Вы находитесь на странице: 1из 53

ADJECTIVAL CONSTRUCTIONS IN ENGLISH

ISSUES OF FORM AND MEANING

By
ATHANASIOS LITSOS

A Thesis
submitted to the School of English
Aristotle University of Thessaloniki
in partial fulfilment of the requirements

for the degree of

MA in LINGUISTICS

Thesis Supervisor: ANGELIKI ATHANASIADOU

Thessaloniki
OCTOBER 2018
i
TABLE OF CONTENTS

LIST OF TABLES III

ACKNOWLEDGEMENTS IV

ABSTRACT V

CHAPTER 1 INTRODUCING THE CLASS OF ADJECTIVES 1

CHAPTER 2 INTERPRETATION IN ATTRIBUTIVE AND PREDICATIVE POSITION 5

2.1 APPROACHES TO THE INTERPRETATION OF ADJECTIVAL CONSTRUCTIONS 5


2.2 PREDICATIVE AND ATTRIBUTIVE POSITION 8

CHAPTER 3 INTERPLAY OF MEANING AND POSITION IN THE PREMODIFIER ZONE 12

3.1 TYPES OF PROPERTIES 12


3.2 SEMANTIC TYPES AND THE SEQUENCE OF ADJECTIVES 14
3.3. THE SUBJECTIVE-OBJECTIVE SPLIT IN THE NP 17
3.4 A FUNCTIONAL TREATMENT OF ADJECTIVES 19
3.5 ELUCIDATING THE FUNCTIONAL TREATMENT 24
3.5.1 Scalar adjectives 24
3.5.2 Deadverbial adjectives 25
3.5.3 Denominal adjectives 26
3.5.4 Determining adjectives 27
3.5.5 Iconicity and position 27
3.5.6 Subjectivity and property stability 28
3.6 COMMENTS ON THE FUNCTIONAL APPROACH 29

CHAPTER 4 PRENOMINAL AND POSTNOMINAL POSITION 31

4.1 ADJECTIVES BEFORE AND AFTER THE NOUN 31


4.2 PARTICIPLES 35

CHAPTER 5 CONCLUSION 39

REFERENCES 45

ii
LIST OF TABLES

TABLES
PAGES
1. The continuum from category to non-restrictive properties. 13
2. Correlations between function, modifier and subjectivity
from the left to the right of the NP. 20
3. The order of adjectives according to the iconic principle
of proximity. 28
4. The premodifier construction in the NP with positions of
adjectives from left to right. 40
5. The continua of property stability and subjectivity and
their relation to the positions of the premodifier construction. 40

iii
ACKNOWLEDGEMENTS

The completion of this thesis was a very interesting but also demanding and challenging
task. It would probably not have come to this point, if it wasn‟t for some very important
people. First and foremost, I would like to thank my supervisor, Angeliki Athanasiadou,
for a number of very important things. First, I want to thank her for introducing me to
linguistics as an undergraduate student and stimulating my enthusiasm for this field of
study, which ultimately led me to pursue postgraduate studies. Second, I want to thank
her for her enthusiasm on my work as well as her points and patience to read really
badly written drafts. Her comments as well as our discussion have been extremely
useful for the completion this work. Any shortcomings of this thesis are completely my
own responsibility. And third, I want to thank her for her unfailing support and
confidence in me in a really tough period of my life.
Further, I would like to thank the department of Theoretical and Applied
Linguistics of the School of English. I have been a student there for seven years,
including my undergraduate studies. The insights that the academic staff has provided
me with over the years on a number of subjects in linguistics have been extremely rich
and continuously furthering my interests in this science. I would like, then, to thank the
following professors, Katerina Nicolaidis, Nikos Lavidas, Eleni Agathopoulou,
Michalis Milapides, Tasos Tsangalidis and Marina Mattheoudakis as well as the
following now-retired professors, Eliza Koutoupi-Kitis, Marianthi Makri-Tsilipakou
and Anna Anastasiadi-Symeonidi from the School of Philology. I am really grateful to
them for all the knowledge they have shared with me over all these years and which has
really shaped my intellectual personality.
Finally, I would like to thank my family, my parents, Xenia and Triandafyllos, as
well as my brother Michael, for having continuous love and faith in me and also dealing
with my problems and stress on a daily basis. I really cannot describe how immensely
grateful I am to them for their never-failing support in both economic and emotional
terms. This thesis is whole-heartedly dedicated to them.

iv
ABSTRACT

This thesis is concerned with the position and interpretation of adjectival constructions
in English. Issues stemming from the interpretation of adjectives are discussed in
conjunction with the positions occupied by them. The positions that are investigated are
the predicative position in the sentence and the premodifier and postmodifier zones in
the noun phrase. Through the analysis, it is evident that not all kinds of meanings are
available in all positions. The conclusion of this thesis is that these three positions are,
in fact, constructions in the fashion of Construction Grammar (Goldberg 1995). In
particular, the premodifier zone possesses a number of specific slots associated with
specific adjective functions, which can express a wide range of interpretations with the
head noun. On the other hand, the predicative position can mostly express meanings
related to the extra-linguistic reference of the noun. Last, the postmodifier zone also
possesses two slots associated with particular interpretations and, thus, the postmodifier
zone can express only some of the meanings available in the premodifier zone.
Furthermore, the expression of permanence and temporariness of properties is
investigated, as particular positions are commonly associated with temporary and
permanent senses. The conclusion is that there are grades of permanence in both
modification zones and that the adjectives‟ positions in these zones have specific
correlations with these grades. On the other hand, the predicative construction is neutral
to this issue, as the evocation of permanence or temporariness in this position needs
further co-textual specification.

v
CHAPTER 1

INTRODUCING THE CLASS OF ADJECTIVES

This thesis is concerned with the part of speech known as adjectives and more
specifically with adjectival constructions. In particular, I am concerned with adjectival
constructions with nouns and I intend to focus on two parameters in my discussion,
namely the position adjectives occupy in these constructions and the meanings that may
arise in these specific positions. The latter does not mean that I will be preoccupied with
the description of individual meanings of specific adjectives. Instead, what I am
interested in is to investigate whether there are recurring patterns of meaning that could
arise in specific positions. This can, in turn, be very helpful in a principled description
of the syntax and semantics of adjectives.
Before I proceed to the exposition and analysis of adjectival constructions, it is
desirable to attempt a short description of this part of speech as well as some properties
associated with it. Properties of adjectives can be found in levels of linguistic analysis
such as phonology, morphology, syntax and semantics.
A morphological feature of English adjectives is their lack of the morphosyntactic
feature of agreement. In some languages, adjectives agree with nouns in the
morphosyntactic features borne by them such as grammatical gender, case and number.
For instance, Modern Greek is a language with adjective-noun agreement and thus in
the noun phrase oreos keros („nice weather‟) the adjective oreos agrees with the noun
keros. The agreement here is marked morphologically by the suffix -os, which
encompasses the morphosyntactic features of masculine gender, nominative case and
singular number, and is adopted by the adjective in the environment of a noun bearing
these features. However, this is not true for English, as adjectives do not even receive a
plurality suffix and thus *reds apples is ungrammatical. In general, languages can
demonstrate the relationship between adjectives and nouns either by agreement or word
order, while a combination of both is also possible (Teyssier 1968: 225). According to
Teyssier (1968: 226), the English adjective, which apparently lacks the property of
agreement, is ruled by word order.

