (Graesser, Singer,
and Trabasso 1994 ) is more in doubt (McKoon and Ratclif 1992 ).
For example, the sentence Te American tour group went to
London for its annual holiday may evoke an inference that the
group traveled by airplane, but whether a reader actually makes
this inference appears to be highly variable. Inferences about
causeefect relations may be more likely than other kinds of
elaborative inferences (Trabasso and Suh 1993 ) .
Comprehension skill is highly variable. Some children
appear to have a comprehension-specifc problem (i.e., without
represented by Japanese Kana, map graphic units to spoken
language syllables. Chinese is usually classifed as logographic
because its graphic units (characters) correspond primarily to
morphemes . However, the fact that characters have compo-
nents that provide syllable-level pronunciation information jus-
tifes an alternative designation of morpho-syllabic (DeFrancis
1989 ) .
Written Word Identifcation
Visual processing of a letter string results in the activation of the
grapheme units (individual and multiple letters) of words. In rep-
resentational (or symbolic) models of reading, words are repre-
sented in the readers mental lexicon . Successful word reading is
a match between the graphic input and the corresponding word
representation. Phonological units are also activated and play an
important role in securing identifcation.
In dual route models of reading, identifcation occurs along
two pathways, a direct route to the word identity and an indi-
rect route through phonological units (Coltheart et al. 1993 ).
Te direct pathway must be used for exception words (e.g.,
iron) for which an indirect phonological route would fail and
may also function for any word that becomes highly familiar.
Te phonological route must be used to read pseudowords
(e.g., nufe ) for which there is no lexical representation and
may be used for words with regular grapheme-to-phoneme
patterns. Single-route connectionist models simulate
these two pathways with a single mechanism that learns how
to read letter strings on the basis of experience (Plaut et al.
1996 ). Alternative models use dynamic resonance mecha-
nisms to capture interactions between existing states and new
inputs (Van Orden and Goldinger 1994 ). In a resonance model,
the patterns of graphic-phonological activation stabilize more
rapidly than do patterns of graphic- semantic activation, sim-
ulating a word-identifcation process that brings rapid conver-
gence of orthography and phonology, with meaning slower to
exert an infuence .
In studies of nonalphabetic reading, research has over-
turned the idea that reading Chinese involves only meaning
and not phonology (Perfetti, Liu, and Tan 2005 ). Even when
single-character words are read silently for meaning, the pro-
nunciation of the character appears to be activated. Tis role
of phonology, where the writing system does not require it,
may refect a universal phonological principle (see phonol-
ogy, universals of ) that is grounded in spoken language.
Nevertheless, neuroimaging studies of the brains imple-
mentation of word reading show diferences as well as similari-
ties between alphabetic and nonalphabetic reading (Siok et al.
2004 ; see also writing and reading, neurobiology of ).
It is interesting to note that English-speaking adults learning to
read Chinese show brain activation patterns that partly overlap
those shown by native Chinese speakers, suggesting that prop-
erties of the writing system recruit specifc brain areas (Perfetti
et al. 2007 ).
Individuals with word-identifcation problems are said to
have a specifc reading disability, or dyslexia, provided they
also show a discrepancy between reading and achievements in
other domains. However, the processes that go wrong in a spe-
cifc disability may not be much diferent from those that go
Reading
701
Frith , Uta. 1985. Beneath the surface of developmental dyslexia.
In Surface Dyslexia: Neuropsychological and Cognitive Studies
of Phonological Reading , ed. K. E. Patterson , J. C. Marshall , and
M. Coltheart , 30130. London : Erlbaum .
Gernsbacher , Morton. A. 1990. Language Comprehension as Structure
Building . Hillsdale, NJ : Erlbaum .
Glenberg , Arthur M., P. Kruley. and W. E. Langston. 1994. Analogical
processes in comprehension: Simulation of a mental model. In
Handbook of Psycholinguistics , ed. M. A. Gernsbacher , 60940.
San Diego, CA : Academic Press .
Gough , Philip B. , and M. L. Hillinger. 1980. Learning to read: An unnatu-
ral act . Bulletin of the Orton Society 20 : 179 96.
Graesser , Arthur C. , M. Singer , and T. Trabasso. 1994. Construction
inferences during narrative comprehension . Psychological Review
101 : 371 95.
Harm , Michael , and M. Seidenberg. 1999. Phonology, reading acquisi-
tion, and dyslexia: Insights from connectionist models . Psychological
Review 106 : 491528.
McKoon , Gail , and R. Ratclif. 1992. Inference during reading .
Psychological Review 99 : 440 66.
Morais , Jose , L. Cary , J. Alegria , and P. Bertelson. 1979. Does awareness
of speech as a sequence of phones arise spontaneously? Cognition
7 : 323 31.
Nation , Kate. 2005. Reading comprehension dif culties. In Snowling
and Hulme 2005, 24865.
Nation , Kate, and M. Snowling. 1999. Developmental diferences in
sensitivity to semantic relations among good and poor compre-
henders: Evidence from semantic priming. Cognition 70 : B113.
Oakhill , Jane V. , and A. Garnham. 1988. Becoming a Skilled Reader .
Oxford : Blackwell .
Perfetti , Charles A. 1992. Te representation problem in reading acquisi-
tion. In Reading Acquisition , ed. P. B. Gough , L. Ehri , and R. Treiman ,
14574. Hillsdale, NJ : Erlbaum .
Perfetti , Charles A. , and L. Hart. 2001. Te lexical basis of comprehen-
sion skill. In On the Consequences of Meaning Selection , ed. D. Gorfein ,
6786. Washington, DC : American Psychological Association .
Perfetti , C. A. , Y. Liu , J. Fiez , J. Nelson , and D. J. Bolger. 2007. Reading in
two writing systems: Accommodation and assimilation in the brains
reading network. Bilingualism: Language and Cognition 10 .2: 13146.
Special issue on Neurocognitive approaches to bilingualism: Asian
languages, edited by P. Li.
Perfetti , Charles A. , Y. Liu , and L. H. Tan. 2005. Te lexical constituency
model: Some implications of research on Chinese for general theories
of reading . Psychological Review 12 .11: 43 59.
Perfetti , Charles A. , J-F. Rouet , and M. A. Britt. 1999. Toward a the-
ory of documents representation. In Te Construction of Mental
Representations during Reading , ed. H. Van Oostendorp and
S. Goldman , 99122. Mahwah, NJ : Erlbaum .
Plaut , David C. , J. L. McClelland , M. S. Seidenberg , and K. Patterson. 1996.
Understanding normal and impaired word reading: Computational
principles in quasi-regular domains . Psychological Review
103 : 56 115.
Rayner , Keith. , B. R. Foorman , C. A. Perfetti , D. Pesetsky , and
M. S. Seidenberg. 2001. How psychological science informs the
teaching of reading. Psychological Science in the Public Interest
2 . 2: 3174. A supplement to Psychological Science. Reviews research
on reading and the acquisition of reading skills, connecting it to read-
ing education.
Read , Charles , Y. Zhang , H. Nie , and B. Ding. 1986. Te ability to
manipulate speech sounds depends on knowing alphabetic reading .
Cognition 24 : 31 44.
Rouet , Jean-Francois. 2006. Te Skills of Document Use . Mahwah,
NJ : Erlbaum .
a decoding problem) that is general across reading and spoken
language (Nation and Snowling 1999; see also disorders of
reading and writing ). Te potential causes for comprehen-
sion problems include failures to make inferences during read-
ing (Oakhill and Garnham 1988 ) and limitations in working
memory functions, among other factors (Nation 2005 ). Unstable
knowledge of word form and meaning (low lexical quality ) also
contributes substantially to comprehension problems (Perfetti
and Hart 2001 ) .
Learning to Read
In an alphabetic writing system, a child learns that letters and
strings of letters correspond to speech segments. For English,
this process is complicated by inconsistent orthography at the
letter-phoneme level, for example, the contrasts between choir
and chore and head and bead . Most European languages tend to
be coded by orthographies that more consistently map graph-
emes to phonemes, and learning to read refects this fact; for
example, childrens errors refect letter-to-phoneme decoding
procedures more than in English (Wimmer and Goswami 1994 ;
see also childrens grammatical errors ).
Important for the alphabetic principle is phonemic aware-
ness (see phonological awareness ), the explicit under-
standing that the speech stream can be segmented into a set of
meaningless units (phonemes). Childrens phonemic awareness
correlates with early reading success, and phoneme segmenta-
tion instruction produces gains in reading. However, alphabetic
literacy experience itself afects awareness of phonemes, as
shown by studies of adults without exposure to alphabetic writ-
ing (Morais et al. 1979 ) and of Chinese who learned to read prior
to the introduction of the alphabetic Pinyin system (Read et al.
1986 ) as well as by longitudinal results that show a bidirectional
relation between phonological sensitivity and literacy (Perfetti
1992 ).
Teories of learning to read have usually referred to a series
of stages (Ehri 1991 , 2005 ; Frith 1985 ; Gough and Hillinger 1980 ).
Alternative theoretical accounts emphasize the incremental
acquisition of decodable lexical representations and the role of
phonology to establish word-specifc orthographic representa-
tions (Perfetti 1992 ; Share 1995 ; see also writing and read-
ing, acquisition of ) .
Charles Perfetti
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Coltheart , Max , B. Curtis , P. Atkins , and M. Haller. 1993. Models
of reading aloud: Dual-route and parallel-distributed-processing
approaches . Psychological Review 100 : 589 608.
Coltheart , Max , K. Rastle , C. Perry , R. Langdon , and J. Ziegler . 2001. Te
DRC model: A model of visual word recognition and reading aloud .
Psychological Review 108 : 204 25. Explains and defends the dual route
theory and its evidence.
DeFrancis , John. 1989. Visible Speech: Te Diverse Oneness of Writing
Systems . Honolulu : University of Hawaii .
Ehri , Linn C. 1991. Learning to read and spell words. In Learning
to Read: Basic Research and Its Implications , ed. L. Rieben and
C. A. Perfetti , 5773. Hillsdale, NJ : Erlbaum .
. 2005. Development of sight word reading: Phases and fndings.
In Snowling and Hulme 2005, 13555.
Reading
702
Gerrig ( 1993 ). Russian Formalists proposed that a literary work
has aspects of fabula and siuzhet , often translated as story
and plot. Te fabula is a story structure: time-ordered events
in the story world. William Brewer and Ed Lichtenstein ( 1981 )
suggested that the siuzhet may best be called the discourse struc-
ture : the ordered set of speech-acts of the writer to the general
reader or listener to prompt the story mentally into being. Oatley
suggested that two further aspects are necessary: One is the sug-
gestion structure , the associations set of by the story idio-
syncratically in individuals. Te other is the realization structure,
the complete mental performance as realized in the mind of the
reader or audience member. Te matter was well put by Virginia
Woolf ( 1957 , 174):
Jane Austen is thus a mistress of much deeper emotion than
appears upon the surface. She stimulates us to supply what is
not there. What she ofers is, apparently, a trife, yet is composed
of something that expands in the readers mind and endows
with the most enduring form of life scenes which are outwardly
trivial.
Te relationship among the four aspects of a piece of liter-
ary prose or poetry can be illustrated by the diagram in Figure 1
(from Oatley 2002 , 45). Te implication of the layout of the dia-
gram is that the event structure starts of a story, usually by means
of a setting, that the discourse structure and suggestion structure
occur simultaneously, and that the realization structure is a
resultant of the other processes .
Keith Oatley
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Brewer , William , and Ed Lichtenstein. 1981. Event schemas, story sche-
mas and story grammars. In Attention and Performance. Vol. 9 . Ed.
J. Long and A. Baddeley , 36379. Hillsdale, NJ : Erlbaum .
Share , David L. 1995. Phonological recoding and self-teaching: Sine qua
non of reading acquisition . Cognition 55 : 151 218.
Siok , Wai T. , C. A. Perfetti , Z. Jin , and L. H. Tan. 2004. Biological abnor-
mality of impaired reading constrained by culture: Evidence from
Chinese . Nature 431 (September 1): 71 6.
Snowling , Margaret J., and C. Hulme , eds. 2005. Te Science of Reading: A
Handbook . Oxford : Blackwell . A compendium of reviews across the
entire spectrum of reading-related research.
Stanovich , Keith E. , and L. S. Siegel. 1994. Phenotypic performance pro-
fle of children with reading disabilities: A regression-based test of the
phonological-core variable-diference model . Journal of Educational
Psychology 86 .1: 24 53.
Trabasso , Tom , and S. Suh. 1993. Understanding text: Achieving explan-
atory coherence through online inferences and mental operations in
working memory . Discourse Processess 16 : 3 34.
Van Dijk , Teun A. , and W. Kintsch. 1983. Strategies of Discourse
Comprehension . New York : Academic Press .
Van Orden , Guy C. , and S. D. Goldinger. 1994. Te interdependence of
form and function in cognitive systems explains perception of printed
words . Journal of Experimental Psychology: Human Perception and
Performance 20 : 1269 91.
Wimmer , Heinz , and U. Goswami. 1994. Te infuence of orthographic
consistency on reading development: Word recognition in English and
German children . Cognition 51 : 91 103.
Zwaan , Rolf. 1996. Processing narrative time shifts . Journal of
Experimental Psychology: Learning, Memory and Cognition
22 .5: 1196 1207.
REALIZATION STRUCTURE
Tis term was coined by Keith Oatley ( 2002 ) to indicate how
one experiences a piece of literature. One does not just receive
or interpret it but realizes it, bringing it into being. Te idea that
fction involves such a realization or mental performance of the
piece by the reader or audience member was discussed in phi-
losophy by Wolfgang Iser ( 1974 ) and in psychology by Richard J.
Event Structure
The events of the story world.
A creation of the author.
Discourse Structure
The text as written by the author, or the
drama as performed. Much of this structure
is in the form of instructions to the reader
or audience as to how to construct the story.
Suggestion Structure
Nonliteral aspects, suggested by the text,
based on the readers or watchers share
of knowledge, experience, emotions, and
ideas.
Realization Structure
The enactment in the mind of the reader or
watcher, which results from the structure
process and suggestion structure
being applied to the discourse structure.
Figure 1.
Realization Structure
703
Such a relation to authority is often the sort of case that writ-
ers on zheng ming had in mind. In his famous treatment of the
rectifcation of names, Hsn Tzu (Xunzi) lamented the result of
verbal confusions wherein the distinction between the noble
and the humble is not clear and similarities and diferences are
not discriminated. When this occurs, ideas will be misunder-
stood and work will encounter dif culty or be neglected ( 1963 ,
125). A similar practical concern is found in Mozi (Mo Tzu), but
with diferent political consequences. Tus, Mozi wrote against
those who distinguish names in the world in such a way as to
promote distinctions and discriminations. He favored those
who distinguish names in the world in such a way as to love
others and beneft others by advocating no discrimination
(quoted in Zhang 2002 , 327).
Te practical, political consequences of the rectifcation of
names are connected with the eight steps set out in the classic
Great Learning. Te steps explain how to manifest ones virtue
by bringing order to a state. One brings order to ones state by
bringing order to ones family. One accomplishes this by devel-
oping oneself. Tat, in turn, results from rectifying ones mind/
heart, which derives from integrating ones thoughts. To do the
latter, one must extend ones knowledge, which is itself efected
by research. (On the eight steps, see Great Learning and Zhang
2002 , 452. Note that the list of steps inverts their order in prac-
tice, with research, therefore, being the frst step.) Knowledge is
at least in part a matter of knowing the right words and apply-
ing them properly. Tus, research is itself in part a matter of the
rectifcation of names. In research, one should seek the relevant
principles and connect them with names so that they will be
understood. Moreover, these principles, and our knowledge of
these principles, are not solely a matter of conceptualization.
As Zhang writes, names should match the reality, and reality
includes the actual use of the object ( 2002 , 424). Zhang is refer-
ring to a Mohist idea. But the link between principle and practice
is much more widespread.
Te eight steps also suggest the manner in which zheng
ming has consequences. In a given situation, our application of
a particular name afects our knowledge of the situation (Step
2). Moreover, it brings our understanding of the situation into
a complex of relations with other ideas and names (or words)
in such a way as to change the way our thoughts are integrated
(Step 3) and our heart or feeling is oriented (Step 4). We may
return here to the previous example. If I follow the lead of the
U.S. government and characterize the U.S. presence in Iraq as
liberation, then I understand its consequences in certain ways.
I also integrate this idea with other aspects of my thought
about the insurgents, about U.S. foreign policy, and so on. Tis,
in turn, has consequences for my emotional response to the
U.S. presence in Iraq, to the U.S. government, and so on ulti-
mately with consequences for my support of that government.
Te consequences are very diferent if I follow the lead of most
Iraqis and categorize the presence as occupation. Determining
which is correct, and using that term consistently in public dis-
cussion, is an instance of research leading to the rectifcation
of names, then to the integration of thought and orientation of
feeling .
Patrick Colm Hogan
Gerrig , Richard J. 1993. Experiencing Narrative Worlds: On the
Psychological Activities of Reading . New Haven, CT : Yale University
Press .
Iser , Wolfgang. 1974. Te Implied Reader: Patterns of Communication
in Prose Fiction from Bunyan To Beckett . Baltimore : Johns Hopkins
University Press .
Oatley , Keith. 2002. Emotions and the story worlds of fction. In
Narrative Impact: Social and Cognitive Foundations , ed. M. Green ,
J. Strange , and T. Brock , 3969. Mahwah, NJ : Erlbaum .
Woolf , Virginia . 1957. Te Common Reader: First Series . London : Hogarth
Press .
RECTIFICATION OF NAMES (ZHENG MING)
Te rectifcation of names is the adjustment of language to ft the
world, specifcally insofar as this bears on action. It is an impor-
tant concept in traditional Chinese philosophy. In the Analects ,
Confucius established the place of this concept in political
thought in particular. Tzu-lu asked Confucius what he would put
frst if he took over state administration. Confucius responded
that it would be the rectifcation of names, stressing the ill con-
sequences if names are not correct (xiii.3). (Tis is not to say
that all schools of thought shared this view. Some downplayed
the importance of language; see Zhang 2002 , 4758) .
Tere has been considerable disagreement among writers in
the Chinese tradition as to just what the rectifcation of names
involves (see Zhang 2002 , 46174). Approaching the topic from
the Western philosophical tradition, the obvious interpretation
is that meanings should be in accord with essences (see essen-
tialism and meaning) . For example, water should be used to
refer to H
2
O. Tere is an element of this in the various Chinese
schools that have debated the topic. Specifcally, a range of writ-
ers suggest that a rectifed term is in accord with li , or principle.
Unfortunately, there is no greater agreement on the meaning of
li than on the meaning of zheng ming . Abstracting somewhat
from the debates, however, we may infer that a principle is what
underlies and unifes a set of otherwise apparently diverse phe-
nomena. Put diferently, it is what manifests itself in the phenom-
enal patterns. Tis is related to the Western notion of essence.
However, it is not identical. For example, the rule complex gov-
erning plural formation in English should count as a principle
in this sense. It produces a patterned set of apparently diverse
phenomena (cats, dogs, bushes, and so on), defning the
unity of those phenomena. Moreover, understood in this way, a
principle may apply not only to real objects but also to ideals.
When a principle underlies an ideal, it defnes a norm .
Tis normative part is crucial. Consider, for example, the
presence of U.S. troops in Iraq. Some speakers, such as repre-
sentatives of the U.S. government, characterize this as liberation
(Chomsky 2006 , 131). Others, including most Iraqis, characterize
it as occupation (Chomsky 2006 , 163). Tese terms difer not only
descriptively but also evaluatively. Tey difer in the way they
apply norms to the situation, and thus in their implications for
appropriate response to that situation. In a case such as this, the
use or misuse of a term is signifcant because it has practical con-
sequences. Specifcally, it bears on our acceptance or rejection of
a particular structure of authority and of particular representa-
tives of authority.
Rectifcation of Names (Zheng Ming)
704
being able to represent the relationship between the repre-
sentation and what it refers to: to understand that a picture of
Niagara Falls stands for that visual scene, or that someones
belief that Santa Claus exists represents that potential state of
the world, or that rocks , the noun, refers to a set of stone objects.
Metarepresentation requires recursive embedding of represen-
tational relationships, but it is not identical to recursive embed-
ding (Stone and Gerrans 2006 ). Metarepresentation may also be
uniquely human (Suddendorf 1999 ) .
Marc Hauser, Noam Chomsky, and W. T. Fitch ( 2002 ) have
ofered the hypothesis that recursion is the defning feature of
language, making it uniquely human. Other features of language,
however, do not follow directly from recursion and also seem to
be uniquely human, such as words, fne phonemic discrimina-
tions, and motor control of mouth, larynx, and so on (enumer-
ated in Pinker and Jackendof 2005 ; Parker 2006 , Chapter 5).
Whether recursion is the single defning feature of language
or not, it might be uniquely human. Testing for recursive capac-
ity directly is dif cult. Instead, researchers rely on demon-
strations of animals ability to do tasks dependent on explicit
recursion. Some claim that animals do implicit recursion in
certain tasks, for example, ants doing dead reckoning, but this is
dif cult to substantiate. Although recursion is an ef cient solu-
tion to many problems, unless one can test for the explicit con-
tent of the recursive steps in a computation, it is always possible
that animal brains solve problems using some other, nonrecur-
sive computational technique . Tus, comparative research uses
tasks believed to depend on explicit recursion: mathematics,
theory of mind, problem solving involving interdependent steps,
mental time travel, or certain kinds of syntax (Corballis 2003 ;
Parker 2006 ). So far, no study has demonstrated that our closest
relatives, great apes, can do any of these tasks with the range and
fexibility of humans (Corballis 2003 ; Hauser 2005 ; Suddendorf
2006 ). For now at least, recursion can join a set of possibly unique
human cognitive capacities: other aspects of language, fexible
control of attention and inhibition, expanded working memory
capacity, and metarepresentation (Suddendorf 1999 ; Kawai
and Matsuzawa 2000 ; Hauser 2005 ; Pinker and Jackendof 2005 ;
Stone and Gerrans 2006 ). Recursion may not be the key to unique
human cognition, but it is no less worthy of study for being one
of many keys .
Valerie Stone
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Anderson , A. K. 2007. Recursion in programming. Available online
at: http://www.usableresults.com. Retrieved January 15, 2007.
Corballis , M. C. 2003. Recursion as the key to the human mind. In
From Mating to Mentality , ed. K. Sterelny and J. Fitness , 15571. New
York : Psychology Press .
Hauser , M. 2005. Our chimpanzee mind . Nature 437 .7055: 60 3.
Hauser , M. , N. Chomsky , and W. T. Fitch. 2002. Te faculty of lan-
guage: What is it, who has it, and how did it evolve? Science
298 .5598: 1569 79.
Kawai , N. , and T. Matsuzawa . 2000. Numerical memory span in a chim-
panzee . Nature 403 .6765: 39 40.
Parker , A.P. 2006. Evolution as a constraint on theories of syntax. Ph.D.
thesis, University of Edinburgh. Available online at: http://www.ling.
ed.ac.uk/~annarp. Retrieved February 4, 2007.
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chan , Wing-tsit. 1963. A Source Book in Chinese Philosophy . Princeton,
NJ : Princeton University Press .
Chomsky , Noam. 2006. Failed States: Te Abuse of Power and the Assault
on Democracy . New York : Metropolitan Books .
Confucius. 1979. Te Analects . Trans. D. C. Lau. New York : Penguin .
Great Learning . 1963. In Chan 1963, 8494.
Hsn Tzu. 1963. Te Hsn Tzu [selections]. In Chan 1963, 11535.
Zhang Dainian. 2002. Key Concepts in Chinese Philosophy . Trans. and ed.
Edmund Ryden. New Haven, CT : Yale University Press .
RECURSION, ITERATION, AND METAREPRESENTATION
Some researchers pinpoint recursion as our species key compu-
tational ability, making humans cognitively unique (e.g., Hauser,
Chomsky, and Fitch 2002 ; Corballis 2003 ). It may give us many
abilities hypothesized to be uniquely human: language, theory of
mind, complex problem solving, mathematics, and mental time
travel (episodic memory/future planning) (Hauser, Chomsky,
and Fitch 2002 ; Corballis 2003 ; Parker 2006 ; Stone and Gerrans
2006 ).
Within psychology and linguistics, recursion is understood as
a property of certain types of representations. Whether internal
to the mind or external, representations that can contain other
representations of the same type are recursive. Language, men-
tal states, mathematical formulas, and spatial representations all
have this property. One can have a thought about someone elses
belief about another persons thoughts, or one can have a pic-
ture of a picture of a picture: Both are recursive representations .
Recursive processing requires that recursive representations be
unpacked in a systematic way, from the highest to lowest level, in
order to produce some output.
Recursion is distinct from the related concept, iteration , but
the two are often confused. Both involve repetition. In program-
ming, iteration is the repetition of a process within a computa-
tion, with a top-level control structure that sees all the steps
involved (Anderson 2007 ). In recursion, however, the number of
steps is unknown to the highest level of the function; all that is
known to that level is whether an end condition has been satisfed
or whether the problem needs further breaking down (Anderson
2007 ; Suh 2007 ). In language, we can construct infnitely long
sentences by iterating elements, for example, I have lived in the
U.S. and England and Australia and Each iterative phrase is
independent, not requiring reference to the other phrases, only
to the top-level clause containing the phrases (Parker 2006 ).
We can also construct infnitely long sentences by using recur-
sively embedded elements, for example, Te blogger said that
Bush thought that Cheney thought that Libby believed that the
reporter did not know that Plame was a spy. Tese elements
are not independent, requiring full unpacking of each embed-
ded level to understand the full meaning of the sentence. Each
level of embedding refers to another level: One cannot know the
semantic value of Cheney thought that without knowing the
semantic value of the clauses it includes .
Recursion should also be distinguished from the related
concept metarepresentation . Some use the terms interchange-
ably, using metarepresentation to mean a representation of a
representation (e.g., Corballis 2003 ). Metarepresentation means
Recursion, Iteration, and Metarepresentation
705
argument, the referent of F is the function that maps the refer-
ent of X onto the referent of F ( X ). Tus, the referent of a complex
expression is always the result of applying the referent of one of its
constituents, as a function, to the referents of its other constitu-
ents, taken as arguments. Te referent of a sentence as a whole is
identifed with its truth value. Tus, the referent of Chomsky is
Chomsky, the referent of is clever the function that maps each
object x onto truth if x is clever and onto falsehood otherwise, and
the referent (truth value) of Chomsky is clever is truth if and
only if Chomsky is clever.
It may seem surprising that the referent of a sentence is its
truth value, but it should be kept in mind that reference is used as
a technical concept within compositional semantics. Given the
use to which the concept is put, this is not an unnatural assump-
tion: Frege was interested in a compositional semantics that
would tell us how the truth values of sentences are determined
by the referents of their parts, and all natural languages have
fragments in which, when a sentence has other sentences as parts,
the truth value of the whole depends only on the truth values of
the constituent sentences. Fragments of languages in which this
is the case, and in which the referent of a complex expression
in general depends only on the referents of its parts, are called
extensional . Tus, in an extensional fragment, expressions hav-
ing the same referent can be substituted in any sentence with-
out altering its truth value (contexts in which such substitutions
preserve truth value are also called extensional). Frege was pri-
marily interested in constructing a semantics for the language
of mathematics, which is extensional, and so choosing truth
values as referents of sentences was natural. However, natural
languages as wholes are not extensional. In contexts involving
propositional attitudes, modality , and counterfactuals,
the substitution of clauses having the same truth value may alter
the truth value of the whole sentence. To account for such con-
texts, Frege held that each sentence or other expression has, in
addition to a referent, another kind of semantic value, which he
called the expressions sense ( Sinn ). Te sense of a sentence is
what he called a thought , or, in contemporary terms, a propo-
sition . In order to maintain a version of the principle of com-
positionality , he held that the truth values of nonextensional
sentences are determined in part by the senses of their constitu-
ents (see sense and reference ).
For various reasons, Freges approach is now considered anti-
quated. Most recent work in formal semantics for natural lan-
guages is inspired by Alfred Tarskis work on the defnability of
truth for formal languages. Richard Montague ( 1974 ) was the
frst to apply Tarskis work productively to (fragments of) natural
languages. Here, extension is the preferred term. Te extension of
a predicate is, again, the set of things to which it applies. Although
terminology varies, in this framework, too, one can speak of the
extension of almost any expression, including a sentence, so that
one identifes a sentences extension with its truth value. Applying
Tarskis approach, the aim is to recursively characterize not only
the truth conditions of sentences but also the entailments (logi-
cal consequences) for a language using the notion of extension: A
sentence S
1
is said to entail a sentence S
2
in language L if and only
if there is no assignment of extensions to the semantically simple
expressions of L (no model of L) under which S
1
is true and S
2
false. On this approach, the logical constants difer from other
Pinker , S. , and R. Jackendof . 2005. Te faculty of language: Whats spe-
cial about it? Cognition, 95 .2: 201 36.
Stone , V. E. , and P. Gerrans . 2006. Whats domain-specifc about theory
of mind? Social Neuroscience 1.3/4 : 309 19.
Suddendorf , T. 1999. Te rise of the metamind. In Te Descent of Mind ,
ed . M.C. Corballis and S. E. G. Lea , 21860. Oxford : Oxford University
Press .
. 2006. Foresight and evolution of the human mind . Science,
312 .5776 : 1006 7.
Suh , E. 2007. Recursion and interation. Available online at: http://www.
aihorizon.com/essays/basiccs/general/recit.html. Retrieved from AI
Horizon February 5, 2007.
REFERENCE AND EXTENSION
Extension and reference are technical terms in the philosophy of
language, formal semantics, and pragmatics. We outline their
roles in three types of theoretical efort compositional seman-
tic theories (which make use of both terms), various theories of
reference , which purport to tell us what it is for a word to have a
certain referent, and views that understand reference as some-
thing people do with words. Te frst two are semantic accounts;
the last conceives of reference as a matter of language use, so of
pragmatics.
Reference and Extension in Compositional Semantics
We begin with the use of reference and extension in composi-
tional semantic theories. In this domain a referent is generally a
thing that a proper noun refers to or names, and an exten-
sion a set of objects to which a predicate applies (the term deno-
tation is sometimes used interchangeably with both reference
and extension). However, compositional semanticists often gen-
eralize one or the other notion so that almost any kind of expres-
sion, including a sentence, can be said to have a referent or an
extension.
With few exceptions, compositional semantic accounts are
versions of truth conditional semantics attempts to
specify the meanings of sentences in terms of their truth con-
ditions. Since natural languages allow for infnitely many sen-
tences, the truth conditions of sentences must be specifed
recursively in terms of the semantic values of their parts, and
referents and extensions are semantic values that enable us to
do just this. For example, we can specify the truth condition of
the sentence John smokes in terms of the referents and exten-
sions of its parts as follows: John smokes is true if and only if
the referent of John (namely, John himself) is a member of the
extension of smokes (the set of things that smoke) .
Te primary historical source for compositional semantics
along these lines is Gottlob Freges ([ 1892 ] 1997) account of
Bedeutung often translated as reference (also as denota-
tion). In it, a referent is assigned to every meaningful expres-
sion. Frege assumed that each complex expression is the result
of combining a functional expression (such as a predicate) with
one or more arguments (such as names) (see predicate and
argument ). Further, he assumed that the referent of a func-
tional expression F is always a function f , and that the referent
of any expression X that F accepts as an argument is the sort
of object that is among the arguments of f . Specifcally, if F is a
functional expression and X an expression that F accepts as an
Reference and Extension
706
maintain uniformity to ensure rigidity in reference at least
with names. In fact, speakers paraphrasing Wittgenstein
play all sorts of games with language.
In taking reference as a form of action and treating refer as
a verb, we come closest to the commonsense idea of a person
referring to or talking about an object. Critical work on
Bertrand Russells analysis of definite descriptions by
P. F. Strawson ( 1950 ) and Keith Donnellan ( 1966 ), as well as
H. Paul Grices work on the semantics/pragmatics distinction,
inspired a distinction between speakers reference and semantic
reference , the latter being central to both compositional seman-
tics and theories of reference of the sorts considered in the pre-
vious section. Te speakers reference of an expression, on an
occasion, is whatever object that speaker uses the expression to
pick out, typically in order to assert (query, etc.) something about
that object. You may use the phrase the man drinking a mar-
tini to refer to a certain person, although the person you have in
mind is, unbeknownst to you, drinking water: He is not, then, the
semantic referent of the phrase.
Some writers hold that semantic reference either does not
exist (Strawson 1950 can be read this way), or if it is to sus-
tain theoretical investigation must be reconceived (Chomsky).