1
Apart from the lack of agreement, English adjectives demonstrate some other
properties. Typically, English adjectives can express grades, i.e. the comparative and the
superlative. Some adjectives achieve this with the morphological suffixes -er and -est
(bigger, biggest), while others use a periphrasis with more and most (more beautiful,
most beautiful). Also, English adjectives can be modified by adverbs like very (e.g.,
very beautiful) and in the sentence they can occur in predicative position, that is, as part
of the predicate after the verb (e.g., the sky is blue), in prenominal or premodifier
position, also called attributive (e.g., the blue sky), as well as in postnominal or
postmodifier position (e.g., the man present).1
The aforementioned properties, however, are not exhibited by all members of the
adjective class. Some adjectives do not express the comparative and superlative forms
like same (e.g., *samer/*samest), and others occur exclusively in predicative (the man
is afraid/*the afraid kid 2) or in attributive position (the only problem/*the problem is
only). Also, some adjectives occur prenominally (the only solution/*the solution only)
or postnominally (the kid afraid of people), while some express different meanings
depending on the position (the peculiar feature „strange feature‟/the feature peculiar
to… „feature characteristic of…‟).
Another morphological attribute of adjectives is their ability to become nouns, the
process being known as conversion (e.g., the blind, the sick, the rich). However, not all
adjectives can felicitously turn to nouns, for example linguistic papers cannot turn to
the noun the linguistic and still mean linguistic papers; and if the linguistic exists, it can
mean something closer to the linguistic phenomenon (cf. the social, the cognitive).
Taking all the features above into consideration, one may conclude that adjectives
do not form a uniform category, as they are quite variable with respect to their formal
properties. In fact, their sole common formal feature is the lack of agreement. Their
syntactic position is also not a property unique to them. Other categories, lexical and
phrasal, can occur in these positions too, as it is shown in the following examples. The
examples under (1) concern the predicative position, (2a) the prenominal and (2b) the
postnominal position.

1
Radden and Dirven (2007: 149) put prenominal and postnominal adjectives under the term attributive,
which, according to them, is a term used for adjectives as modifiers to nouns. In contrast, Teyssier (1968:
242) uses the term attributive only for prenominal adjectives, while reserving the term appositive for
postnominal adjectives. I will use the term attributive in the fashion of Teyssier, because it seems to me
that we mostly use this term with prenominal adjectives to contrast them with predicative adjectives.
2
This is typical of adjectives with a- (e.g., afraid, aware, etc.). These adjectives are not further studied
here.
2
(1) Terry is in the garden. (PP)
Bob is a dog. (NP)
The question is whether Mary has found a job. (clause)
(2) a. It was an inside joke. (adverb)
„The dog‟ is a noun phrase. (noun)
Don‟t give me an I-don’t-know look. (clause)
The People’s Republic of China. (noun in genitive case)
b. The dog outside is barking. (adverb)
The man in the garden is my uncle. (PP)
Tina is a person I used to trust. (relative clause)
Furthermore, adjectives seem to be a hybrid category in that they oscillate between
nouns and verbs, sharing properties typical of both classes. As a category close to
nouns, adjectives possess these features (Radden & Dirven 2007: 148-149):
(a) They can become nouns (conversion).
(b) In languages like Greek, Latin and French, they demonstrate agreement with
nouns.
(c) They can appear in predicative position, just like nouns (snow is white).
(d) They can express quite stable properties. Just as a dog is likely to remain a dog, a
big black dog is expected to remain big and black.
As a category that resembles verbs, adjectives have these features (Radden & Dirven
2007: 148-149):
(a) In English they can be morphologically identical to verbs (to open/open) as well
as derived from verbs, either as participles (to close/closed, to amaze/amazing)
or new lexemes (removable).
(b) Like verbs, they are intensifiable (very attractive/it attracts me very much), and
gradable (more attractive/it attracts me more), while some can receive
complements (I am excited about getting married).
(c) In semantic terms, they express relations, like verbs, and thus are conceptually
dependent, for example, big in big dog designates a relation between the dog and
the domain of size.
On the basis of their exhibiting the previously mentioned formal properties, we can
posit a distinction, following Radden and Dirven (2007: 150), between prototypical and
non-prototypical adjectives. Prototypical adjectives are gradable, intensifiable, and can
occur in both attributive and predicative positions, while less prototypical adjectives do
3
not exhibit all these features. It is also possible to consider that their hybridity as a
category is reflected in their inconsistency of formal properties.
The interplay between the position and the semantics of adjectives is going to be
the central topic of what follows. Specifically, the issues with which I am going to be
concerned include, but are not limited to, the following matters:
1. The occurrence of the adjectives in predicative and attributive position and the
meanings that arise in such environments.
2. The canonical position of the English adjective, that is, the premodifier zone of
the NP.
3. The prenominal and postnominal position in association with the evocation of
permanent senses in prenominal position and temporary senses in postnominal
position.
The purpose of discussing the issues above is to examine whether there can be found an
account that can relate semantic interpretation and syntactic position in a principled
manner.

4
CHAPTER 2

INTERPRETATION IN ATTRIBUTIVE AND PREDICATIVE


POSITION

2.1 Approaches to the interpretation of adjectival constructions

Adjectives bear a syntactic and semantic relation to nouns either as premodifiers or


postmodifiers to the head noun in the noun phrase (NP), or in the predicative position as
part of the predicate in which they ascribe the property to the subject NP. The
relationship between noun and adjective has been described by Teyssier (1968: 226) as
one of exchange: the noun uses the adjective to express an additional nuance of
meaning, while the adjective uses the noun for grammatical support. In his own words,
“the relation from noun to adjective is one of semantic sufficiency, whereas the relation
from adjective to noun is one of grammatical necessity” (p. 226). Thus, as he further
notices, an adjective with the ability to sufficiently modify or even alter the meaning of
the noun will tend to be grammatically closer to the noun (p. 226). Therefore, the nature
of the adjective-noun semantic relationship has repercussions on the position of
adjectives in relation to the noun. This is a first observation on the importance of
meaning in determining the distance of the adjective from the noun. More on this will
be mentioned in Chapter 3.
Another aspect of the Adjective-Noun relationship is that adjectives are, in
essence, qualifiers of nouns. According to Radden and Dirven (2007), qualification is
employed in order to specify things or instances of things and it can be expressed not
only by adjectives but also by prepositional phrases (PPs), genitive phrases, and relative
clauses (p. 141). As I observed above in (1) and (2), these kinds of qualifiers can occur
in the same positions as adjectives. This perhaps shows that these positions are
generally associated with qualifiers. In particular, adjectives as qualifiers express
“single qualitative features that are related to a thing or an instance of a thing”, which
are called properties (p. 146).
The issue that arises here is how properties are interpreted in the environment of
the noun, that is, how properties relate to the noun referent. An approach to this problem
would be to consider a compositional account for how adjectival constructions are