Chomsky points out that natural language use (not in-house use
of the symbols of mathematics or natural science, where practi-
tioners constrain their actions) displays creativity, where this
is thought of in terms not only of the uncaused production of
novel expressions but also of their free use for any number of
purposes (appropriateness). Because referring is a form of
free action and cannot sustain naturalistic study, Chomsky pro-
poses the elimination of the semantic study of natural languages
as usually conceived (ofering theories of wordworld relation-
ships), placing the study of reference in a part of pragmatics
that resists theoretical investigation, and placing the study of
what he calls meaning (a psycholinguistic version of Fregean
senses) in syntax broadly conceived as the study of the intrinsic
properties of the mind/brain. Te study of meaning semantics
reconceived becomes a psycholinguistic enterprise focusing
on the natures of mind-internal elements, such as lexical items,
their semantic features, and the computations in which they
fgure. Chomsky ( 2000 , 38f) points out that this kind of study
might employ a theoretical device called relation R, construed
as a postulated relationship between theoretically defned
expressions and objects in some introduced, stipulated domain.
Relation R is not reference outside the head that is not apt
for naturalistic study. Relation R and the domain D are, rather,
construed to be part of syntax theoretical devices aiding the
naturalistic study of syntax conceived as language in the head.
Te members of D could be stipulated to be semantic values.
Tis might allow Chomsky to absorb the insights of Montague
and other developing theories within formal semantics into
syntax. It would also emphasize a view Chomsky maintains for
other reasons: Semantic compositionality is syntactic computa-
tion. Whether absorbing formal semantic accounts of composi-
tionality in this way suits the intuitions and aims of those who
want their semantic eforts to provide explications of truth con-
ditions is another matter .
James McGilvray and Juhani Yli-Vakkuri
expressions in that they are not assigned extensions/referents
(see logic and language ).
Montagues approach difers notably from Freges: It does
not assign senses to expressions to account for nonextensional
contexts. Instead, it employs the tools of possible worlds
semantics to this end. One can, however, within Montagues
framework defne objects corresponding roughly to Freges
senses: Te sense of an expression could be thought to corre-
spond to the function that maps each possible world onto the
extension that the expression has in that world. Such functions
are often called intensions (see intension and extension ) .
Theories of Reference, New and Old
Te second set of theories (often called theories of reference)
in which the terms reference and extension are found appear in
the works of philosophers of language who aim to describe and
explain the wordworld relations that compositional semantic
theories of the sort discussed previously take for granted.
In this area, too, a classical source is Frege. According to
Frege, a word has a specifc referent because its users associate it
with a particular sense something like a conceptual represen-
tation of its referent. Applied to proper names, his view was that
a name, say, George W. Bush, is associated by its users with a
certain descriptive condition, say, being the 43d president of the
United States , and that its referent is that object (if any) which
uniquely satisfes this condition.
Another view, the so-called new theory of reference (in vogue
since the 1970s), maintains that at least some expressions do
not have senses, but simply refer. Proper names are paradigm
examples. According to the approach, what cements the relation
between a name and its referent is not a mediating conceptual
representation in the speakers mind but a causal and historical
relationship between the names user and its referent. Te idea,
articulated by Kripke (1972), is that a name is introduced by an
initial baptism, which involves a causal interaction between a
speaker and the referent itself, and reference for all other speakers
is preserved in chains of communication in which each speaker
intends to use the name to refer to the same object as those from
which he or she acquired the name. Extending Kripkes view in
ways suggested by Kripke himself, Hilary Putnam ( 1975 ) pro-
posed baptism + history as an account of how natural kind
terms come to have and maintain their extensions (see essen-
tialism and meaning ) .
Reference as Action
Te third view we discuss maintains that reference depends
essentially on individual speakers (and possibly interpreters)
with variable interests: An appropriate slogan might be Words
dont refer; people do. One root of this view is found in the
work of the later Ludwig Wittgenstein, another in Descartes.
Tose who defend it point out that it is dif cult to fnd cases
of uniform wordworld relationships in the use of natural
languages. Tey grant that the practices of mathematicians
display uniformity, but these practices aside, reference varies
with time, context, speakers interests, and so on. Tey also
grant that some who ofer theories of reference, such as Kripke
(1972), acknowledge a role for speaker intentions. But Kripke
and others incorrectly assume that ordinary speakers desire to
Reference and Extension
707
grammatical markings, such as agreement morphology: subject-
verb agreement , person , number , and gender (see gender
marking) agreement; the switch-reference system (Amele); or
topic/subject markers (Japanese). Discourse analysis (see dis-
course analysis [linguistic]) fnds that reference usage
follows the constraints of information fow: Te grammatical
subject of a transitive clause tends to be coded with a pronoun
in English, or zero anaphora in Chinese or Japanese, to present
given or accessible information ( the light subject constraint ), for
instance, He in the example, whereas the grammatical object
tends to be an NP carrying new information (e.g., one plant). Tis
allows easy processing of accessible information early in an utter-
ance, while the rest of the utterance introduces the new referent,
thus facilitating reference tracking and discourse processing .
Experimental studies fnd that the discourse pattern of a
language engenders specifc reference tracking strategies in its
native speakers. Terefore, speakers of diferent languages may
develop diferent cognitive strategies to track reference during
discourse processes .
Liang Tao
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chafe , Wallace. 1994. Discourse, Consciousness, and Time: Te fow
and Displacement of Conscious Experience in Speaking and Writing .
Chicago : University of Chicago Press .
Kintsch , Walter. 1988. Te role of knowledge in discourse compre-
hension: A construction-integration model . Psychological Review
95 : 163 82.
Tao , Liang , and Alice Healy . 2005. Zero Anaphora: Transfer of Reference
Tracking Strategies from Chinese to English . Journal of Psycholinguistic
Research 34 : 99 131.
REGISTER
Speakers of a language use diferent words and grammatical
structures in diferent communicative situations. For example,
we do not use the same words and structures to write an aca-
demic term paper that we would use when talking to a close
friend about weekend plans.
Researchers study the language used in a particular situation
under the rubric of register : a language variety defned by its situ-
ational characteristics, including the setting, interactiveness, the
channel (or mode) of communication, the production and pro-
cessing circumstances, the purposes of communication, and the
topic.
Although registers are defned in situational terms, they can
also be described in terms of their typical linguistic characteris-
tics; most linguistic features are functional and, therefore, they
tend to occur in registers with certain situational characteristics.
For example, frst and second person pronouns ( I and you ) are
especially common in conversation. Speakers in conversation
talk a lot about themselves, and so they commonly use the pro-
noun I . Tese speakers also interact directly with another person,
often using the pronoun you .
Tere are many studies that describe the characteristics of
a particular register, such as sports announcer talk (Ferguson
1983 ), note taking (Janda 1985 ), classifed advertising (Bruthiaux
1996 ), and scientifc writing (Halliday 1988 ). Other researchers
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Chomsky , Noam. 2000. New Horizons in the Study of Language and Mind .
Cambridge : Cambridge University Press .
Donnellan , Keith. 1966. Reference and defnite descriptions.
Philosophical Review 75 .3: 281 304.
Frege , Gottlob. [1892] 1997. On Sinn and Bedeutung. In Te Frege
Reader , ed. M. Beaney . Oxford : Blackwell .
Kripke , Saul. 1972. Naming and necessity. In Semantics of Natural
Language , ed. Donald Davidson and Gilbert Harman, 253355, 76369.
Dordrecht, the Netherlands: Reidel.
. 1976. Speakers reference and semantic reference. Midwest
Studies in the Philosophy of Language 2 : 25576.
. 1980. Naming and Necessity . Cambridge : Harvard University
Press .
Montague , Richard. 1974. Formal Philosophy . New Haven and
London : Yale University Press .
Putnam , Hilary. 1975. Te meaning of meaning. In Language, Mind,
and Knowledge , ed Keith Gunderson, 13193. Minneapolis : University
of Minnesota Press .
Strawson , P. F. 1950. On referring. Mind 59 .235: 320 44.
REFERENCE TRACKING
Reference tracking, or ANAPHORA resolution, concerns how
language users track who or what the speaker is referring to in
discourse. Because everyday language use generally concerns
who does what to whom, reference tracking is important in
studying human language and cognition.
Anaphora devices include noun phrases (NPs), pronouns,
and zero anaphora, whose identities depend on their anteced-
ents in discourse. In the example Isabel went to China, and this
volunteer/she helped with midwifery training, Tis volunteer
is an NP that refers back to its antecedent Isabel. It could be
replaced by the pronoun she or by an empty slot (zero anaphora)
as in the sentence Isabel went to China and ____ helped with
midwifery training.
Pronouns and zero anaphora give less explicit information
than full NPs. Still, the reader/hearer benefts from the ef ciency
of these devices in conveying information that has been intro-
duced/given in the prior discourse or can be accessed/inferred
from the context. Tese devices are crucial for global cohesion
and local coherence in discourse. Experimental studies fnd that
without a specifc need, replacing a pronoun with an NP for given
information may hinder understanding .
A discourse topic provides a basic means for tracking the
identity of a pronoun or zero anaphora because the topic tends
to recur as given information continuously. Cross-linguistic
studies fnd that people can track the identity of a pronoun or
zero anaphora even when its referent is not in the immediately
preceding clause but in the prior context. Terefore, although
language production may be linear due to human physical limi-
tations, language processing and reference tracking are hierar-
chical cognitive processes.
Reference tracking requires the hearer to make inferences
from world knowledge about likely events, especially for lan-
guages that have no morphological markings (Chinese) yet
allow abundant zero anaphora, as in He grew only one plant,
but ___ blossomed well. Many languages (e.g., French, German,
and Turkish) make reference tracking easier with specifc
Reference Tracking Register
708
(with its allomorphs) is productive in English, not just because it
is frequent and ordered, but because it is the internalized pattern
that speakers automatically employ when a new verb, such as
supersize enters the language. We all agree that the past is super-
sized and not, for instance, * supersoze . Regularization is studied
diachronically (see synchrony and diachrony ) as well as in
contemporary settings; and psycholinguistic research inves-
tigates and attempts to account for regularization processes,
particularly for paradigmatic regularization (the regularization
of infectional paradigms), in the language of individual speak-
ers. It is easier to agree on what has happened in the language
than on the underlying psycholinguistic processes in the minds
of speakers.
Tere is controversy among linguists as to whether speak-
ers rely on abstract morphophonological algorithms or on more
general analogical processes when they infect or derive words.
Difering models of mental representation and cognitive func-
tioning can explain the productivity that ultimately makes regu-
larization both possible and likely.
Diachronic Evidence of Regularization
historical linguistics reveals ways that regularization
occurs over time. Modern English refects regularization in the
infectional system that started hundreds of years ago. Over time,
English changed from a highly infected language to a more ana-
lytic language, and the infections that remained were largely
regularized, with some notable exceptions in the verb system.
Old English had diferent infectional paradigms for strong verbs
(verbs that changed their vowels to form a past) and weak verbs
(verbs that maintained their stem and added a past tense suf x).
A typical kind of paradigmatic regularization occurred in Middle
English when many strong verbs like helpan to help, which had
various past forms ( healp, hulpe, hulpon, hulpen ), began to fol-
low the pattern of the more regular weak verbs, leading to just
one past form, helped. Much of this took place in Middle English
between the years 1150 and 1500 (Baugh 1957 , 189). In early
Middle English, plurals formed with -s were common, but there
were also many plurals formed with -en. During the fourteenth
century, the -s pattern became the regular plural, and today,
of the tens of thousands of nouns in Modern English, we can
point to very few ( children, oxen, brethren) that form the plural
with -en . Many irregular words became regularized during this
period or fell into disuse, thus making English a more transpar-
ent and consistent language in general. In particular, many Old
English strong verbs simply disappeared while some others were
changed into weak or regular verbs. Tus, over a period of sev-
eral hundred years, various kinds of regularization took place in
English. Regularization was not omnipresent, however: Te most
frequent verbs in English (such as to be, to have, and to do ) have
roots in Old English and are irregular. Highly frequent words tend
to resist regularization in other languages as well as in English .
Social Forces That Contribute to Regularization
Te historical and societal forces that contribute to change in
spoken language are usually out of the awareness of speakers,
except when those in power exert control through standard-
ization or some other kind of language policy . A stan-
dardized language is usually one particular variety designated
take a comparative approach, studying the patterns of register
variation , which seems to be inherent in human language .
corpus linguistics has been an especially productive
analytical approach for studying register variation. For example,
the Longman Grammar of Spoken and Written English (Biber et
al 1999 ) applies corpus-based analyses to show how any gram-
matical feature can be described for both its structural character-
istics and its patterns of use across spoken and written registers.
In the multidimensional approach to register variation, corpus-
based analysis is combined with sophisticated statistical analy-
sis to analyze the patterns of linguistic variation that distinguish
among registers (see, e.g., Biber 1988 , 2006 ; Conrad and Biber
2001 ) .
Some studies have distinguished between registers and
genres. In these studies, the term register refers to a general kind
of language associated with a domain of use, such as a legal regis-
ter , scientifc register , or bureaucratic register . Register studies
have usually focused on lexico-grammatical features, showing
how the use of particular words and grammatical features vary
with the situation of use. In contrast, the term genre has been
used to refer to a culturally recognized message type with a
conventional internal structure, such as an af davit, a biology
research article, or a business memo. Genre studies have usually
focused on the conventional discourse structure of texts or the
expected sociocultural actions of a discourse community .
Douglas Biber
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Biber , Douglas. 1988. Variation across Speech and Writing .
Cambridge : Cambridge University Press .
. 2006. University Language: A Corpus-Based Study of Spoken and
Written Registers . Amsterdam : John Benjamins .
Biber , Douglas , Stig Johansson , Geofrey Leech , Susan Conrad , and
Edward Finegan . 1999. Longman Grammar of Spoken and Written
English . London : Longman .
Bruthiaux , Paul . 1996. Te Discourse of Classifed Advertising . New
York : Oxford University Press .
Conrad , Susan , and Douglas Biber , eds. 2001. Variation in English: Multi-
Dimensional Studies . London : Longman .
Ferguson , Charles A. 1983. Sports announcer talk: Syntactic aspects of
register variation . Language in Society 12 : 153 72.
Ghadessy , M., ed. 1988. Registers of Written English: Situational Factors
and Linguistic Features . London : Pinter .
Halliday , M. A. K. 1988. On the language of physical science. In
Ghadessy 1988, 16278.
Janda , R. 1985. Note-taking English as a simplifed register . Discourse
Processes 8 : 437 54.
REGULARIZATION
Regularization is the process of becoming more regular or
rule governed. In a general sense, it implies that events in the
world (or in individuals) become more predictable and orderly.
Regularization in language is one of many kinds of language
change. It occurs when forms in the language are under pressure
to be consistent with a common pattern. Te patterns that speak-
ers use in the regularization of the existing language are typically
the same as those that are used with new forms that enter the lan-
guage and are referred to as productive . Tus, the past tense -ed
Regularization
709
for new words, following a complex set of morphophonological
rules. In order to form a plural, one must observe the ending of
the stem of the word, and then add / -s/ or / -z/ or / -iz/ depending
on whether the stem ends in a voiceless sound, a voiced sound,
or a sibilant (backs, bugs, busses ). When new words such as hard-
scape , abdominoplasty , and rifage recently entered English, they
were all treated as regular plurals. At some point, adults appear to
have acquired the rules for making plurals, making it possible to
apply them to new words or to regularize existing words. Adults
are also able to produce appropriate irregular forms in English,
for instance, the past tenses of words like sing, keep, and catch.
An adequate psycholinguistic model must account for both reg-
ular and irregular infections.
Studies of childrens acquisition of morphology provide some
insights into the development of these mental representations
(see morphology, acquisition of ). It has also long been
observed that young children acquiring language make typical
errors of overregularization : Tey regularize words that are
irregular and say things like my foots, two mouses, and My
teacher holded the baby rabbits and we patted them, clearly not
as imitations of adult models (Gleason 1966 ). Tese errors are
overgeneralizations that children make as they acquire the lin-
guistic system. Overregularized words are evidence that the child
has knowledge about language that goes beyond the memoriza-
tion of heard forms. Studies of spoken language have shown a
U-shaped curve in childrens acquisition of irregular words. In
the frst stage, young children produce correct irregular forms
along with uninfected regular words. Somewhat later, at about
the same time that children begin to produce correct regular
pasts and plurals, the overregularized forms appear, and fnally
both regular and irregular infections are used (Marcus et al.
1992 ). One explanation is that initially, they produce the irregu-
lars correctly as rote unanalyzed forms, but when they begin to
acquire the rules for the regular infections, they overapply those
rules, unaware of the exceptions. Only later are they able to han-
dle both the regular forms and the irregular ones .
Experimental studies have shown that children as young as
four have systematic knowledge about the infectional system.
When presented with pictures of creatures with novel names and
told Tis is a wug. Now there is another one. Tere are two ___?
they accurately produced the plural wugs (Berko 1958 ). When
given novel verb forms (Tis is a man who knows how to spow.
Yesterday he ___?), the children uniformly produced the regular
form spowed. Children were remarkably consistent in producing
the most frequent, productive, and uncomplicated infections of
nouns and verbs.
Tere are several theoretical explanations for these phe-
nomena. Some researchers contend that speakers operate with
abstract higher-order rules when infecting regular forms (Pinker
1991 ). According to this view, the infectional algorithm allows
speakers to deal with any regular word, whether it is familiar or
not. Irregular words, however, must be committed to memory.
Other scholars disagree in the interpretation of how it is that
speakers are able to handle new words, and whether activa-
tion of internalized rules adequately characterizes the process.
connectionist approaches to language do not rely on rules,
nor do they divide the lexicon into two groups of words, regu-
lar and irregular. In connectionist models, the mental processor
the of cial or preferred language, and it does not follow that
the chosen dialec t will be any more regularized than less val-
ued varieties. Standardized spelling , by contrast, is typically
a move to simplify and regularize spelling conventions (Wijk
1977 ). Tere are cases, however, where of cial bodies are del-
egated to choose or change words or expressions in order to
provide conformity with the rules of a language: For instance,
the Academy of the Hebrew Language in Israel is charged with
replacing words borrowed from other languages with words
based on Hebrew semantics and morphology , in this way
providing a kind of regularization of the lexicon of contem-
porary Hebrew. Te French Academy (lAcadmie franaise )
compiles a dictionary of the French language and makes rec-
ommendations as to the admissibility of loan words, usually in
favor of words that are more French for instance, courier lec-
tronique instead of e-mail . Tese attempts to control language
by decree are often not successful, especially in keeping out loan
words that are pervasive and international. In any language,
loan words themselves are subject to regularization that brings
them into conformity with the morphology and phonotactics of
the target language. For example, the English word baseball has
been incorporated into Japanese as beisuboru , and the umpires
call to play ball is rendered as purei boru.
A powerful regularizing force in a language is the communi-
cation pressure that results from becoming more cosmopolitan,
whereas an insular language is more likely to remain irregular
and complex. When a society becomes more multiethnic, there
is often language contact . In addition, a rapidly changing
society may have new technologies and terminologies, a mobile
population carrying language to new communities, and many
speakers who do not know the language well. All of these fac-
tors lead to linguistic economy, simplifcation, and regulariza-
tion. When new words are coined, their infections are regular.
Loan words from contact languages are regularized. And, like
frst language learners, second language learners (see second
language acquisition ) of whatever age tend to regularize
the language they are acquiring. It is very common for learners
of French, for instance, to regularize the irregular second per-
son plural of dire (to say), producing vous disez instead of vous
dites. Tis regularized form is not permissible in standard French,
but it occurs in various nonstandard dialects and is the norm in
Louisiana Cajun French. When a language is spoken in a homo-
geneous, isolated community with few non-native speakers and
little outside contact, there is apparently a greater tendency for
complexity to remain in the infectional system. A comparison
of modern Icelandic and Norwegian provides a good example of
this tendency: Both have Old Norse beginnings, but the relatively
isolated Icelandic has preserved complex morphology, including
many of the strong verbs of Old Norse, whereas comparatively
cosmopolitan Norwegian shows more regularization, with sim-
pler infectional morphology and many fewer Old Norse strong
verbs (Kusters 2003 ) .
Psycholinguistic Studies
Some of the best evidence we have for the processes of paradig-
matic regularization has come from psycholinguistic studies of
both children and adults. We know that adult speakers of English,
for instance, are able to produce appropriate infectional endings
Regularization
710
input may deliver a variety of diferent types of cognitive efects to
the system: It may, for instance, combine inferentially with exist-
ing assumptions to yield new conclusions (known as contextual
implications ), it may provide evidence that strengthens existing
beliefs, it may contradict and eliminate already held information,
or it may rearrange the way information is stored. Such efects
may or may not be benefcial to an individual; that is, they may
increase or decrease the accuracy of the cognitive systems infor-
mation about the world and may make useful information easier
or harder to access. An input is relevant to a cognitive system only
if it benefts that system, that is, only if it has positive cognitive
efects. Te other crucial factor afecting the degree of relevance
of an input (whether an external stimulus or an internal men-
tal representation) is the processing efort it requires: Deriving
efects from any given input requires a mobilization of cognitive
resources, including attention, memory, and various processing
algorithms and heuristics. Tus, the relevance of any input is a
trade-of between the positive cognitive efects it yields and the
processing efort it requires: Te greater the ratio of efects to
efort, the greater the relevance of the input.
Te basic claim of the framework is that human cognition is
oriented toward maximizing relevance (known as the cognitive
principle of relevance ). Te evolutionary idea underlying this
claim is that as a result of constant selection pressure toward
increasing cognitive ef ciency, humans have evolved proce-
dures to pick out potentially relevant inputs and to process them
in the most cost-efective way (Sperber and Wilson [1986] 1995b ).
Human communicative behavior, including verbal communica-
tion and comprehension, exploits this prevailing cognitive drive
for relevance in a particular way .
Relevance and Linguistic Communication
Te starting point for a pragmatic theory (that is, an account
of how speakers and their addressees manage to converge on
a shared interpretation) is the question concerning how hear-
ers are able to bridge the gap between the encoded linguistic
meaning and the speakers intended meaning. Te most obvi-
ous manifestations of this gap are nonliteral uses of language,
such as metaphor, metonymy or irony, and cases where the
speaker communicates, in addition to the proposition explicitly
expressed, a further proposition known as a conversational
implicature , exemplifed by speaker Ys utterance in (1).
(1) X: We need your written report now.
Y: Ive been very busy recently.
Implicating: Y hasnt written the report.
Tere is also a range of tasks involved in determining the
proposition explicitly expressed, including disambiguation,
assignment of referents to indexicals , and flling in miss-
ing constituents, as in (2), and various other enrichments or
adjustments of encoded content, as indicated in the examples
in (3): (3b) involves a narrowing down of take time, and (3c) a
loosening of the concept encoded by boiling. In each case, of
course, the particular proposition explicitly expressed is just one
of indefnitely many other possibilities:
(2) He has taken enough from her.
Expressing: Jim has endured enough abusive treatment from Mary.
relies on exemplars or analogies : Speakers hear a variety of
words over time, some more than others, and their mental rep-
resentation refects the weight of these frequencies, the features
of the words, and the circumstances of their use. Frequently
encountered features are recognized and associated, and ulti-
mately the learner produces language that matches the language
that has been heard .
Regardless of the theoretical model, it is clear that individuals
are sensitive to the characteristics of the language around them
and are able to generalize those characteristics to new instances.
By the time children are of preschool age, they have suf cient
knowledge of the most regular features of language to be able to
extend them to words they have never heard before. Tis kind of
knowledge underlies both productivity and regularization, and
it has implications for language change in general. Languages
everywhere are moved to become more regular in response to
communicative pressure. Speakers carry within themselves the
linguistic tools of regularization, which is a process that reveals
a fundamental characteristic of the way humans organize
information .
Jean Berko Gleason
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Baugh , Albert C. 1957. A History of the English Language . New
York : Appleton-Century-Crofts .
Berko , Jean. 1958. Te childs learning of English morphology . Word
14 : 150 77.
Gleason , Jean Berko. 1966. Do children imitate? Proceedings of the
International Conference on Oral Education of the Deaf 2 : 441448.
Washington, DC : Alexander Graham Bell Association for the Deaf .
Kusters , Wouter. 2003. Linguistic complexity: Te infuence of social
change on verbal infection . Ph.D. diss., University of Leiden.
Marcus , Gary , S. Pinker , M. Ullman , M. Hollander , T. J. Rosen , and Fei Xu .
1992. Overregularization in language acquisition . Monographs of the
Society for Research in Child Development 57: 1182 .
Pinker , Steven . 1991. Rules of language . Science 253 : 530 5.
Wijk , Axel . 1977. Regularized English: A Proposal for an Efective
Solution of the Reading Problem in the English-Speaking Countries .
Stockholm : Almqvist and Wiksell .
RELEVANCE THEORY
Relevance theory (RT) is best known for its account of verbal
communication and comprehension, which is employed by
people working in pragmatics . It also sets out a broad pic-
ture of the principles driving the human cognitive system as a
whole, and this plays a crucial role in underpinning the particu-
lar claims made about communication (Sperber and Wilson
[ 1986 ] 1995a; Wilson and Sperber 2004 ).
Relevance and Cognition
According to the RT framework, human cognitive processing quite
generally is geared toward achieving as many improvements to
its representational contents and their organization as possible,
while ensuring that the cost to its energy resources is kept as low
as reasonably possible. At the center of the theory is a techni-
cally defned notion of relevance , where relevance is a potential
property of any input to any perceptual or cognitive process. An
Relevance Theory
711
as: Y might trip over, Y should take small steps, Y should keep an
eye on the path, and so on. Te result is a concept, which we can
label uneven* , whose denotation is a proper subset of the deno-
tation of the lexical concept uneven .
A distinctive RT claim in this context is that metaphorical
and hyperbolic uses of words involve a kind of concept broaden-
ing (loose use), and so fall within this single process of lexical
meaning adjustment. For instance, an utterance of the sentence
in (5) could be taken as an ordinary broadening (if, say, the run
referred to was a little less than 26 miles) or as hyperbolic (if it
was considerably less than the length of a marathon) or as met-
aphorical for a long, arduous, exhausting experience, whether
physical or mental:
(5) It was a marathon .
(For much more detailed exemplifcation of the RT-based
account of lexical adjustment, resulting in concept broadening,
or narrowing, or a combination of the two, see Carston 2002 ;
Wilson and Sperber 2002 ; Wilson and Carston 2007 ).
As these examples indicate, on the RT view, the contribution of
pragmatics to the proposition expressed by an utterance (i.e., its
truth conditional content) is very wide-ranging, going well
beyond the role of simply providing contextual values for indexi-
cals. So, unlike the standard view of semantics and pragmatics
in the philosophy of language, the RT position on the semantics/
pragmatics distinction is that it does not coincide with the dis-
tinction between explicit utterance content and implicature but,
rather, with the distinction between context-free linguistically
encoded meaning and what is communicated .
Relevance Theory in a Broader Perspective
Te output of pragmatic interpretation is a conclusion about
speakers communicative intentions, and the inferential process
may rely on assumptions about other mental states of speak-
ers (their beliefs and desires). So there is a close relationship
between utterance comprehension and the theory of mind
capacity, both of which depend on a metarepresentational abil-
ity an ability to represent representations and attribute them
to others. Dan Sperber and Deirdre Wilson ( 2002 ) argue that the
pragmatic capacity is a subsystem of the mental system respon-
sible for interpreting peoples behavior in terms of the underly-
ing mental states that cause it. Tey further argue, in line with
current views in evolutionary psychology on cognitive
architecture , that it is an autonomous domain-specifc sys-
tem, operating just on ostensive stimuli (in particular, utter-
ances), and has its own dedicated procedures, as outlined here.
Tus, it is a modular system, a submodule of the more general
theory-of-mind module.
Tere is currently a strong emphasis on spelling out the
empirical predictions of relevance theory, comparing them with
other theories of cognition and/or pragmatics and subjecting
them to experimental testing. Tere are three broad areas of
such empirical work. Te frst involves using familiar psycho-
linguistic techniques (e.g., measuring participants reaction
times in a range of on-line tasks) to investigate the time course
of processing and the relative allocation of time/efort to difer-
ent aspects of pragmatic interpretation (see, for instance, Noveck
and Sperber 2004 ) . Te second is research into the development
(3) a. Ive eaten.
Expressing: Ive eaten dinner tonight.
b. Your knee will take time to heal.
Expressing: Te knee will take a substantial amount of time to
heal.
c. Te water is boiling.
Expressing: Te water is very hot [not necessarily strictly at
boiling point] .
How, then, is an addressee able to infer the intended meaning
from the encoded linguistic meaning, which is just a schematic
guide or set of clues? According to RT, the answer lies with a spe-
cial property of overtly communicative acts, which is that they
raise certain expectations of relevance in their addressees, that
is, expectations about the cognitive efects they will yield and the
mental efort they will cost. Quite generally, an utterance comes
with a presumption of its own optimal relevance ; that is, there
is an implicit guarantee that the utterance is the most relevant
one the speaker could have produced, given his or her compe-
tence and goals, and that it is at least relevant enough to be worth
processing. Tis is known as the communicative principle of rel-
evance and it follows from the cognitive principle of relevance in
conjunction with the overtness of the intention that accompanies
an utterance: Te speaker openly requests efort (attention) from
the addressee, who is thereby entitled to expect a certain quality
of information requiring no gratuitous expenditure of efort.
Tat utterances carry this presumption licenses a particular
comprehension procedure, which, in successful communica-
tion, reduces the number of possible interpretations to one.
According to the relevance-theoretic comprehension procedure :
(a) Follow a path of least efort in computing cognitive
efects: Test interpretive hypotheses (disambiguations, ref-
erence resolution, lexical adjustments, implicatures, etc.) in
order of accessibility.
(b) Stop when your expectations of relevance are satisfed.
Tis procedure is automatically applied in the on-line process-
ing of verbal utterances: Taking the schematic decoded linguis-
tic meaning as input, processes of pragmatic enrichment at the
explicit level occur in parallel with the derivation of the implica-
tions of the utterance.
Central to the working of the procedure is a subprocess of
mutual adjustment of explicit content and contextual implica-
tions, a process guided and constrained by expectations of rele-
vance. Here is a brief example involving the adjustment of explicit
content in response to expected implications and where the out-
come is a narrowing down of a lexically encoded meaning:
(4) X (to Y): Be careful. Te path is uneven .
Given that the frst part of Xs utterance warns Y to take care, Y is
very likely to expect the second part of the utterance to achieve
relevance by explaining or elaborating on why, or in what way,
he should take care. Now, virtually every path is, strictly speak-
ing, uneven to some degree or other (i.e., not a perfect plane),
but given that Y is looking for a particular kind of implication,
he or she will enrich the very general encoded concept uneven
so that the proposition explicitly communicated provides appro-
priate inferential warrant for such implications of the utterance
Relevance Theory
712
speak about God in language suggestive of (Soskice 1985 , 51)
a father, mother, master, king, shepherd, rock, fre, or wind but
without intending these references to be taken literally. To assist
refection and argument, religions sometimes develop more sta-
ble, extended, systematic (and often complementary) linguistic
models from such illuminating metaphors for example, God
as personal; the death of Christ as a victory or sacrifce.
Traditional accounts argue that such models can truly represent,
although they do not literally describe, divine reality, but more
radical thinkers treat them as no more than imaginative fctions
whose purpose is to motivate and direct peoples spiritual and
moral lives and experiences .
Religious analogies are sometimes regarded as diferent
only in degree from religious metaphors, but many insist that
they apply to God literally and function as legitimate extensions
of their normal application. So God is really wise, merciful, and
living, but in a stretched sense appropriate to Gods very difer-
ent nature when compared with humans. (Te same may be said
of the meaning of clever when applied to dogs or computers.)
It is possible partly to specify and thus partly replace analogi-
cal religious language, providing that we know the ways in which
(say) Gods love is similar to that of a loving person and the ways
it is diferent. It is only when analogies are specifed that they
can be reliably used in argument; metaphors are frequently too
vague to allow inferences to be drawn from them .
Extended metaphors in narrative form are used to speak of
the actions of divinities. In myths, God or the gods are pictured
as interacting with this world; in parables , this activity is said to
be like a wholly human situation (many are explicit similes). In
both cases, the stories are often self-involving and possess con-
siderable emotional power and salience .
Performative Religious Language
In addition to this cognitive (fact-asserting) function, reli-
gious language is also used to perform many noncognitive
tasks through speech-acts in which attitudes or feelings are
expressed, commitments, vows and requests are made, or obli-
gations prescribed. J. L. Austin distinguished this illocution-
ary force from an utterances consequences ( perlocution ).