5
interpreted. Compositionality is the “process of computing meanings of wholes from
meanings of parts… This computation requires knowledge of both meanings of parts
and of rules of their combination (semantics and syntax)” (Kitis 2012: 118). The
addition of the meanings of parts gives a composite term, for instance the meaning of
red house is composed of the things that are a house and things that are red (Taylor
1992: 1). In set-theoretic formulation, the phrase denotes the set of things which is the
intersection of the set of things which are [Adj] and the set which is [N]. However, there
are phrases that are problematic to being treated in set-theoretic terms, for example this
is a fake Picasso does not entail this is a Picasso, and this is a mere child does not entail
this is mere (Taylor 1992: 1). Such phrases pose a problem for compositional
interpretation, as they cannot be semantically explained by a simple „summation‟ of
meanings of the terms they are composed of (Taylor 1992: 2). As it seems then, the
interpretation of adjectival constructions is quite complex.
In an attempt to show the complexity of the matter, Athanasiadou (2006) states
that interpretation is determined by the adjectives‟ “inherent properties, the meaning of
the noun the adjectival properties are assigned to, the manner in which these different
meanings are related, and also the linguistic and pragmatic context” (p. 210). Therefore,
there are other factors that may contribute to the interpretation of these constructions,
such as active zones, the context and background knowledge. Sweetser (1999: 147)
argues that nouns profile entities, while adjectives elaborate a specific active zone of the
entity profiled. For instance, in red ball the noun profiles the entity ball, while the
adjective profiles the color of its surface. However, this is only the simplest
interpretation, as a red ball can also be a type of ball filled with red paint or a ball with
red marks. Sweetser (1999: 147) even provides the interpretation „the ball used by the
team with red uniforms‟ in which the team‟s uniforms constitute an accessible active
zone of the ball. Thus, she argues that aspects of the frames linked to the entity are also
included in the active zone elaborated by the adjective. This also shows the power of
context on the interpretation of the adjective. As Sweetser says, “the same word can
represent very different complex meaning structures in different contexts and may alter
flexibly depending on the meanings surrounding it” (1999: 136). Moreover,
comprehension of adjectives seems to involve complex background knowledge (Radden
& Dirven 2007: 147). This can be demonstrated even in simple cases, for example in
Sylvia is tall (Radden & Dirven‟s example) the adjective is understood along a scale of
height in relation to a norm which is defined by the frame „young girl‟.
6
A distinction that is discussed by Taylor (1992) and introduced by Vendler (1967,
1968) is that between absolute reading and synthetic reading. Absolute reading is
defined as the interpretation derived at in a compositional manner, while synthetic
reading concerns interpretations that are not possible to be arrived at compositionally.
This distinction is well manifested in ambiguous phrases like beautiful dancer. In its
absolute reading, the phrase means „a dancer who is beautiful, as a person‟, while in its
synthetic reading it means „a person who dances beautifully‟. In the first sense, the
adjective is directly linked to the noun through the copula, while in the second sense
there is an indirect connection through the intervening verb (Taylor 1992: 5).
Not only can the same phrase be ambiguous between an absolute and synthetic
reading but also depending on the noun, the same adjective can express different
meanings. Taylor (1992) discusses the senses of the adjective old. An old friend can be
„an aged friend‟ (absolute reading) or „a friend of long standing‟ (synthetic reading).
There are more kinds of synthetic interpretation expressed by old. An old girlfriend
usually means „a former girlfriend‟ and an old regime can be „a regime that no longer
exists‟ (p. 2). Also, there are phrases that can have more than one synthetic reading, like
old colleague which can mean „colleague of long standing‟ or „former colleague‟, apart
from the absolute reading „aged colleague‟. But, according to Taylor, there are phrases
like old box and old car which do not seem to evoke a synthetic reading (p. 2) and also
phrases such an old wife which seem impossible to have synthetic interpretations, even
though they are conceptually possible (as a „person who used to be someone‟s wife‟ or a
„person who has been someone‟s wife for many years‟) (p. 3). The problem is, then,
how we can account for the variability of interpretation.
Taylor (1992) presents an analysis within the framework of Cognitive Grammar
and claims that the different senses of old are specific instantiations of an abstract and
schematic sense OLDo, which denotes the entity‟s existence at some time before
reference time (p. 20). Also, he observes that synthetic readings arise due to the
relational character of nouns colleague, friend, girlfriend (p. 20). This means that the
designation of a noun is important for what an adjective can mean.
Here it is useful to make a distinction between two terms, intension and extension.
The designation of a noun can be said to be its intension, which involves the meaning of
an expression without considering its extra-linguistic reference. The latter is known as
extension (Kitis 2012: 114). Intension, which is equated with the term sense, accounts
for non-synonymous terms which, however, can be co-referential. Thus, the phrases the
7
Morning star and the Evening star have the same reference, the planet Venus, but are
felt to be non-synonymous (ibid., p. 114). According to Carnap, “the concepts of sense
and of intension refer to meaning in a strict sense, as that which is grasped when we
understand an expression without knowing the facts” while reference and extension are
dependent on facts (1956: 125). In the previous phrases with old, the intension
expressed in the nouns was important for the evocation of synthetic readings, while
their extension seemed to be evoked in their absolute readings. Therefore, we can
conclude that there are two correlations, one between absolute reading and extension
and the other between synthetic reading and intension.
The distinction between two types of interpretations and their relation to
extension/intension was also noticed by Bolinger (1967), who introduced the terms
reference-modification and referent-modification. The first concerns the application of
the modifier on the „reference‟ system of the noun (cf. intension), while the second
involves modification as applied to the referent (cf. extension)3. For instance, the
ambiguous phrase criminal lawyer can mean „a practitioner of criminal law‟, which is a
field of law (reference-modification, synthetic reading, intensional meaning of the noun)
or „criminal practitioner of law‟ (referent-modification, absolute reading, extensional
meaning of the noun). As it is evident, in referent-modification we do not need the
designation of the referent as a lawyer to describe it as criminal, but this information is
crucial for reference-modification. According to Bolinger, these two interpretations
arise due to the double nature of noun as namers of classes and designators of
individuals; in the first they take reference-modification, in the second they take
referent-modification (p. 23).

2.2 Predicative and attributive position

Till now it seems that the semantic notions of intension and extension play a role in how
adjectives are interpreted. But I need to introduce the factor of position, namely the
attributive and the predicative position.
At first glance, it seems that predicative and attributive adjectives are just optional
alternatives, for instance compare the house is nice/the nice house, the sky is blue/the