In worship, religious persons expresses their trust, gratitude,
awe, and longing and makes prayerful requests and promises.
Tis language is intended to afect God, but it often also evokes
and deepens the faith of other worshippers as well as producing
a reinforcing efect on the language users themselves. In many
religions, especially nontheistic faiths, the language of medita-
tion including the repetition of mantras serves to evoke mys-
tical experiences or spiritual illumination of various kinds: for
example, the realization of the oneness of the self with absolute
Brahman in Hinduism or the enlightenment experience that
releases from dukkha (the unsatisfactoriness or sufering of
life) in Buddhism .
Religious Language and Religious Life
According to Ludwig Wittgenstein , all language is rooted within
human activity, as part of a form of life . Some argue that reli-
gion itself is a form of life, or at least that some of its distinctive
activities for example, its rituals are examples of this category.
Much of the language used in these contexts may be thought of
of communicative competence in children and its relation to
their linguistic maturation, on the one hand, and to their devel-
oping theory-of-mind capacity, on the other. Tere is evidence
of a close relation between mature pragmatic competence (e.g.,
ability to interpret nonliteral uses of language, such as metaphor
and irony) and mature theory of mind (e.g., ability to attribute
to others beliefs that one knows to be false). But it is also clear
that children can communicate ostensively from around the age
of two years, and so this early ability does not depend on a full-
fedged metarepresentational capacity, which comes some years
later, but perhaps on some earlier emerging component of the-
ory of mind, such as a capacity for joint attention. Finally, there is
empirical investigation of people with atypical or impaired com-
municative capacities, including those with autism , Williams
syndrome, or schizophrenia. Te results so far provide tentative
support for RT predictions about the diferences in the process-
ing of literal, metaphorical, and ironical uses of language (see
Wilson 2005 ; Carston and Powell 2006 ) .
Robyn Carston
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Carston , Robyn . 2002. Toughts and Utterances: Te Pragmatics of Explicit
Communication . Oxford : Blackwell .
Carston , Robyn , and George Powell . 2006. Relevance theory new direc-
tions and developments. In Te Oxford Handbook of Philosophy of
Language , ed. Ernie Lepore and Barry C. Smith , 34160. Oxford : Oxford
University Press .
Noveck , Ira, and Dan Sperber , eds. 2004. Experimental Pragmatics .
Basingstoke, UK : Palgrave .
Sperber , Dan , and Deirdre Wilson . [1986] 1995a. Relevance: Communica-
tion and Cognition . 2d ed. Oxford : Blackwell.
. [1986] 1995b. Postface. In Sperber and Wilson 1995, 25579.
. 2002. Pragmatics, modularity and mind-reading . Mind and
Language 17 : 3 23.
Wilson , Deirdre . 2005. New directions for research on pragmatics and
modularity . Lingua 115 : 1129 46.
Wilson , Deirdre , and Robyn Carston . 2007. A unitary approach to lexical
pragmatics: Relevance, inference and ad hoc concepts. In Advances in
Pragmatics , ed. Noel Burton-Roberts . Basingstoke, UK : Palgrave .
Wilson , Deirdre , and Dan Sperber . 2002. Truthfulness and relevance .
Mind 111 : 583 632.
. 2004. Relevance theory. In Te Handbook of Pragmatics , ed.
Laurence Horn and Gregory Ward , 60732. Oxford : Blackwell .
RELIGION AND LANGUAGE
Te topic of religion and language could potentially fll a book.
In order to make it more manageable, this entry concentrates
mainly on theistic, especially Judeo-Christian, usage.
Descriptive Religious Language
In speaking and writing of God, or however they conceptualize
the ultimate, religious people attempt to portray a mystery that
transcends empirical reality by using language that is normally
applied to things or (particularly) human beings. Islam in par-
ticular emphasizes the transcendence of God.
Figurative forms of language draw on the associations of lan-
guage in its home in the mundane world to highlight similarities
with this mysterious divine world. Using metaphors , people
Religion and Language
713
Tiselton , Anthony C . 1992. New Horizons in Hermeneutics .
London : HarperCollins .
Wittgenstein , Ludwig . 1980. Culture and Value . Ed. G. H. von Wright ,
trans. Peter Winch. Oxford : Blackwell .
REPRESENTATIONS
Defnition
Te core concept of a representation is made up of two distin-
guishable ideas.
(1) Any representation, r , must be about , or mean , or have as
its content , some distinct item s ;
(2) r must be employable, in the absence of s , to guide the
behaviors of the consumers of r organisms or systems that
use r with respect to s .
To say that these ideas are distinguishable is not to say that they
are separable. On some accounts of representation, (1) is a func-
tion of (2). Tat is, a representation can have a distinct item as its
content only because of the way it is utilized by its consumers.
Te relation between (1) and (2) is, however, disputed .
Original and Derived Representations
It is common to distinguish original and derived representations.
Te distinction is intuitively clear, but its application less so. Te
distinction pertains to how representations acquire their con-
tent. Some seem to acquire their content in virtue of intentional,
hence representational, states of subjects. Many people think
this is true of language: A form of words has content only because
of the intentions and understanding of its users (see inten-
tionality ). When persons use a form of words in a context
that is not metaphorical, ironic, or sarcastic they intend their
words to be taken in a certain way, and hearers/readers of their
words understand that they intend their words to be taken in this
way, and so on (Grice 1957; see communicative intention ).
Language, then, would consist in derived representations: repre-
sentations whose content derives from the intentional states of
its users. Intentional states, on the other hand, would be original
representations: representations whose content is not similarly
derived .
The Problem of Representation
Tis conception of the relation between original and derived
representation has exerted an immense infuence on the way the
concept of representation has been developed in recent decades.
It entails that if we want to understand representations, then our
focus must be on intentional states. If we understand the content
of derived representations as issuing from original representa-
tions, then we must understand how the latter can have their
content. All representation ultimately reduces to mental repre-
sentation. Te dif culties inherent in this project were identifed
by Ludwig Wittgenstein ( 1953 ).
Wittgenstein argued that the content of an intentional state
cannot be determined by any fact conscious or unconscious
about the states subject. Consider, frst, an imagistic concep-
tion of content. One might suppose, for example, that when I
use the word dog an image of a dog must somehow fash before
my mind. However, as Wittgenstein showed, such an image
as religious language-games , such as praying or confessing,
that possess a particular set of rules or logical grammar very
diferent from that operating in some other language-games.
(Tus, predestination is said to be less a theory than a sigh, or
a cry [Wittgenstein 1980 , 30e]) . For Wittgenstein, the forms of
life are given, for this is what people do, and the justifcation of
a language game lies in our acting. Te meaning of religious
belief, therefore, may be thought of as thoroughly grounded in
religious behavior.
Christian theologians infuenced by Wittgenstein sometimes
treat Christianity as a learned cultural-linguistic system, with
doctrine as the rules of its grammar; others have developed his
refections into a wholly noncognitive view, in which belief in
God is nothing more than allegiance to a set of spiritual values.
Religious Readers and Religious Writers
Hermeneutics (the theory of interpretation; see philology and
hermeneutics ) is key to understanding how sacred texts are
read. Friedrich Schleiermacher saw the readers task as that of
uncovering the authors intended meaning, but many insist that
writing distances the author from the text and that the reader has
his or her own part to play. Certainly, readers are always inter-
preters of the text and never approach it with innocent eyes
(without their own preconceptions and worldview). For Hans-
Georg Gadamer , the texts horizon of meaning does not swamp
that of the reader; rather, the two interact and fuse, resulting in a
new interpretation that goes beyond them both .
While religious conservatives seek to preserve a single under-
standing of the meaning of a text (and sometimes insist on its
literal interpretation), more liberal and radical scholars regard
it as perfectly legitimate for religious readers to fnd other mean-
ings there that go beyond what the original author consciously
intended, even if that can be recovered. Tis second perspective
works well for some sacred poetry, and it fts the claim that reli-
gious responses are at their most authentic when they are most
personal. It is more dif cult to sustain, however, for religious his-
tory and biography and wherever readers care about what the
author meant to say in using this language. Such considerations
raise theological questions about the divine inspiration of the
text and its role as a medium of revelation, in which God speaks
through sacred scripture (whether by infallible dictation or more
generally perhaps by superintending its writing, or merely by
authorizing or appropriating the fallible authors language) .
Jef Astley
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Astley , Jef . 2004. Exploring God-Talk: Using Language in Religion .
London : Darton, Longman and Todd .
Caird , George B . 1980. Te Language and Imagery of the Bible . London :
Duckworth .
Dhavamony , Mariasusai . 1973. Phenomenology of Religion . Rome :
Gregorian University Press .
Soskice , Janet Martin . 1985. Metaphor and Religious Language .
Oxford : Clarendon .
Smart , Ninian . 1996. Dimensions of the Sacred: An Anatomy of the Worlds
Beliefs . London : HarperCollins .
Stiver , Dan R . 1996. Te Philosophy of Religious Language: Sign, Symbol,
and Story . Oxford : Blackwell .
Religion and Language Representations
714
account of what representation is . One prominent response to
this legacy is the attempt to naturalize representation to explain
what makes something a representation by appealing only to
states that are nonintentional or less than fully intentional. Tese
attempts can be divided, roughly, into informational and teleo-
logical approaches.
Fred Dretske ( 1981 ) argued that representation can be
explained in terms of information : An item qualifes as rep-
resentational only if it carries information about some item
extrinsic to it. Information reduces to relations of conditional
probability though precisely which relations is a matter of dis-
pute. According to the stringent version, defended by Dretske,
information requires a conditional probability of 1. Tat is, for r
to carry the information that s , the probability of s given r must
be 1 (i.e., given r , s must be certain). Other versions (e.g., Lloyd
1989 ) identify information only with an increase in conditional
probability, though not necessarily to the level of 1. On this view,
r will carry information about s if the probability of s given r is
greater than the probability of s given not r .
Te primary drawback with the informational account is
its well-documented problems accommodating an essential
feature of representations: normativity (Dretske 1986 ; Fodor
1990 ) Consider the mental representation of a dog dog . If dog
occurs, then the world should , in an appropriate way, contain a
dog. Te representation, therefore, makes a normative claim: a
claim about the way the world should be, given that the repre-
sentation is instantiated. However, it is likely that dog can be
caused by things that are not dogs foxes in the distance or cats
on a dark night, for example. So, what dog is most reliably cor-
related with is not dogs but a disjunction of dogs or foxes-in-the-
distance or cats-on-a-dark-night. But relations of conditional
probability the core of the concept of information are a func-
tion of reliable correlation. So, the pure informational account
of representation cannot distinguish between the way the world
should be when dog is instantiated and the way the world in fact
is. Representation is normative in a way that information is not.
Central to teleological approaches is the concept of proper
function (Millikan 1984 , 1993 ; Papineau 1984 ). Tis is a norma-
tive concept: Te proper function of a mechanism, trait, or state
is what it is supposed to do, what it has been designed to do, what
it ought to do. It is not what that mechanism generally does or is
disposed to do. What something does or is disposed to do is not
always what it is supposed to do. First, any mechanism, trait, or
process will do many things, not all of which are part of its proper
function. Secondly, a mechanism, trait, or process can have a
proper function even if it never, or hardly ever, performs it. Tird,
a mechanism, trait, or process may have a proper function and
yet not be able to perform it properly .
Te normativity of proper functions is grounded in their his-
tory . Te proper function of a heart is to pump blood because it is
their doing this in the past that explains why hearts proliferated
and exist today. Tey did not proliferate because of their abil-
ity to make noise or produce wiggly lines on an electrocardio-
gram. Tus, the proper function of an item is determined not by
the present characteristics or dispositions of that item but by its
history : Proper function is essentially historical.
Te core idea of the teleological approaches to representation
is that the mechanisms responsible for mental representation are
is neither necessary nor suf cient for content. It is not neces-
sary because, typically, when I use the word dog , no such image
appears to me. It is not suf cient because the image is, itself, just
another symbol, and its content needs to be interpreted no less
than that of the word. An image of a dog can stand for many
things dog, furry creature, creature with four legs, and so on
and so the image can have content only if there is a further act
of interpretation on the part of its subject: an act that serves
to disambiguate the image. However, an act of interpretation is
itself an intentional state. Terefore, we have made no progress
in understanding how an intentional state has content; we have
simply replaced one intentional state with another.
Tese problems remain when we shift to more sophisticated
accounts of content. In one such account, content is understood
in terms of the following of rules : In using a sign, I consciously
or unconsciously follow a rule, a rule that determines how the
sign is to be applied in particular cases. Wittgenstein argued that
there is no fact about an individual that determines that he or
she is following one rule rather than another. Any behavior in
which I engage is compatible with an indefnite number of rules.
In continuing the mathematical sequence 2, 4, 6, 8, I could be fol-
lowing the n + 2 rule or the n + 2 if and only if n is less than 32,
if so n + 4 rule. And so on for an infnite number of n s. Similarly,
in applying the word dog , I could be following the Apply dog to
all and only dogs rule, or the Apply dog to all and only dogs
unless the dog is frst seen after 2020 in which case apply cat,
and so on for an infnite number of permutations (see pro-
jectibility of predicates ). Crucially, this point also extends
to mental rehearsals of a rule themselves just more subtle forms
of behavior. It would be implausible to suppose that whenever I
apply a rule I must mentally rehearse all the possible situations
in which the rule might apply for this would entail that when-
ever I followed a rule, I must be simultaneously thinking an inf-
nite number of thoughts.
Tese examples are outlandish but are merely ways of making
graphic a simple point: Any rule is, logically, no diferent from
a word. Te rule is just another symbol and, as such, stands in
need of interpretation if it is to have content. But interpreting is
a representational state. And so, in our attempt to understand
original representation, we have merely substituted the problem
of understanding the content of one representation with that of
understanding the content of another.
Wittgensteins response to this rule-following paradox
turns on the appeal to practice . Tis response is, however, deeply
problematic at least if understood as a constructive attempt to
explain the foundations of meaning. A practice is, as he put it,
what we do . But doing seems to be a form of acting . And actions
are essentially connected to, and presuppose, prior intentional
states of a subject. Tat is, both the status of an event as an action
and its identity as the particular action that it is depend on its
relation to a subjects intentional states. Terefore, the appeal to
action seems to presuppose representational states and so can-
not explain what representations are (McDowell 1992 ; Hurley
1998 ; Rowlands 2006 ) .
Naturalizing Representation
Wittgensteins legacy, therefore, is a convincing account of
what representation is not , coupled with a highly questionable
Representations
715
McGinn , Colin . 1991. Te Problem of Consciousness . Oxford : Blackwell .
Millikan , Ruth Garrett . 1984. Language, Tought and Other Biological
Categories . Cambridge, MA : MIT Press .
. 1993. White Queen Psychology, and Other Essays for Alice .
Cambridge, MA : MIT Press .
Papineau , David . 1984. Representation and Reality . Oxford : Blackwell .
Rowlands , Mark . 2006. Body Language: Representation in Action .
Cambridge, MA : MIT Press .
Wittgenstein , Ludwig . 1953. Philosophical Investigations . Oxford :
Blackwell .
RHETORIC AND PERSUASION
Rhetoric is traditionally a discursive skill, either written or oral,
used to produce a desired efect on an audience. Alternatively,
it is a study that focuses on the techniques a person should opti-
mally employ in given contexts to produce such an efect. Tis
second sense echoes Aristotles emphasis, set out in his Art of
Rhetoric , pertaining to the detection of the persuasive aspects
of each matter ( 1991 , 6970).
From an etymological perspective, the English noun rhetoric
is derived from the Greek word rhma (a word), which is linked
to rhtor (a teacher of oratory). Both are ultimately derived
from the Greek verb eir (I say). Te notion of rhetoric, there-
fore, is grounded in language. Te English noun persuasion is as
much grounded in cognition as it is in language. Even though the
suade element of the word goes back to the Latin sudre to
advise, the Greek word for persuasion derives from the cogni-
tive verb to believe.
Te formal codifcation of rhetoric as a heuristic-like system
is something that was frst written down in circa 475 b.c. by
Corax of Syracuse and involved a simple but efective four-part
structure: introduction, background, arguments, and conclu-
sion. Rhetoric can only operate when individuals are given the
opportunity to speak and persuade in a public forum. In ancient
societies that had tyrannical, royal, aristocratic, or oligarchic
systems, freedom of speech was either restricted or nonexis-
tent, and as a result, rhetorical practice was very limited. It was
only with the advent of democracy, championed by Cleisthenes
in the city-state of Athens around 510 b.c. , that individuals
were given the opportunity to practice persuasive oratory for
their own ends. Tree distinct genres developed from circa 500
b.c. through to the formal end of Athenian democracy around
320 b.c. Tese were forensic oratory (the rhetoric of the law
courts), deliberative oratory (the rhetoric of the political arena),
and epideictic oratory (the rhetoric of praise or blame). Each
of these genres had its own separate focus expressed in their
means and topics . Some of the great orators of the time were
Gorgias (an epideictic, sophist orator), Lysias and Isocrates
(two forensic orators), and Demosthenes (the great delibera-
tive orator). Famous Roman orators and writers of rhetorical
handbooks were Cicero ( de Oratore and de Inventione ) and
Quintilian ( Institutio Oratoria ). Both Isocrates and Quintilian
were also great educators .
Rhetoric is about structure and strategy. Structure refers
to both the arrangement of the whole process and that of the
speech itself. Te frst is expressed by means of the fve canons of
rhetoric, pertaining to the discovery ( herisis / inventio ); arrange-
ment ( txis / dispositio ), stylization ( lxis / elocutio ), memorizing
evolutionary products, and that we can therefore understand rep-
resentation in terms of the concept of proper function. Suppose
we have a representational mechanism M capable of going into
a variety of states or confgurations. Te direct proper function of
mechanism M is, let us suppose, to enable the organism to track
various environmental contingencies for example, the presence
of predators. In the event that a predator is present, M goes into a
particular confguration, F. Tis state F of the mechanism has the
derived proper function deriving from the proper function of the
mechanism M of indicating the presence of a predator. Tat is,
it has the content, roughly, predator, there!
One strength of the teleological approach is the elegant man-
ner in which it satisfes the normativity constraint. Another is
that it does not rely on the questionable assumption that all rep-
resentation derives from mental representation. Te historical-
normative account of representation can be applied directly to
linguistic forms independently of their connection to the inten-
tional states of subjects. However, all attempts to naturalize
representation are controversial, and their success or failure is
a matter of continuing debate. Te worries surrounding these
attempts can be divided into technical and foundational . With
respect to teleological approaches, for example, one technical
worry would be whether the approach can capture the fneness of
grain of certain content attributions, for example, whether it can
distinguish between the content of predator! and tiger! and
thing that will eat me! Tere is every reason to think that with
suf cient ingenuity, answers to these sorts of technical worries
will be (or, indeed, already have been) forthcoming.
Foundational worries are more serious. Tese concern
whether the sorts of natural relations invoked by these accounts
are the right sorts of things to explain the nature of representation.
Even if we were to identify items that satisfed ones preferred
naturalistic model of representation, it is argued, it would still be
an open question whether these items were, in fact, represen-
tations. Tus, some (e.g., McGinn 1991 ) argue that naturalistic
accounts provide only a criterion of representational content a
criterion that allows us to determine when one thing is about
another, and what thing it is about. But they fail to explain what
representation actually is . More positively, if we are to properly
understand representation, we also need to understand con-
sciousness . And the prospects for a naturalistic interpretation
of the former are, therefore, tied to a naturalistic interpretation
of the latter .
Mark Rowlands
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Dretske , Fred . 1981. Knowledge and the Flow of Information .
Oxford : Blackwell .
. 1986. Misrepresentation. In Belief , ed. R. Bogdan , 1736
Oxford : Oxford University Press .
Fodor , Jerry . 1990. A Teory of Content and other Essays . Cambridge,
MA : MIT Press .
Grice , Paul . 1957. Meaning . Philosophical Review 66 : 377 88.
Hurley , Susan . 1998. Consciousness in Action . Cambridge : Harvard
University Press .
Lloyd , Dan . 1989. Simple Minds . Cambridge, MA : MIT Press .
McDowell , John . 1992. Meaning and intentionality in Wittgensteins
later philosophy. Midwest Studies in Philosophy 17 : 3042.
Representations Rhetoric and Persuasion
716
voice, rhythm , and gesture , something the Roman orators
made into an art .
Troughout the latter half of the twentieth century, rhetoric
has often been viewed in university language departments as an
archaic system. One area where rhetoric did fourish was argu-
mentation analysis. In the 1950s, infuential work was conducted
by Chaim Perelman and Lucie Olbrechts-Tyteca . What they
called the new rhetoric challenged the validity of dialectic and
logico-rational approaches to argumentation, claiming instead
that real arguments are not grounded in absolute or perfect
contexts but, rather, in situations that are characterized by the
possible, plausible, and probable. Hence, a much more prag-
matic linguistic approach to evaluating arguments is required.
Other important philosophical and theoretical scholars in this
period were I. A. Richards , with his work in literary criticism,
and Kenneth Burke , with his rhetoric of motives, based on his
notions of identifcation and appeal, which focuses more on
the reception and context of the persuasive discourse than on
the actual production.
Today, rhetoric and persuasion have dissipated into a
number of associative language-based domains that include
composition, word and image studies, philosophy, psychol-
ogy, communication studies, argumentation analysis, and
stylistics. Rhetoric is crucial to freshman composition courses
in universities. This has traditionally been the case in the
United States but is becoming increasingly so in the rest of the
world. The structural and systematic nature of rhetoric lends
itself perfectly to teaching writing and oral composi-
tion . It is here that the didactic and pedagogical quality of
rhetoric as a skill, in the true Aristotelian sense of techn , to be
applied to other subjects rather than a subject in itself, comes
to the fore.
In communication studies, the term rhetoric has all but disap-
peared, even if what happens there is essentially still rhetoric as
the ancients described it. Tis social scientifc approach to mod-
ern rhetorical theory tends to focus more on cognition and emo-
tion than on language, more on reception than production. Tis
can be summed up in Herbert S. Simonss claim that persuasion
is about winning beliefs, not arguments ( 2001 , xxii). Tis high-
lights the need of a persuader to position himself/herself closer
to the persuadee. Tis attempt at creation of common ground,
which Simons terms a coactive approach, must ultimately
draw on the classical notions of ethos and pathos .
Rhetoric is also very much evident in the domain of critical
discourse analysis , where the ordering, presentation, and
omission of information in news discourse is crucial for persuad-
ing and manipulating mass audiences. Te same can be said in
the word and image domain of advertising discourse, arguably
the most fecund example of rhetoric at work in society in the
modern age. Rhetoric also continues to fourish in argumenta-
tion analysis, as can be seen in the work of Frans H. van Eemeren
and R. Grootendorst ( 2004 ), whose pragma-dialectic model
of argumentation provides a means of resolving diferences of
opinion based on what they term critical discussion.
One current approach that warrants further elaboration is
STYLISTICS : a feld which in the twentieth century replaced
and expanded on the earlier study of elocutio in classical rheto-
ric (Wales 2001 , 372). Its object of analysis is primarily literary
( mnm / memoria ), and delivery ( hupkrisis / pronunciatio ) of
persuasive arguments in a speech or essay.
Once a proposition has been decided on, a speaker or writer
needs to go about gathering, discovering, and generating argu-
ments in support of it. A main strategy is to turn to the topics
( topoi / loci ). Tese are places where arguments can be found.
Te fact that they are already out there in the world and only
need to be used led Aristotle to refer to them as nonartistic
proofs. Topics can be internal, external, or special. Internal top-
ics are textual strategies for generating arguments, such as def-
inition, comparison, analogy, cause and efect, and testimony.
External topics are literal places and objects where people can
go to fnd arguments, like reference books in libraries, or search
engines on the Internet, to give a modern example. Special top-
ics are systems that are particular to the three genres of rhetoric.
Deliberative oratory uses special topics like the worthy and
the advantageous in order to persuade people to act or think
in a certain way. Forensic oratory uses special topics appropri-
ate for either defense or prosecution in the law courts. One such
strategy is stasis , a Roman judicial heuristic, which addresses
the three questions of whether something happens (evidence),
what something is (defnition) and the quality of what happened
(motives). Epideictic oratory uses strategies relevant for either
amplifying or playing down the virtues and vices of individuals
or institutions .
A second mode of persuasion in this frst canon of rhetoric is
what Aristotle termed the artistic proofs , incorporating the three
appeals: logos , pathos , and ethos . Logos centers on whether argu-
ments are deductive or inductive, fallacious or nonfallacious,
syllogistic or enthymemic. Pathos deals with the psychology of
persuasion, focusing on the ways that emotions are triggered
and then channeled in an audience for the speakers own ends.
Ethos is concerned with the moral character of the speaker, the
trust or admiration an audience has for a speaker or writer. It
also includes ethics in the strategies employed in the language
use itself, like dealing with counterarguments, opponents, and
minorities in a fair and balanced way.
Te second canon is concerned with arrangement. Tis occurs
at two levels: that of the entire speech and of the arguments
themselves. One famous system is that set out in the anonymous
frst century b.c.e. Roman manual Rhetoric ad Herennium . Tis
work, which also deals with the last three canons, stipulates six
parts to a speech: introduction ( exordium ), background/scene
setting ( narratio ), a brief list of arguments ( divisio ), the argu-
ments in favor ( confrmatio ), the counterarguments ( confutatio ),
and conclusion ( peroratio ). Te second level of arrangement
concerns sections four and fve: which arguments a speaker puts
forward frst, which last, and which placed in the middle and,
most importantly, why .
Te third canon deals with style and is realized by the use of
style fgures ( tropes and schemes ) to produce difering linguis-
tic and cognitive levels of parallelism and deviation in order to
draw in, delight, and ultimately persuade listeners and hearers.
It also suggests diferent styles that are suitable for specifc occa-
sions: high, middle, or low. Te fourth and ffth canons set out the
performance aspects of rhetoric and involve oral, rather than
written, production. Tese are the memorizing and delivery of
a speech. Te latter puts much focus on intonation , prosody,
Rhetoric and Persuasion
717
others by such repetition. Rhymes are often classifed by the
kind of segment repeated, the exactness of the repetition, and
the location of the repetition within surrounding segments.
Masculine rhyme pairs words on fnal stressed syllables, like
kind/blind . Feminine rhyme pairs words on stressed syllables
with one or more identical unstressed syllables following, like
litter/bitter or authority/majority . In grammatical rhyme , the
word-fnal syllable(s) are similar infectional suf xes. In exact
or full rhyme , the repeated segments are highly similar while
the syllable onsets are not. Slant , half , or near rhyme allows vari-
ous deviations from strict similarity: in the syllable coda, such
as time/nine or light/rights ; in the nucleus, like want/faunt ; or
even in syllabic structure, as in lying/mine . End rhymes occur at
the ends of poetic lines (see verse line ), while internal rhymes
occur irregularly within a group of lines .
Assonance is the repetition of like vowels in neighboring syl-
lables. Nigel Fabb ( 1997 ) observes that some instances of asso-
nance could be reclassifed as rhyme in traditions that allow
sets of phonologically similar consonants in codas. Rhyme
and assonance usually, but not necessarily, involve stressed
syllables .
Rhyme and assonance are essentially perceptual phenom-
ena, and classifcations like those here typically serve to facili-
tate descriptions of their perception by listeners and readers.
Masculine rhymes register more vigorously with listeners and
feminine rhymes less so, and grammatical rhymes register very
little or not at all. Slant rhymes are less vigorous than full rhymes,
while internal rhymes are ordinarily less noticeable than end
rhymes. Te important question of efective distance between
repeated segments has not been closely researched. With asso-
nance, like alliteration , repetition outside of a rather narrow
structural window, such as a line or two adjacent lines, is not
likely to be perceived as echoic. Rhyme that demarcates poetic
lines has a larger efective window because the segmentation is
overdetermined by other operating schemes.
Outside of poetry and verbal play, rhyme and assonance
fgure little in linguistic organization. Tey sometimes help
motivate word coinage, as, for example, in the so-called
rhyme-motivated reduplications of English, like mumbo-
jumbo , teeny-weeny , or claptrap . Nevertheless, the human
facility for rhyme and rhyme recognition serves as an impor-
tant tool for linguistic investigation. psycholinguistics has
revealed that rhyme words are associated in memory, so that a
word prompts recall of its rhyme cohorts, and the presence of
a rhyming competitor of a target item delays visual identifca-
tion of that words referent. Phonologists have used the poetic
classifcation of half rhymes to investigate the perceptual basis
for phonological similarity judgments. Eve Sweetser ( 2006 )
suggests that rhyme relations between words can prompt
conceptual blending .
Claiborne Rice
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Fabb , Nigel . 1997. Linguistics and Literature: Language in the Verbal Arts
of the World . Oxford : Blackwell .
Sweetser , Eve . 2006. Whose rhyme is whose reason? Sound and sense in
Cyrano de Bergerac . Language and Literature 15 .1: 29 54.
discourse, under the premise that literary discourse is either
consciously or subconsciously stylized to order to induce cer-
tain persuasive emotive and aesthetic efects in readers. Te
rhetorical notion of foregrounding commonly through
such techniques as parallelism, repetition, and deviation is
central to this approach. Stylistics is the analytic side of the
literary rhetorical coin, the other being production, which
currently fnds form in university creative writing programs.
Stylistics has also made a journey similar to that of rhetoric
and persuasion: from a focus on production and form, to one of
interaction and then reception, and fnally on to the domain of
the mind, beliefs, and emotion. Tis can be seen in recent work
being conducted on cognitive stylistics, also often referred to
as cognitive poetics . Tis augmentation shows how lin-
guistic and cognitive approaches to stylistics are not contrary
but complementary (M. Burke 2005 , 194). Cognitive stylistics
thus explicitly seeks to integrate the language-oriented nature
of mainstream stylistic analysis with the mind-oriented nature
of emotion and cognitive linguistics . Tis is a line of
development that is at the very heart of twenty-frst century
rhetoric and persuasion .
Michael Burke
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Aristotle . 1991. Te Art of Rhetoric . Ed. Hugh Lawson-Tancred .
London : Penguin .
Burke , Kenneth . 1969. A Rhetoric of Motives . Berkley and Los
Angeles : University of California Press .
Burke , Michael . 2005. How cognition can augment stylistic analysis .
European Journal of English Studies 9 .2: 185 95.
Cockcroft , Robert , and S. Cockcroft . 2005. Persuading People: An
Introduction to Rhetoric . London : Palgrave . Tis practical book
explores persuasive techniques in English and provides clear links to
stylistics.
Eemeren , Frans H. van , and R. Grootendorst . 2004. A Systematic
Teory of Argumentation: Te Pragma-Dialectic Approach .
Cambridge : Cambridge University Press .
Kennedy , George A . 1994. A New History of Classical Rhetoric . Princeton,
NJ: Princeton University Press. Tis book brings together several of
Kennedys previous works to provide an engaging and informative
account of the history of rhetoric and persuasion.
Perelman , Chaim , and L. Olbrechts-Tyteca . 1969. Te New Rhetoric: A
Treatise on Argumentation . Notre Dame, IN : University of Notre Dame
Press .
Simons , Herbert, S . 2001. Persuasion in Society . London: Sage. Taking a
social sciences approach to communication, this book looks at rhet-
oric and persuasion in a number of everyday domains, such as adver-
tising, politics, law, and psychology.
Wales , Katie . 2001. A Dictionary of Stylistics . 2d ed. London : Longman .
RHYME AND ASSONANCE
Rhyme and assonance are patterns of phonetic similarity that
function primarily as devices of poetic euphony. In phonology ,
rhyme (sometimes distinguished by the spelling rime ) denotes
the unit comprising a vowel the nucleus and optional follow-
ing consonants the coda that follows an onset in a syllable .
In poetics , rhyme refers to the repetition of similar segments
beginning with a nucleus or to a word linked perceptually to
Rhyme and Assonance
718
consonantal intervals. On the contrary, European Spanish lacks
vowel reduction, and its syllabic structure is simpler, resulting in
a reversed pattern with a relatively higher proportion of vocalic
intervals and a lower duration variability in consonant intervals.
Several experimental studies have emphasized the salience of
these indices in language discrimination tasks performed by
human subjects (e.g., Ramus, Dupoux, and Mehler 2003 ), and
an abundant literature has followed. However, further investiga-
tion is still needed for understanding some dynamical aspects
of rhythm (metric patterns, speech rate, etc.) and the possible
interaction among intensity, pitch , and duration, explicitly in
terms of rhythm .
RELATION AMONG RHYTHM, METRICS, AND STRESS. As Anthony
Fox pointed out, [rhythm] is rarely taken into account in a for-
mal way in phonological theory and description ( 2000 , 86).