3
This is a terminological difference. Bolinger uses the term reference to talk about intension and the term
referent to talk about extension.
8
blue sky. During the early days of Transformational Grammar, it had been proposed that
attributive adjectives are transformationally derived from predicative adjectives
(Chomsky 1957: 72; Bolinger 1967: 1-2). Bolinger (1967) showed, however, that this is
not possible for a number of reasons, one of them being that many attributive adjectives
cannot be placed in predicative position, e.g. the main reason/*the reason is main, a
fond old man/*the old man is fond, a total stranger/*the stranger is total, a medical
man/*the man is medical (p. 2). The problems he discusses show that we cannot
account for predicative and attributive positions as transformationally related.
There is further support for this observation and this concerns the distinction
between intension/extension and synthetic/absolute. Taylor (1992) notices that when old
occurs in predicative position, only the absolute reading is available (p. 2). The
following sentences are not possible with old in a „synthetic‟ sense:
(3) *The girlfriend is old. („former‟)
(4) *The regime is old. („non-existent‟)
(5) *The friend is old. („of long standing‟)
In the sentences above we notice that they could be acceptable if the adjective meant
„aged‟, i.e. an „absolute‟ sense. If we also add the distinction between
intension/extension to the picture, we notice that predicative adjectives evoke meanings
that are extensional or referent-oriented, while attributive adjectives evoke not only
extensional but also intensional meanings (Taylor 1992: 7). For instance, old friend can
have both meanings, either as „aged friend‟ (absolute reading, extensional meaning) or
„friend of long standing‟ (synthetic reading, intensional meaning). On the other hand,
my friend is old can only have an extensional meaning, the others being impossible to
be evoked. Therefore, the syntactic position (predicative/attributive) seems to be a
factor for semantic interpretation (absolute/synthetic, extension/intension). Taylor (ibid.,
p. 31) concludes that unprofiled relations that are salient in the semantic structure may
arise inside the NP and, thus, more intrinsic properties related to the noun‟s designation
can be expressed, while predicative adjectives can express only extrinsic properties on
the profile of the noun. In short, an attributive adjective can bring a broader range of
interpretations for an NP than a predicative one. This distinction between attributive and
predicative adjectives is reflected in Radden and Dirven (2007: 150), who place
attributive adjectives under qualification and predicative adjectives under assignment.
Therefore, attributive adjectives broadly qualify the noun, whereas predicative
adjectives assign properties.
9
Using the account given by Taylor, we can understand that previous examples,
which presumably cannot evoke synthetic readings, like old box or old wife, can indeed
be interpreted in such a manner. In principle, old box can mean „former box‟ in a
situation where something needs to be kept inside a box and the present box perhaps has
become unsuitable to contain it, thus needing a new one. The sense „box of long
standing‟ is admittedly peculiar but, in principle, it could be interpreted as such. The
problem is that we do not usually find appropriate situations in which this kind of
meanings could be realized. Therefore, we do not usually relate ourselves to boxes as to
provide this meaning to them. This is also the case with old wife which can mean
„former wife‟. Such a concept would be usable in a society where more than one
marriage is acceptable, but in another society forbidding it, it would not, although this
kind of interpretation could still be evoked. Consider also the phrase old father. It can
mean „former father‟ in principle, but this is not a commonly usable sense, because in
our society we do not change fathers. The concept father is normally related to
consanguinity, that is, it is connected with a blood relation. Consanguinity, however, is
eliminated if one is a father of an adopted child. Its elimination is expressed in
stepfather or foster father and even in the figurative instance of he was a second father
to me (example given in Taylor, 1992: 4). This is consonant with a theory of concepts
with fuzzy borders and prototype structure, as family resemblance categories and not as
governed by necessary and sufficient conditions (Taylor 1995). Therefore, more
possible interpretations are available than the ones we usually employ in our everyday
affairs. Moreover, the grammar of the language gives speakers the ability to express
these relations with attributive adjectives, while proper context may facilitate their
activation. This statement is consonant with Taylor‟s conclusion that unprofiled
semantic relations can be expressed in the premodifying string of the NP.
A crucial question that follows is why there is a difference between the attributive
and predicative positions or, in other words, why an attributive adjective can express a
richer set of interpretations than a predicative adjective. Here I think it is useful to
consider the nature of the groupings; the nominal or NP and the predicate. An
attributive adjective occurs inside the NP, where the noun is the head. As such,
premodifying adjectives are part and parcel of the process of nomination (Teyssier
1968: 242). On the other hand, a predicative adjective is a part of the predicate, which
says something about the nominal. As such, it is irrelevant to the process of nomination.
Rather, it imposes an attribute on a previously determined constituent, which is also the
10
subject of the sentence (ibid., pp. 242-243). Because premodifier adjectives participate
in the nomination process as parts of the NP, they have the ability to express more
relations with the noun they modify. In contrast, unprofiled relations are not accessible
for the predicative adjective, as the latter is part of the VP and is related with the head
noun of the subject NP only through the copula verb. Attributive adjectives do not need
the mediation of the copula to express various relations with the noun and, thus, they are
not restricted to assigning properties, unlike predicative adjectives.
From the previous discussion, we also noticed that adjectival constructions can
have various synthetic readings. However, the term „synthetic reading‟ is too broad and
encompasses many different interpretations of adjectives, as we observed. It would be
better if we could find a way to systematize these synthetic interpretations, which, as we
noticed, involve the intension of the noun. Furthermore, we noticed that synthetic
readings tend to go hand-in-hand more easily with attributive adjectives. But it seems
that the synthetic reading in beautiful dancer can perhaps be expressed with a
predicative adjective in the dancer is beautiful. These facts show that the distinction
synthetic/absolute and its correlation with intension/extension and
attributive/predicative are in need for a finer treatment.

11
CHAPTER 3

INTERPLAY OF MEANING AND POSITION IN THE


PREMODIFIER ZONE

The previous discussion shows that distinctions like absolute/synthetic reading and
referent-/reference-modification are perhaps too broad to discuss the problems that
emerge in describing and explaining the interaction between syntactic position and
semantic interpretation of adjectives. I will now turn to approaches that engage with
categorizing the adjectives according to some criteria. Perhaps, these can be a key to
resolving this puzzle.

3.1 Types of properties

As I previously mentioned, adjectival modifiers are a type of qualification. There are


two types of qualification, restrictive and non-restrictive qualification (Radden &
Dirven 2007: 142-143). Restrictive qualification is applied to things or instances of
things and restricts their referential range, as in the following example:
(6) Simone de Beauvoir was an influential French feminist writer.
The items in italics restrict the reference of the noun writer. Such qualifications create
subcategories for which we may not have specific names. Radden and Dirven (2007)
make an important observation about the lack of (sub)category-specific names: “In fact,
we constantly lack simple words to express the infinite number of subcategories that we
can think of and want to talk about. For example, there are no single words available in
English to express subcategories such as „law-abiding citizen‟, „big surprise‟ or „the
man in the street‟” (p. 142).
On the other hand, non-restrictive qualification offers non-categorizing, additional
information about a thing or an instance of a thing (ibid., p. 143). Essentially, it
provides evaluative information. Such properties are also called epithetical. Consider
the authors‟ example:
(7) That cute little spaniel with its frizzy ears is still watching us.

12
The referent of the spaniel and its ears are easily identifiable within the deictic situation.
Therefore, the modifiers cute, little and frizzy are not necessary for identification but
provide additional, evaluative information.
Restrictive and non-restrictive adjectives can be shown to belong to a continuum
from fully entrenched categories to epithetical, non-categorizing properties (Radden &
Dirven 2007: 146-147). The following Table by Radden and Dirven (ibid., p. 146)
shows the range from category to non-restrictive property:

Entrenched Subtype Categorizing Epithetical


category of category property property
Simple noun Compound Restrictive adj. Non-restrictive adj.
story short story long story stupid story

Table 1. The continuum from category to non-restrictive properties.