However, nonlinear approaches, and especially metrical pho-
nology (see meter ), take rhythm into consideration by investi-
gating both the structure and the weight of rhythmic constituents
(e.g., Hyman 1985 ; Blevins 1995 ) and their relation to metric and
stress patterns (Hayes 1995 ).
WHY ARE LANGUAGES RHYTHMIC? Tere is extensive evidence
in support of speech rhythmicity as fundamental for speech
communication. From the production side, no uncontrover-
sial position has yet emerged. For instance, Peter MacNeilage
( 1998 ) proposed an evolutionary scenario deriving speech
rhythmicity from cycles of mandibular oscillation during
ingestion; Robert F. Port ( 2003 ) proposed that neurocognitive
oscillators could synchronize the production of prominent
events with perceptual attention, renewing approaches initi-
ated in psychology (for a review, see Evans and Clynes 1986 ).
Furthermore, quite a few experiments have assessed human
awareness of rhythm diferences for neonates, young infants,
or adults, and several studies suggest that rhythm plays an
important role in segmenting the speech stream and thus for
language acquisition (see, among others, Morgan and Demuth
1996 and Mehler and Nespor 2004 ; see also speech percep-
tion in infants ) .
Franois Pellegrino
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Abercrombie , David . 1967. Elements of General Phonetics .
Edinburgh : Edinburgh University Press .
Blevins , Juliette . 1995. Te syllable in phonological theory. In
Handbook of Phonological Teory , ed. John A. Goldsmith, 20644.
Oxford : Blackwell .
Dauer , Rebecca M . 1983. Stress-timing and syllable-timing revisited .
Journal of Phonetics 11 : 51 62.
Evans , James R. , and M. Clynes , eds. 1986. Rhythm in Psychological,
Linguistic and Musical Processes . Springfeld, IL : Charles C. Tomas .
Fox , Anthony . 2000. Prosodic Features and Prosodic Structure: Te
Phonology of Suprasegmentals . Oxford : Oxford University Press .
Grabe , Esther , and E. L. Low . 2002. Durational variability in speech and
the rhythm class hypothesis. In Papers in Laboratory Phonology 7 , ed.
Carlos Gussenhoven and Natasha Warner , 51546. Berlin : Mouton de
Gruyter.
Hayes , Bruce . 1995. Metrical Stress Teory . Chicago : University of Chicago
Press .
RHYTHM
Te notion of rhythm is widely present in language sciences and
an abundant literature ranging from acoustics to phonological
theory and neuropsychology is available, leading to several and
sometimes conficting defnitions. Nonetheless, most would
agree that rhythm involves the temporal organization of speech
and results from a threefold complex interaction among
the nature of the rhythmic atomic constituents;
the use of alternations between more and less prominent
constituents;
the pattern of regularity for the grouping of the constituents
into longer units.
According to this defnition, rhythm is fundamental to languages
(it seems that no language may be defned as arrhythmic, even if
the last two proposed dimensions may be irrelevant in specifc
languages) .
In the twentieth century, phonetics searched mainly for the
acoustic correlates of rhythm units, while phonology with the
exception of metrical phonology usually considered rhythm as
a mere sequence of timing slots on which linguistic properties
are cast. In addition, cognitive science addressed distinct ques-
tions, namely, why languages are rhythmic and whether rhythm
plays a role in the cognitive processing of language .
Current State of the Field
RHYTHMIC TYPOLOGY. Te long-lasting view that speech rhythm
would consist in isochronous occurrences of some acoustic
event or unit along the speech stream, popularized by Kenneth
L. Pike ( 1945 ) and David Abercrombie ( 1967 ), is now widely
rejected. Explaining how rhythm is perceptually salient despite
the absence of objective regularity and why it seems neverthe-
less possible to gather languages into a few rhythmic categories
is consequently a challenging issue. Tese categories, initially
known as stress-, syllable- and mora- timed , have been renamed
stress-, syllable- and mora- based in Laver ( 1994 , 5289). Tis
distinction is enlightening about the change from a discrete
to a continuous approach to rhythm variation across lan-
guages: Rhythmic typology has to cope with languages that
do not strictly match categorical prototypes, and there is now
general agreement that this typology better refects tendencies,
rather than mutually exclusive categories (Roach 1982 ; Dauer
1983 ). According to Rebecca M. Dauer, the diference between
stress-timed and syllable-timed languages has to do with difer-
ences in syllable structure, vowel reduction and the phonetic
realization of stress and its infuence on the linguistic system
( 1983 , 51). In other words, she states that typological diferences
in rhythm are side efects of the phonological characteristics of
languages.
Over the last decade, the durational correlates of rhythm
types have been thoroughly investigated, highlighting in par-
ticular distributional properties of vocalic and intervocalic seg-
ment durations (Ramus, Nespor, and Mehler 1999 ) and the
pairwise variability of these segment durations (Grabe and Low
2002 ). For example, British English exhibits both vowel reduc-
tion and fairly complex syllable structure, yielding a low propor-
tion of vocalic intervals and a high variation in the duration of
Rhythm
719
neuroimaging in various modalities often shows stronger
brain activity in the LH than in the RH during most stan-
dard language tasks. However, there usually is some activity
in homologous regions of the RH, even if it is weaker (e.g.,
Mazoyer et al. 1993 ).
All of this information (derived from lesions, divided inputs
to LH and RH, and neuroimaging) reveals that the RH seems
important for processing many aspects of prosody and the
melodic contour of language used to stress some elements,
conveying emotion, or (in many languages) demarcating
questions and sentences (Joanette, Goulet, and Hannequin
1990 ; cf. Schirmer and Kotz, 2006 ).
Likewise, a variety of evidence demonstrates that the RH
strongly contributes to some language tasks, particularly in
higher-level language comprehension.
Tis fnal point concerns the vast majority of work on RH lan-
guage processing, and it demands elaboration. Lesion studies,
neuroimaging, and behavioral evidence all indicate that RH lan-
guage processing is important for drawing connective inferences
(Beeman 1993 ; Brownell et al. 1986 ; Mason and Just 2004 ; Virtue
et al. 2006); understanding jokes (Coulson and Wu 2005 ; Goel
and Dolan 2001 ) and at least certain types of metaphors ( Brain
and Language special issue in 2007; see metaphor, neural
substrates of ); integrating new input with previously com-
prehended text (Robertson et al. 2000 ); and comprehending the
gist of stories or conversations (St. George et al. 1999 ). In other
words, many of the truly communicative aspects of language
seem to rely in some good measure on RH function. Moreover,
paradigmatic tasks known to elicit primarily LH brain activity
can be slightly adjusted to elicit increases in RH activity, if more
distant semantic associations are involved. For instance, gen-
erating the frst verb that comes to mind when reading a noun
elicits predominantly LH frontal activity, but generating a novel
or unusually associated verb predominately increases activity in
the RH, compared to generating the frst verb (Abduallaev and
Posner 1997 ; Seger et al. 2000 ).
Any successful theory of language asymmetries must account
for all of these facts. One long-favored view is that only the LH
possesses the requisite neural machinery for language process-
ing. A common explanation is that evolution favored a single
hemisphere to control speech, and that other aspects of lan-
guage processing could interact with speech more ef ciently if
organized within the same hemisphere. By evolutionary chance
(or perhaps for reasons associated with right-hand tool use), the
LH gained control of speech, and thus the brain evolved to have
language centers (performing various component processes that
continue to be enumerated) only within the LH. On this view, all
contributions of the RH for understanding language are merely
paralinguistic, that is, outside the realm of pure language pro-
cessing. It is easy to see how this applies to prosody; moreover,
it is often also applied to the higher-level language processes for
which the RH seems important, described previously.
This LH-only view has the advantage of easily explaining
the association of classic aphasia with LH lesions. However,
it suffers from several disadvantages. First, it is merely
descriptive, failing to explain how the LH, but not the RH, is
able to process language. It may be true that the left planum
Hyman Larry M . 1985. A Teory of Phonological Weight . Dordrecht, the
Netherlands : Foris .
Laver , John . 1994. Principles of Phonetics . Cambridge : Cambridge
University Press .
MacNeilage , Peter . 1998. Te frame/content theory of evolution of
speech production . Behavioral and Brain Sciences 21 : 499 546.
Mehler , Jacques , and M. Nespor . 2004. Linguistic rhythm and the devel-
opment of language. In Structures and Beyond: Te Cartography of
Syntactic Structures , ed. A. Belletti , 21321. Oxford : Oxford University
Press .
Morgan , James L. , and K. Demuth , eds. 1996. Signal to Syntax: Boot-
strapping from Speech to Grammar in Early Acquisition . Mahwah,
NJ : Lawrence Erlbaum .
Pike , Kenneth L. 1945. Te Intonation of American English . Ann
Arbor : University of Michigan Press .
Port , Robert F. 2003. Meter and speech . Journal of Phonetics
31 : 599 611.
Ramus , Franck , E. Dupoux , and J. Mehler . 2003. Te psychological real-
ity of rhythm classes: Perceptual studies. In Proceedings of the 15th
International Congress of Phonetic Sciences , ed. Maria-Josep Sol ,
Daniel Recasens , and Joaqun Romero , 33742. Barcelona: Causal
Productions.
Ramus , Franck , M. Nespor , and J. Mehler . 1999. Correlates of linguistic
rhythm in the speech signal . Cognition 73 : 265 92.
Roach , Peter . 1982. On the distinction between stress-timed and syl-
lable-timed languages. In Linguistic Controversies , ed. David Crystal,
739. London : Edward Arnold .
RIGHT HEMISPHERE LANGUAGE PROCESSING
Te human brain is composed of two cerebral hemispheres,
which are anatomically very similar but have well-known
asymmetries in terms of their functional capacities. One of
the best-known facts about the interface between the brain
and cognition is that the left hemisphere (LH) is dominant
for language. (Tis description applies to the vast majority of
right-handed people, and probably at least two-thirds of left-
handers.) However, the nature of this left hemisphere language
dominance, and of right hemisphere (RH) contributions to lan-
guage, remains far from clear, and several schools of thought
persevere. What is known:
aphasia, a marked defcit in language function following
brain damage, is almost always associated with damage to
the LH, whereas similar damage to the RH is not associated
with overt aphasia.
Yet, people with aphasia following extensive LH brain damage
often recover at least some linguistic function, and in some
(but not all) cases, homologous areas of the RH (those areas
mirroring the damaged LH language areas) seem responsible
for recovery (Blasi et al. 2002 ; Lef et al. 2002 ). Indeed, chil-
dren who have their entire LH removed as a last resort to treat
a crippling form of epilepsy often go on to develop language
abilities within a normal range (Vargha-Khadem et al. 1997 ).
Neurologically intact individuals can perform most language
tasks better (faster and more accurately) when processing lin-
guistic stimuli received via the right visual feld or the right
ear, both of which project most directly to the LH, compared
to stimuli received via the left visual feld or left ear, which
project to the RH.
Right Hemisphere Language Processing
720
means a part of the body or a unit of measurement. Also, LH
relatively fne semantic coding is unlikely to maintain activa-
tion of semantic features peripherally related to foot, such as
the fact that it is susceptible to injury.
In contrast, the theory asserts that relatively coarse cod-
ing of information by RH language areas poorly serves such
selection, because multiple lexical and phonological repre-
sentations remain active. Tis is clearly disadvantageous for
speech output. Coarser semantic coding explains why the
RH shows more sensitivity to distant semantic associations in
behavioral priming (Chiarello et al. 2003 ) and electrophysi-
ological studies (Federmeier and Kutas 1999 ). Tus, coarser
coding allows for multiple word meanings, shades of mean-
ing, and seemingly distant associations to remain active and
contribute to the aspects of language comprehension that
are relatively dependent on RH processing. For example, RH
semantic processing is conducive to recognizing semantic
overlap from multiple words that are each distantly related
to a primed concept (Beeman et al. 1994 ). Tus, when people
hear Joan walked barefoot near some glass, then felt a stab
of pain, coarser coding may help detect the semantic overlap
that helps people infer that she cut her foot; as noted, multiple
lines of evidence suggest the RH contributes to such connec-
tive inferences in important ways .
Some theorists also claim that computational asymmetries
in language processing arise from documented asymmetries in
neural microcircuitry of language areas, such as broader den-
dritic branching that expands input felds to pyramidal neurons
(which constitute the majority of excitatory neurons in the cor-
tex), thus proposing a causal mechanism (Jung-Beeman 2005 ;
see also Hutsler and Galuske 2003 ). Furthermore, recovery from
aphasia is readily explained by small changes in previously
asymmetric microcircuitry, such as a slight retuning of dendritic
branching, rather than wholesale rewiring of areas that were pre-
viously completely nonlinguistic, or merely paralinguistic .
Mark Beeman
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Abduallaev , Yalchin , and Michael I. Posner . 1997. Time course of acti-
vating brain areas in generating verbal associations . Psychol. Sci .
8 : 56 9.
Beeman , Mark . 1993. Semantic processing in the right hemisphere may
contribute to drawing inferences during comprehension . Brain Lang .
44 : 80 120.
Beeman , Mark, and Christine Chiarello , eds. 1998. Right Hemisphere
Language Comprehension: Perspectives from Cognitive Neuroscience .
Mahwah NJ : Lawrence Erlbaum . Tis book contains chapters on hemi-
spheric diferences in many diferent language functions, focusing on
right hemisphere contributions.
Beeman , Mark . et al . 1994. Summation priming and coarse coding in the
right hemisphere . J. Cog. Neurosci . 6 : 26 45.
Blasi , Valeria , et al. 2002. Word retrieval learning modulates right frontal
cortex in patients with frontal damage . Neuron 36 : 159 70.
Brain and Language 100 .2. 2007. Special issue on metaphor processing.
Brownell , Hiram H. , et al . 1986. Inference defcits in right brain-dam-
aged patients . Brain Lang . 29 : 310 21.
Chiarello , Christine , et al. 2003. Priming of strong semantic rela-
tions in the left and right visual felds: A time course investigation .
Neuropsychologia 41 : 721 32.
temporale (near Wernickes area) or the left Brocas area, or
both, are slightly larger than their RH counterparts, but these
asymmetries seem imperfectly correlated with language lat-
erality (e.g., Gannon et al. 1998 ; Moffat, Hampson, and Lee
1998 ). Second, the LH-only view has difficulty parsimoniously
accounting for RH involvement in language recovery from
aphasia following LH brain damage (or entire hemispherec-
tomy in children). Although the brain is famously plastic, there
are limits to plasticity, and it seems prima facie unlikely that a
brain area having nothing to do with language prior to injury
(e.g., one that processes visuospatial information) could, after
LH injury, rewire itself to suddenly perform language process-
ing. Furthermore, the LH-only view does not incorporate lan-
guage comprehension subprocesses to which the RH is known
to contribute.
Te LH superiority view is a slightly less extreme theory,
maintaining that the LH is greatly superior to the RH at most
or all language processing, although there may be a range of
RH linguistic ability (or bilaterality) among diferent individu-
als. Some theorists hold a hybrid view, that the RH is inferior at
some processes but completely lacking in others, such as syn-
tactic processing or speech production. Such LH superior views
describe fairly well, without explaining, the aforementioned
phenomena regarding language asymmetries, and may better
account for RH involvement in language recovery. Specifcally,
patients who sufer extensive damage to the LH early in life
may recover language very well because developmental plas-
ticity permits a rewiring of the rudimentary language systems
in the RH to take over that function. Tis is more likely than,
say, rewiring a visuospatial processing system to suddenly per-
form language function. However, while plausible, it seems
uneconomical for large cortical regions to be poorly function-
ing backup systems that only retune to serve useful purposes
in the event that the primary language systems, all in the LH,
are severely impaired .
Finally, a more recent class of theories is that the two hemi-
spheres both process all (or nearly all) aspects of language but
compute the information in slightly diferent manners, each type
of computation conveying advantages for diferent sets of tasks.
Such computational asymmetry views do not posit equality in
language ability, but rather assert that when cooperating with
an intact LH, the RH makes unique and signifcant contributions
to language processing. Furthermore, they provide, to varying
degrees, some causal explanation for the observed asymmetries
of language function .