As Table 1 shows, an entrenched category is expressed by a simple noun. Category


subtypes are expressed by lexicalized phrases called compounds like short story (ibid.,
p. 146). Compounds result from a process known as lexicalization (Adamson 2000: 60).
Semantically, this compound expresses a particular type of a piece of literature, that is, a
subtype of a category, just like love story. Because short does not denote any specific
length, it is possible to say a long short story, or even a short short story. Also, the
compound does not exhibit the formal properties associated with prototypical adjectives
(Radden & Dirven 2007: 146). Phonologically, the phrase has a unitary stress pattern,
while morphologically the adjective cannot form comparatives and superlatives
(*shorter story, *shortest story). Syntactically, the compound cannot receive intensifiers
(*very short story) and the adjective cannot be separated from the noun by another
adjective (*short interesting story). I should also add that the compound cannot be
separated by placing the adjective in predicative position (*the story is short). If the
phrase is not a compound, the expected morphosyntactic behavior applies well. For
instance, long story behaves as expected (e.g., longer story, very long story, a long
science-fiction story, the story is long). Semantically, it does not express a particular
subtype of a thing but categorizes story as being long in comparison to other stories
(ibid., p. 146). On the contrary, stupid story expresses a non-restrictive, non-
categorizing, additional property, meant as an evaluative comment (Radden & Dirven
2007: 146).

13
The discussion above shows that the manner in which an adjective is used, as
subcategorizing a category or a property of a category, has consequences on its formal
behavior (i.e., loss of morphosyntactic properties). Moreover, this is not a property of
specific adjectives, as short can be used either in the specific compound (short story) or
as a free adjective (short man). We also notice that in a phrase like a short short story
only the first short implies a description about length while the second constitutes part
of the compound. These meanings cannot be reversed, with the second short implying
length. Therefore, the meanings are derived by the specific arrangement of the
adjectives. This is a hint that perhaps that there are specific syntactic positions in the NP
with an associated semantic interpretation.

3.2 Semantic types and the sequence of adjectives

In order to investigate the existence of such positions we can consider the way
adjectives are placed in a sequence inside an NP. In general, we know that more than
one adjective can occur with a head noun (e.g., a nice red dress, a hot tasty soup). We
should, then, consider in which order the adjectives are placed.
It is generally known that adjectives tend to occur in a specific order. Observe the
following sequences (from Adamson 2000: 43):
(8) a. horrible jealous old man
b. ?jealous old horrible man
(9) a. nice large soft green cushions
b. ?green soft nice large cushions
Examples in (a) appear more acceptable than those in (b), but the latter are not
ungrammatical or totally unacceptable. This leads us to the conclusion that there are
preferred and non-preferred adjective orders. (Adamson 2000: 43; Dixon 1982: 24-25).
But what makes an order preferred? Is the meaning of the adjective related to this?
Discussing the order of adjectives in the NP, Dixon (1982: 24-25) proposed the
following general sequence of the premodifying elements from the left to the right of
the NP: Pre-adjectival modifiers – Adjectives – Post-adjectival modifiers. The three
categories include the following elements:

14
a) Pre-adjectival modifiers: logical qualifiers (all, some etc.), determiners (the, this),
possessives (my, John’s), superlatives (best, cleverest), ordinal numbers (fourth),
cardinal numbers (four).
b) Adjectives and
c) Post-adjectival modifiers: origin/composition (oatmeal dog food),
purpose/beneficiary (oatmeal dog food).
For instance, four large leather collars.
Within the categories a, b and c, items are not freely ordered. For instance, the
pre-adjectival order all four is more acceptable that four all, and post-adjectival leather
dog collars is more acceptable than dog leather collars (Adamson 2000: 43). Regarding
the order of adjectives, Dixon (1982) categorized adjectives into seven types (pp. 15-
16). The order of the adjectives follows the order of their semantic classes, which is 1 to
7 from left to right thus (p. 24):
1. Value: good, bad, proper, perfect, excellent, fine, delicious, atrocious, poor, etc.
2. Dimension: big, large, little, small; long, short; wide, narrow, etc.
3. Physical property: hard, soft; heavy, light; rough, smooth; hot, cold, etc.
4. Speed: fast, quick, slow, etc.
5. Human propensity: jealous, happy, kind, clever, generous, gay, cruel, rude, etc.
6. Age: new, young, old
7. Color: black, white, red, etc.
It is important to note that with adjectives of the same type there is no basic underlying
order. Therefore, order exists only between semantic types, not within a type.
The order above concerns unmarked cases, but Dixon says that there are
deviations in marked circumstances. Consider again the order of the aforementioned
examples in (8) and (9) now repeated as (10) and (11) along with the semantic classes of
the adjectives (as given in Adamson 2000: 43):
(10) a. horrible jealous old man [1-5-6: value, human propensity, age]
b. ?jealous old horrible man [5-6-1]
(11) a. nice large soft green cushions [1-2-3-7: value, dimension, physical
property, color]
b. ?green soft nice large cushions [7-3-1-2]
A phonological factor that has been discussed as playing a role in distinguishing
unmarked from marked sequences is stress. According to Whorf (1956: 93), the normal
stress pattern is to have stress on the noun (steep rocky ˈhill) or on the more „inherent‟
15
adjective (pretty ˈFrench girl). Reversing the order of adjectives means that the stress
should move as well, thus ˈFrench pretty girl which would contrast with ˈSpanish pretty
girl. Likewise, Dixon (1982: 24) says that marked stress patterns with stress on the first
adjective are compatible with adjectives in either order, for example, ˈnew white house
(vis-à-vis ˈold white house), ˈwhite new house (vis-à-vis ˈred new house), but with
normal stress pattern the preferred left-to-right order is followed. Adamson (2000: 43)
also notices that non-preferred orders have marked intonation patterns, such as comma
disjuncture between the adjectives or contrastive stress.
According to Quirk et al. (1972: 924), adjectival modification is essentially a type
of restriction of the scope of the class of referents denoted by the noun. Consider the
phrase a tall thin young black-haired Scottish woman. As Dixon writes, adjectives
“directly qualify the head noun” (1982: 25). This is in fact true, as all adjectives provide
direct qualification to the noun woman. There is no adjective that restricts reference for
the next one, but instead they all apply to the noun and restrict its own referential scope.
Thus, Dixon (ibid., p. 25) correctly observes that the phrases a clever brave man/a
brave clever man have similar cognitive meaning. Note, however, that both adjectives
clever and brave belong to the same semantic type, namely human propensity, which
means that, according to Dixon, their order is not fixed.
The discussion above means that adjectives are not recursive, in the way meant by
Quirk et al. (1972: 924), who suggest that “in part the preferences seem to correspond to
the „natural‟ order of recursive qualification”. Adamson (2000: 43) correctly notices that
the word natural above “avoids, rather than provides, an explanation for the particular
constraints on ordering”. According to her, Dixon‟s order of semantic types represents a
scale of inherence, in which some properties are more inherent than others.
As Adamson claims (2000: 43), there is no notion of recursive qualification in
Dixon‟s account. However, Dixon states that there are two exceptions. Pre-adjectival
modifiers, such as all, two etc. qualify all items appearing after them in the NP, and
hence adjectives as well (Dixon 1982: 25). Consider Dixon‟s phrases the cleverest two
men and the two cleverest men. The first selects the cleverest pair of men among other
pairs of men, while the other selects two men who are the cleverest among various other
clever men. The second exception concerns the adjectival semantic class „value‟.
Although each adjective directly qualifies the noun, an adjective denoting a value
property cannot directly qualify the noun but instead a non-value property of the noun
referent (ibid., pp. 25-26). In a phrase like a good fast new car, the car is good as
16
regards its speed, while in a good new fast car the car is good with regard to being new.
Of course, value adjectives can occur without any other adjective, as in a good car. In
this case, the car is not good with regard to being a car, but instead the adjective
modifies an aspect of the car which has been left unsaid. As Dixon puts it, the value
adjective “effectively qualifies some implicit non-value adjective” (p. 26), the latter
really qualifying the noun. The meaning of good, then, is vague, as it basically means
„good for X‟, where X stands for any relevant property which could be evaluated as
good. Of course, such vagueness can be resolved in language use through context,
linguistic or situational.

3.3. The subjective-objective split in the NP

Dixon‟s categorization of adjectives into seven types is basically a semantic typology


that purports to explain why adjectives occur the way they do in preferred sequences.
Under this rubric, it is thought that adjectives belong to a specific semantic category.
However, there are adjectives that tend to be polysemous and are used quite flexibly.
This means that they should belong to more than one semantic class (Adamson 2000:
45), for instance sweet belongs to physical property class but can be used as an
evaluative adjective, thus sweet wine/sweet baby (physical property/value), fair hair/fair
jury/fair performance (color/human propensity/value).
According to Adamson (ibid., pp. 45-46), senses are predictable from the head
noun, but whether an adjective, whose range covers both descriptive and affective
meanings, selects one or the other depends on its position in the NP. Consider these
examples (taken from Adamson, p. 46):
(12) a. short rotten planks
b. rotten short planks
(13) a. little yellow devils
b. yellow little devils
It is evident that in both cases, (a) and (b) can have the same meaning. However, they
need not. Adamson claims that phrases in (b) can mean something else too (p. 46). In
(12) rotten can mean „in a state of decay‟, a descriptive sense under the physical
property class, but in (b) it can also have the evaluative meaning „worthless‟; in fact,
under this meaning the planks may not be in a decaying state. In (13) we find the same

17
pattern; yellow in (a) evokes only a descriptive sense under the color category but in (b)
it can evoke the evaluative meaning „coward‟.
These findings support the division of the premodifying string in the NP between
a slot to the left evoking evaluative meanings and a slot to the right evoking descriptive
meanings. The specific slot may determine the way the adjective ought to be
interpreted. But the reverse can also be true; if the meaning of the adjective is more
evaluative it will tend to be placed in the leftmost position compared to descriptive
adjective meanings (Adamson 2000: 50).
According to Athanasiadou (2006), the different positions of an adjective are
associated with the viewpoint of the speaker and demonstrate a difference in construal
(p. 210, 223). Following Langacker‟s work (1990: 7), Athanasiadou (2006: 223) defines
subjectivity and objectivity “as viewing relations between a perceiver and an object of
perception. A viewing situation is characterized as subjective when the perceiving
subject is implicit (offstage) whereas in the objectively construed viewing situation the
observer is salient and the focus of attention (onstage). For Langacker, an entity or an
event is said to be objectively construed to the extent that it goes „onstage‟ as an
explicit, focused object of conception. Subjective construal of an entity entails that it
remains „offstage‟ as an implicit unselfconscious subject of conception”.
The influence of subjectivity on adjectives can be observed by contrasting the
following sentences (ibid., pp. 217-218):
(14) The complete works of Shakespeare.
(15) He is a complete stranger to me.
In both sentences, complete evokes quantity. In (14) it describes a quantity of a set of
things, but in (15) it is quantity of a quality that is talked about. According to
Athanasiadou, the difference lies in terms of objective and subjective construal. (14) is
construed in an objective fashion, as any observer can have access to this aspect of the
object, but (15) implies a subjective assessment of the thing talked about. Although the
first is conceptualizer-independent, the second depends on the conceptualizer‟s
assessment and their ability to make an abstraction from the objective configuration in
space. Further, Athanasiadou observes that complete in (15) cannot be turned into a
predicative adjective as in *the stranger is complete. This shows the effect of the
change from the objective construal to the subjective on the adjective‟s properties.
The previous discussion shows that adjectives are not contained within fixed
semantic categories. There are adjectives that can have both a descriptive and an
18
affective sense and, when used in the latter, they tend to be placed in the leftmost part of
the phrase. This makes Adamson posit a split in Dixon‟s semantic classes with the value
class being affective and speaker-oriented while the rest being essentially descriptive
and referent-oriented (2000: 44). For instance, a nice dress is an evaluation of the dress
by a specific speaker, but a red dress is a description of the dress irrespective of the
speaker. This split is motivated by Adamson‟s criticism to Dixon‟s notion of a value
adjective qualifying an implicit non-value adjective. Adamson argues that value
adjectives are “subjective in the same sense as deictic terms: their referential meaning is
largely dependent on their speaker‟s identity” (p. 45). Consider the phrase good book
uttered by someone who has read a specific book. The only way to know in what aspect
the book is good is to know what the speaker values in reading books.
Such a polarity between the objective and the subjective has also been proposed
by Quirk et al. (1972: 924-925) to account for the order of all premodifying items:
“modifiers relating to properties which are (relatively) inherent in the head of the noun
phrase, visually observable, objectively recognizable or assessible, will tend to be
placed nearer to the head and be preceded by modifiers concerned with what is a matter
of opinion, imposed on the head by the observer, not visually observed and only
subjectively assessible.” This is also corroborated by other studies that show a clear
correlation between „objective‟ adjectives and noun proximity, while distance from the
noun is observed with „subjective‟ adjectives (Hetzron 1978; Halliday 1985; Langacker
1991; McGregor 1997; Bache 2000). Halliday (1985, cited in Athanasiadou 2006: 220)
notices that adjectives closer to the determiner express more ephemeral properties,
temporally restricted, while rightmost placed adjectives express more permanent
properties. This observation is important if we consider the commonly held view that
premodifier adjectives tend to express stable properties only ( also expressed in Radden
& Dirven 2007: 150). As a conclusion, we see that there is evidence for the existence of
a division inside the NP between referent-description and speaker-comment, the former
at the right of the NP and the latter at the left (Adamson 2000: 45).

3.4 A functional treatment of adjectives

Till now we have noticed that the position of an adjective can play a role in the way it
modifies a noun, for instance an adjective closer to the noun can have the ability to

19
subcategorize the noun and, consequently, displays a reduction in formal properties. We
also noticed that there are adjectives used both in an evaluative and in a descriptive
fashion and that this distinction also has a basis on their position and distance from the
noun. The features that we noticed perhaps are not associated with a closed class of
adjectives, but instead several adjectives can be used in different ways. It seems, then,
that we need an account in which adjectives are not placed in fixed categories but one
which can distinguish their different functions depending on the position they occupy.
Such a functional account has been offered by Teyssier (1968), who classified the
premodifying elements into three functions from the left to the right of the NP. These
functions, which are associated with positions, are the identifying function, the
characterizing function and the classifying function (p. 232). The order of premodifiers
is, then, explained by the functions they bear in the NP. In Table 2 notice that each
premodifier class (pre-adjectival, adjective, post-adjectival) is associated with a
canonical function, while they are also associated with a subjectivity cline (Adamson
2000: 56).