For instance, one theory (Jung-Beeman 2005) asserts that rel-
atively fne coding of information by LH language areas permits
rapid categorization and selection of information (Tompson-
Schill et al . 1997 ) that is fundamental to many language func-
tions. When speaking a word, for instance, people must select
one lexical and phonological representation to articulate; for
example, they must choose between cup and mug , but the pre-
cise duration, stress, and accent are less important. During
comprehension, upon hearing the word foot , LH fne coding
will strongly activate a small set of semantic features strongly
related to the context (or to the dominant meaning, in absence
of context). Tis is clearly benefcial, as comprehenders would
quickly get bogged down if they fail to discern whether foot
Right Hemisphere Language Processing
721
theoretical and descriptive constructs derive from the investiga-
tion of many non-Indo-European languages, yielding a theory
with rather diferent approaches to the analysis and explanation
of morphosyntactic phenomena. RRG posits only a single level
of syntactic representation, corresponding to the actual form of
the sentence; there are no abstract levels of representations or
derivations, as in Chomskyan generative grammar. Moreover,
no phonologically null elements are allowed in syntactic rep-
resentations. Te name of the theory comes from its view that
much of grammar is an interaction between semantics (role) and
discourse-pragmatics (reference). Te organization of the theory
is in Figure 1 . Tere is a mapping between the syntactic and
semantic representations, mediated by the linking algorithm.
Discourse-pragmatics plays a role in this mapping, but the role
varies across languages, leading to important typological difer-
ences among languages. It is represented in part by discourse
representation structures (DRSs) taken from discourse repre-
sentation theory .
Te syntactic representation is called the layered structure of
the clause and is composed of three interlocking projections: the
constituent projection (predicate, arguments, adjuncts), the
operator projection (grammatical categories like aspect, nega-
tion, tense), and the focus structure projection (information
structure of the sentence, that is, the potential focus domain
of the clause [where focus can potentially fall], the actual focus
domain, and the presupposed, non-focus part). Te semantic
representation is an Aktionsart -based decompositional rep-
resentation. thematic relations are defned in terms of posi-
tions in the decompositions. More important are the semantic
macroroles of actor and undergoer, which correspond to the
two primary arguments in a transitive predication, either one of
which can be the single argument of an intransitive predicate.
Based on extensive typological evidence, RRG rejects traditional
grammatical relations and instead posits a construction-specifc
concept, the privileged syntactic argument (PSA) of a particular
construction .
Te linking between syntax and semantics is bidirectional,
refecting production and comprehension. In semantics-to-
syntax linking, the elements in the semantic representation are
mapped into positions in a syntactic template, following both
universal and language-specifc principles. An example from
English is given in Figure 2 . In syntax-to-semantics linking,
information about the syntactic-semantic function of arguments
is derived from the overt morphosyntactic markings, and this is
linked to the argument positions in the lexical representation of
the predicate; a simple example is in Figure 3 .
Coulson , Seana , and Wu Y. C . 2005. Right hemisphere activation of joke-
related information: An event-related brain potential study . J. Cog.
Neurosci . 17 : 494 506.
Federmeier , Kara D. , and Marta Kutas . 1999. Right words and left
words: Electrophysiological evidence for hemispheric diferences in
meaning processing . Cogn. Brain Res . 8 : 373 92.
Ferstl , Evelyn C. , et al . 2005. Emotional and temporal aspects of situation
model processing during text comprehension: An event-related fMRI
study . J Cog. Neurosci . 17 : 724 39.
Gannon , Patrick J. , et al. 1998. Asymmetry of chimpanzee planum tem-
porale: Humanlike pattern of Wernickes brain language area homo-
log . Science 279 : 220 2.
Goel , Vinod , and Raymond J. Dolan . 2001. Te functional anatomy of
humor: Segregating cognitive and afective components . Nature
Neurosci . 4 : 237 8.
Hutsler , Jefrey , and R. A. W. Galuske. 2003. Hemispheric asymmetries
in cerebral cortical networks . Trends Neurosci . 26 : 429 36.
Joanette , Yves , Pierre Goulet , and Didier Hannequin . 1990. Right
Hemisphere and Verbal Communication . New York: Springer-Verlag.
Jung-Beeman , Mark . 2005. Bilateral brain processes for comprehending
natural language . Trends in Cognitive Sciences 9 : 512 18. Tis paper
proposes a neural microcircuitry explanation of computational asym-
metries in language processing.
Lef , Alexander , et al . 2002. A physiological change in the homotopic
cortex following left posterior temporal lobe infarction . Ann. Neurol .
51 : 443 558.
Mason , Robert A. , and Marcel Just . 2004. How the brain processes causal
inferences in text: A theoretical account of generation and integration
component processes utilizing both cerebral hemispheres . Psychol.
Sci . 15 : 1 7
Mazoyer , B. M , et al . 1993. Te cortical representation of speech . J. Cog.
Neurosci . 5 : 467 79.
Mofat , Scott D. , Elizabeth Hampson , and Donald H. Lee . 1998.
Morphology of the planum temporale and corpus callosum in left
handers with evidence of left and right hemisphere speech representa-
tion . Brain 121 : 2369 79.
Robertson , David A. , et al . 2000. Functional neuroanatomy of the cog-
nitive process of mapping during discourse comprehension . Psychol.
Sci . 11 : 255 60.
Schirmer , Annett , and Sonja A. Kotz . 2006. Beyond the right hemi-
sphere: Brain mechanisms mediating vocal emotional processing .
Trends in Cognitive Sciences 10 : 24 30.
Seger , Carol A. , et al . 2000. Functional magnetic resonance imaging
evidence for right hemisphere involvement in processing unusual
semantic relationships . Neuropsy. 14 : 361 9.
St. George , Marie , et al . 1999. Semantic integration in read-
ing: Engagement of the right hemisphere during discourse process-
ing . Brain 122 : 1317 25.
Tompson-Schill , Sharon L. , et al. 1997. Role of left inferior prefrontal
cortex in retrieval of semantic knowledge: A re-evaluation . Proc. Nat.
Acad. Sci . 94 : 14792 7.
Vargha-Khadem , Faraneh , et al . 1997. Onset of speech after left hemi-
spherectomy in a nine-year old-boy . Brain 120 : 159 82.
Virtue, S., J. Haberman, Z. Clancy, T. Parrish, and M. Jung-Beeman. 2006.
Neural activity of inferences during story comprehension. Brain
Research 1084 : 10414.
Xu , J. et al. 2005. Language in context: Emergent features of word, sen-
tence, and narrative comprehension . NeuroImage 25 : 1002 15.
ROLE AND REFERENCE GRAMMAR
Role and reference grammar (RRG) is a theory of the syntax-
semantics-pragmatics interface (Van Valin 2005 ); its
Linking
Algorithm
SYNTACTIC REPRESENTATION
D
i
s
c
o
u
r
s
e
-
P
r
a
g
m
a
t
i
c
s
SEMANTIC REPRESENTATION
Figure 1. The organization of RRG.
Role and Reference Grammar
722
WORKS CITED AND SUGGESTIONS FOR FURTHER READING
Bornkessel, Ina, Matthias Schlesewsky, and Robert D. Van Valin, Jr. 2004.
Syntactic templates and linking mechanisms: A new approach to
grammatical function asymmetries. Available online at: http://lin-
guistics.buffalo.edu/people/faculty/vanvalin/rrg/vanvalin_papers/
rrg_adm_CUNY04.pdf.
Van Valin , Robert D. , Jr. 1998. Te acquisition of wh-questions and
the mechanisms of language acquisition. In Te New Psychology of
Language: Cognitive and Functional Approaches to Language Structure ,
ed. Michael Tomasello, 22149. Mahwah, NJ: Lawrence Erlbaum.
Available online at: http://linguistics.bufalo.edu/people/faculty/van-
valin/rrg/vanvalin_papers/acqwh.pdf.
. 2001. Te acquisition of complex sentences: A case study in the
role of theory in the study of language development. Chicago Linguistic
Society Parasession Papers 36 : 51131. Available online at: http://lin-
guistics.buffalo.edu/people/faculty/vanvalin/rrg/vanvalin_papers/
Acq_of_complex_sent.pdf.
. 2005. Exploring the Syntax-Semantics Interface .
Cambridge : Cambridge University Press .
. 2006. Semantic macroroles and language processing. In Semantic
Role Universals and Argument Linking: Teoretical, Typological and
Psycho-/Neurolinguistic Perspectives , ed. Ina Bornkessel, Matthias
Schlesewsky, Bernard Comrie, and Angela Friederici, 263302.
Berlin: Mouton de Gruyter. Available online at: http://linguistics.
buffalo.edu/people/faculty/vanvalin/rrg/vanvalin_papers/Sem_
Macroroles_&_Lang_Processing.pdf.
Te best source for work and references relating to role and ref-
erence grammar is the RRG Web site. Available online at:
http://linguistics.bufalo.edu/research/rrg.html.
Not only has work been done using RRG on a wide vari-
ety of languages, but there has also been psycholinguistic and
neurolinguistic work done as well. RRG takes a cognitivist
approach to language acquisition (see Van Valin 1998, 2001),
as well as references on the RRG Web site). The application
of RRG to the study of sentence processing is investigated in
Bornkessel, Schlesewsky, and Van Valin (2004) and Van Valin
(2006).
Robert D. Van Valin , Jr.
LEXICON
SYNTACTIC
PERIPHERY
INVENTORY
ACTOR Non-Macrorole UNDERGOER
PSA:NOM ACTIVE: 3sg to: ACC ACC at: ACC
be-at(party , [[do(Sandy , )] CAUSE [BECOME have(Chris , flowers )]]
ACV ACV
INA INA
CLAUSE
CORE <
PRED
SENTENCE
NUCLEUS
V
NP
PP NP PP
SPEECH ACT
Presupposition Assertion
Predicate focus
present x to y
w, x, y, z
Sandy(w)
flowers(x)
Chris(y)
party(z)
w present x to y at z
w, z
Sandy(w)
party(z)
w P at z
at the party Sandy presented the flowers to Chris
Figure 2. Linking from semantics to syntax. Constituent and focus structure projections shown, along with discourse
representation structures.
did Mary give to John yesterday What
Parser
Lexicon
Actor
NP
ADV
V
PRED
NUC
<---------
CLAUSE
SENTENCE
PrCS
Voice? -- Active
PSA = Actor
CORE
NP PP
PERIPHERY
[do(x, )] CAUSE [BECOME have (y,z)]
Figure 3. Linking from syntax to semantics.
Role and Reference Grammar
723
following manner: When I learned to add, and so learned what
plus (and +) mean, I didnt merely memorize the answers to
a fnite list of addition problems. I learned a rule for addition (and
so for using plus or +) that determines the sum of any pair of
numbers, regardless of whether I happen to have yet encoun-
tered them. Tat the correct answer to What is 68 + 57? is 125,
rather than 5, is determined by this rule a rule that, Kripke sug-
gests, might have taken the following form :
Take a huge bunch of marbles. First count out x marbles in one
heap. Ten count out y marbles in another. Put the two heaps
together and count out the number of marbles in the union thus
formed. Te result is x + y . (Kripke 1982 , 15)
Te problem with this sort of answer to the skeptics chal-
lenge is not that I may be misremembering the rule the inter-
pretation of + that I learned as a child. Te skeptic is happy
to grant that as a child I learned precisely this formulation and
that I remember it clearly to this day. But he questions how this
rule lets call it R ought to be understood. While the skep-
tic allows that R is the interpretation that Ive long assigned to
the plus sign, he claims that I am, at present, misapplying R .
He suggests that given what, in the past, I meant by the word
count, the result that a correct application of R would yield,
given 68 + 57, is 5.
At this point, I might reply to the skeptic by recalling an
interpretation of the word count that I also internalized as a
child. But he will just make the same sort of move again: Hell
grant me the remembered interpretation and suggest that I
am, at present, misapplying it. Tus, each new interpretation
that I adduce seems to require another one standing behind it,
and an infnite regress threatens. Te moral the problem with
what might be called interpretationalism may be put as fol-
lows: However tempting it is to say that our words derive their
meanings from rules or interpretations, saying this just leaves
us with the question of where these rules or interpretations get
their meanings.
Dispositionalism
Another kind of reply to Kripkes skeptic would appeal to dis-
positions rather than to interpretations or rules. Tus, the skep-
tics challenge might be answered as follows: While I have never
before added 68 and 57, its nonetheless true that had I been
asked to add these numbers, I would have arrived at the answer
125. My having meant plus (rather than quus ) by plus consists
in my having been so disposed disposed to answer questions
of the form What is x plus y ? with the sum (and not the quum)
of x and y .
A problem with this sort of dispositionalism becomes appar-
ent when one considers the fact that speakers are sometimes
disposed to make mistakes. Imagine that when Im sleepy, I
am disposed to answer 115 if someone asks me to add 68 and
57. We dont want an account of meaning according to which
it would follow from this that when Im sleepy, what I mean by
plus is not addition but, instead, some function that yields 115,
given 68 and 57 as arguments. Or consider a nonmathematical
example: It might be that I have a disposition to answer the ques-
tion Is that a horse? af rmatively when Im shown an albino
RULE-FOLLOWING
In Ludwig Wittgensteins late writings, one fnds numerous
interconnected remarks having to do with meaning, under-
standing, and rule-following remarks in which Wittgenstein
can seem to be continually circling around these topics with-
out ever arriving at any defnite conclusion. An exceptionally
clear and compelling way into this material was provided by
Saul Kripke in his infuential 1982 book, Wittgenstein on Rules
and Private Language . By presenting Wittgensteins concerns
in the form of a single extended argument, Kripke brought
many readers to see that Wittgensteins discussions of rule-
following bear signifcantly on central issues in (among other
things) contemporary philosophy of language. As Kripke
recasts Wittgensteins remarks, they amount to an argument
for a radical form of skepticism a skepticism according to
which there are no facts about what we mean by any of our
utterances or inscriptions. In what follows, a sketch is provided
of the skeptical argument that Kripke fnds in Wittgensteins
remarks on rule-following. Tat done, an efort is made to
distinguish the actual, historical Wittgenstein from Kripkes
reconstruction of him.
The Skeptics Challenge
Suppose that Im asked to add two numbers that I happen never
before to have added together: Im asked (lets say) Whats 68
plus 57? I think for a moment and reply 125. Kripke imag-
ines a bizarre sceptic ( 1982 , 8) who contests this answer. Te
skeptic suggests that, given what Ive always meant by the word
plus in the past, the correct answer to Whats 68 plus 57? is
not 125. For in the past, he says, what I meant by plus might
have been a function call it the quus function that yields
5, rather than 125, given the arguments 68 and 57. By hypoth-
esis, the quus function is consistent with all of the sums that I
have computed before today. (Imagine that the only diference
between the quus function and the plus function is that 68 quus
57 that is, the quum of 68 and 57 is 5.) Kripkes skeptic chal-
lenges me to show that, in the past, when I used the word plus
(or addition or sum), what I meant was plus ( addition, sum)
rather than quus ( quaddition, quum ) . Im to show this by citing a
fact or set of facts about my former self.
Although the skeptics challenge concerns my use of a math-
ematical expression, it is important to understand that whats
at issue has nothing especially to do with mathematics. I might
just as well have been challenged to show that in the past I
meant horse by horse. If the skeptic can secure his conclu-
sion concerning my past uses of plus, the point will general-
ize from plus to other words and from past uses to present
ones. Te conclusion he seeks is that there are no facts of the
matter about what I, or anyone, ever meant, or mean, by any
utterance or inscription. Te argument may be understood to
proceed by a process of elimination. Kripke adduces various
answers to the skeptics challenge and, one by one, shows each
to be unsatisfactory .
Interpretationalism
Perhaps the most natural reply to the skeptic would make appeal
to the notion of a rule or interpretation in something like the
Rule-Following
724
behind our in-themselves-empty utterances and inscriptions
super interpretations that, unlike mere vocalizations or marks
on a page, neither need nor brook any further interpretation.
Wittgenstein characterizes the impulse toward this sort of posi-
tion when he writes, What one wishes to say is: Every sign is
capable of interpretation; but the meaning mustnt be capable of
interpretation. It is the last interpretation. (Wittgenstein 1958 ,
34). Platonism is a target of criticism in his late writings. But it is
a mistake to read Wittgenstein as holding that in order to reject
reductionism of the sort that Kripkes skeptic presupposes, one
must embrace Platonism.
Getting Wittgensteins response to Platonism into view
requires that one appreciate how much the semantic skeptic,
the interpretationalist, the dispositionalist, and the platonist
have in common . All four see our communicative activities
as the production of in-themselves-meaningless movements,
noises, and marks. All four agree, in effect, with Wittgensteins
interlocutor at Phil