Function Identifying Characterizing Classifying


Modifier Pre-adjectival Adjectival Post-adjectival
Subjectivity Subjective --------------------------------------- Objective

Table 2. Correlations between function, modifier and subjectivity from the left to the right of the NP.

The Table is explained by Adamson (2000: 56) as follows. In the leftmost part of the
NP, the identifying function is canonically realized by pre-adjectival modifiers (ordinal
and cardinal numbers, superlatives, quantifiers), which along with determiners identify
the relevant instance of the noun referent. At the rightmost end, the classifying function
is canonically realized by post-adjectival modifiers, such as nouns modifying head
nouns (cheese omelet). This function is associated with maximal objectivity, because it
specifies a particular class of the noun, i.e. it subcategorizes the noun. When a class
becomes conventionalized, we can observe lexicalization as a compound noun
(postman). In middle position between classifiers and identifiers lies the characterizing
function, canonically associated with adjectives. They express some attribute of the
noun referent, deemed to be salient and relevant in the specific situation. According to
Teyssier (1968: 229-230), they are the true instances of qualifiers and form, in principle,
an unlimited class, even though there are some adjectives more prone to
20
characterization. Such adjectives are usually morphologically marked with a suffix (-
some, -ful, -ous, -ic, -able etc.) and can appear with proper names as well (e.g., little
Mary) (ibid., p. 230).
However, there is no one-to-one correlation between category and function, as the
classifying and identifying functions can also be realized by adjectives. Some adjectives
(e.g. fiscal, dental) act mostly as classifiers (dental surgeon, fiscal policy) (Adamson
2000: 57). As such, they are placed closer to the noun (Teyssier 1968: 227). Teyssier
notices the similarity between classifying adjectives and compounds like blackbird,
blackboard, blueprint (p. 228). This is the position-function where adjectives can be
lexicalized with the noun (cf. short story). Teyssier also notes that both identifying and
classifying adjectives apply to class nouns, that is, nouns which allow a degree of
limitation, like child, boy, girl, man, woman and not, for instance, proper names (p.
228). Some adjectives are identifiers, showing some affinity to determiners by
intensifying determination (the same girl) or relating to the time of the utterance (the
former president) or to the speaker‟s epistemic stance (likely candidate) (Adamson
2000: 57). According to Teyssier, these adjectives are devoid of descriptive meaning
and require the use of definite determiners in the establishment of the uniqueness of the
noun (1968: 226-227). The adjectives that have specialized in these functions and occur
exclusively as identifiers or classifiers demonstrate a reduction or loss in syntactic
properties compared to the central characterizing adjectives (Adamson 2000: 57). Such
properties are modification by intensifiers (*a very former, *very monetary) and
placement in predicative position (*the girl is same, *the surgeon is dental).
If we take Radden and Dirven‟s (2007: 150) division between prototypical and
non-prototypical adjectives into consideration along with Teyssier‟s tripartite
distinction, we observe that prototypical adjectives are characterizers, who demonstrate
all formal properties of adjectives, while non-prototypical adjectives are classifiers and
identifiers and lack these properties. Also, prototypical adjectives occupy the middle
position in the premodifying string, while non-prototypical adjectives, i.e. identifiers
and classifiers, are placed in the extreme left and right respectively. Last, prototypical
adjectives are more or less subjective or objective while non-prototypical adjectives
demonstrate either maximal subjectivity as identifiers or maximal objectivity as
classifiers.
As Adamson (2000: 55) points out, the distinction between identifying and
classifying functions has a correlate with the distinction between reference and
21
denotation, as given in Lyons (1977: 208, see also Kitis 2012: 107-109). Denotation is a
relation between lexemes and the world, independently of specific situations, while
reference is a relation that holds between lexical items and a specific object in the
world. As a simplistic example, cat denotes, but the cat refers or, more correctly, it has
the ability to refer. Cat denotes a class of animals, but if I say the cat is on the mat, the
NP refers to a specific object in the world. Therefore, a noun denotes but an NP refers4.
More specifically, a noun denotes a type while an NP refers to a token of a type.
Regarding the distinction between identifying and classifying function, an identifier is
related to reference, because it works together with the determiner in delimiting the
referential scope of the noun, that is, establishing a specific token of the type, as in the
first man to walk on the moon, where first establishes a particular token of the type
“men who walked on the moon”. In contrast, a classifier works together with the noun
to restrict its denotative scope, that is, it establishes a certain type or class, as in civil
law, where civil establishes a type of law.
As we noted before, the same adjective can be multifunctional (Adamson 2000:
57, Teyssier 1968: 232), for example in (very) nasal pronunciation „nasal‟ is a
characterizer, while in (*very) nasal cavity it is a classifier, just as short in short story
„type of literary work‟ and short story „story short in length‟. As a characterizer, it
demonstrates all formal properties but not as a classifier. To illustrate this, consider the
sentence “Here was a young, impulsive, over-curious young woman” (cited in
Adamson 2000: 58), in which young is used twice, in the leftmost position as a
characterizer, at the rightmost as a classifier of a class of women. The multifunctionality
of adjectives can be a source for ambiguity, as in the following sentences (from
Adamson 2000: 57):
(16) My first disastrous marriage.
(17) Bloggs is our only criminal lawyer.
(18) Cat is a common noun.
These sentences seem to be quite clear in the meaning they have but, in actual fact, they
are ambiguous. In all sentences, the adjective in italics is ambiguous between a
classifier and characterizer interpretation. In (16) the adjective might characterize the
relationship of a married couple (characterizer), but it might also denote a class of

4
An NP does not always refer. Compare the tiger is an aggressive animal where the NP the tiger denotes
the class of tigers and not a token of the class. Therefore, it is more correct to say that an NP has the
ability to refer but need not always. I owe this important observation to Michalis Milapides.
22
marriages (as a classifier) in which case the ordinal number could imply that the speaker
also has a second or a third disastrous marriage. In (17) criminal may denote a specific
type of lawyer (classifier) or describe a non-law-abiding lawyer (characterizer). In (18)
common denotes a class of nouns in opposition to the class of proper nouns (classifier),
but it can also describe an often-used noun (characterizer). Thus, we observe that
adjectives evoke ambiguous interpretations due to ambiguity of function.
The distinction between classification and characterization can be demonstrated
phonologically by the use of stress. In characterization both the adjective and the noun
receive a full stress but in classification the classifier adjective receives full stress while
the noun a weak one (Teyssier 1968: 232), for instance English in ˈEnglish ˈman and
ˈEnglishman. The former is a characterization describing the origin of a person, but the
latter denotes a „type‟ of person, with a particular culture, habits etc. In this way, the
adjective is a classifier, which has gone so far as to become lexicalized, that is, part of a
compound.
Moreover, it is important to note the diachronic interplay between the three
functions. Adamson (2000: 59-60) observes that the adjectives mere and criminal
follow different change patterns. Mere used to be a characterizer but evolved into an
exclusive identifier, while criminal developed a classifier sense (criminal lawyer „expert
of criminal law‟) from a characterizer without abandoning its role as a characterizer
(criminal lawyer „criminal person, who is a lawyer‟). Therefore, the more rightwards an
item moves in the NP, the more de-subjectivized it becomes. The closer to the noun it
is, the more lexicalized it becomes (e.g., blackbird < black bird). On the contrary, the
more leftwards it moves, the greater the subjectivity it displays. The closer to the
determiner it moves, the more grammaticalized it can be said to become (p. 60).
The functional treatment of the adjectives seems to solve problems that previously
discussed approaches could not. It captures the fact that several adjectives demonstrate
variability as modifiers of nouns and relates positions in the NP with different functions
performed in the nomination process. Thus, it provides a systematic explanation for the
unprofiled relations that can arise in the NP, as Taylor (1992) observed. It also captures
the previously mentioned distinctions between absolute and synthetic reading (Vendler
1967, 1968), as well as referent- and reference-modification (Bolinger 1967). As
Adamson (2000: 63) notices, the classifying and characterizing functions can be equated
with reference- and referent-modification respectively, but classifiers and identifiers are
subsumed under reference-modification. This is why these distinctions were too broad
23
and thus inadequate to explain the full complexity of the phenomenon, even though they
captured the basic problem of variability in the description of adjectives.

3.5 Elucidating the functional treatment

Teyssier‟s account (1968) and its elaboration by Adamson (2000) captures very well the
functional distinctions of the heterogeneous class of adjectives and relates them well
with formal properties. However, it falls short of explaining some other facts, for
instance the ambiguity of adjectives in phrases like old friend, or beautiful dancer. Old
friend, as we have seen, can mean „friend of long standing‟ or „aged friend‟ and
beautiful dancer is ambiguous between „a beautiful person, who dances‟ and „a person
who dances beautifully‟. These interpretations do not seem to arise from different
functions, as in both cases the adjectives function as characterizers. Therefore, the
characterizing function perhaps contains some distinctions that would be desirable to be
further elaborated.
A more elucidating account of adjectives is offered by Radden and Dirven (2007:
150-154) who distinguish these types of premodifying adjectives: scalar, denominal,
deadverbial, and determining.

3.5.1 Scalar adjectives

Scalar adjectives are the prototypical adjectives, as they exhibit all features associated
with adjectives (ibid., p. 150). They express properties that can be placed along the
scales of comparison and intensity (ibid., p. 151). Comparison involves the positive,
comparative and superlative grades (ibid., p. 151). Shorter adjectives express the latter
two morphologically (bigger, biggest), while longer adjectives use more and most (more
beautiful, most beautiful) and less and least (less beautiful, least beautiful), in a similar
fashion to noun quantification (ibid., p. 151). Intensity is expressed by adverbs like
pretty, very, extremely, terribly, and awfully (ibid., p. 151).
The notions of gradability and intensity bring in a semantic type of adjectives,
called complementary adjectives. These express complementary properties in pairs such
as alive/dead, male/female, full/empty. They complete each other and probably belong
to the same conceptual domain, the only difference lying in a single feature (Kitis 2012:
24
214). Also, the affirmation of the one simultaneously negates the other, for instance
alive equals not dead. They are deemed as not freely gradable or intensifiable. For
example, Radden and Dirven (2007: 151) hold that phrases such as ?a fuller glass, *the
emptiest bottle or *a very dead man are seldom or never spoken, because these
adjectives denote extreme or „absolute‟ properties. Kitis (2012: 215) also holds that they
either have the property or not and consequently they are non-gradable, (*he’s deader,
*aliver) and non-intensifiable by very, a little etc. But she concedes that there are cases
which are not „either-or‟, for example half-open window, half-closed eyes. This shows
that their „scalarity‟ exists in terms of an implicit scale. The use of emphasizers, like
absolutely/total, is also possible (e.g., he’s absolutely dead, no doubt about it, an
absolute/total resounding success) (ibid., p. 216). Radden and Dirven (2007: 151) also
concede that degrees can be expressed, as in half (full), almost (round), completely
(dead). Furthermore, there are examples such as the emptiest arms in the world, very
dead matter, more male brain, very much alive which undermine the strong contention
of an „either-or‟ property and show that these adjectives can be used in a scalar fashion.
As Kitis says on this matter, “the magic of language can enjoin otherwise under certain
circumstances. We must never forget that language is just a tool to serve our purposes,
and, indeed, it is a very adaptable one” (2012: 216n). I think the previous examples can
demonstrate that scalarity is an aspect manifested even by adjectives not assumed to
involve a scale when studied on their own. Therefore, the term scalar adjective
designates more of a function displayed by adjectival structures rather than a strict
semantic class, in accordance with the functional treatment I have discussed so far.

3.5.2 Deadverbial adjectives

Deadverbial adjectives describe the manner of an action or the setting of a situation


(Radden and Dirven 2007: 152), for example hard worker „someone works hard‟, early
riser „someone rises early‟, heavy smoker „someone smokes heavy‟. Observe that these
„deadverbial‟ meanings are not possible in the predicative position, in which the
adjectives would evoke different meanings, for instance the smoker is heavy would
evoke only a scalar interpretation of the referent of smoker, that is, the referent‟s weight
(i.e., extensional meaning). Aspects of the setting include the time of the situation (the
late Marilyn Monroe, the former USSR) and reality status (possible effects, the likely

25
winner). Here I notice that the latter adjectives can be placed in predicative position,
although they may need a complement (the effects are possible to observe), while this is
impossible with the former, as in *the USSR is former and *Marilyn Monroe is late
does not refer to her death.
The fact that we are concerned with adjective uses and not strict categories is also
demonstrated by the fact that the same adjective can be used either as scalar or
deadverbial, for example, an elegant girl/an elegant dancer (examples given in Radden
& Dirven 2007: 152). The first phrase displays a scalar use, while the second relates to
the manner of dancing, meaning „a person who dances elegantly‟ (p. 152). This is also
seen in my old colleague (also authors‟ example) which has two meanings: i) scalar „an
aged colleague of mine‟, ii) deadverbial „someone who used to be a colleague of mine
in the past‟. In the predicative position only the first meaning would be possible. This
means that deadverbial uses of adjectives, with the exception of those denoting reality
status, are generally not acceptable in this position.
We now understand why beautiful dancer is ambiguous between two readings.
The adjective can be related to the noun in two manners. In its absolute reading the
adjective is used in a scalar fashion, „a dancer who is beautiful, as a person‟, while in its
synthetic reading it is used in a deadverbial fashion, „a person who dances beautifully‟.
The distinction between scalar and deadverbial adjectives is the one that was subsumed
under Teyssier‟s (1968) characterizing function.

3.5.3 Denominal adjectives

Denominal adjectives are categorizers, for example legal advice, financial advice,
medical advice, and as such they are similar to noun-noun compounds like health
advice, career advice. These attributes have also been previously discussed (Teyssier‟s
(1968) classifiers) but a crucial feature Radden and Dirven (2007: 152) capture is that
these adjectives relate to participants of events and, thus, are given specific roles. The
same adjective may play different semantic roles when modifying a different noun, for
instance, presidential decision (president as the agent), presidential advisor (president
as the recipient), presidential election (president as the theme), presidential candidate
(president as the goal).

26
3.5.4 Determining adjectives

Determining adjectives specify a thing or ground a referent (Teyssier‟s (1968)


identifiers). They can show that the category expressed by the noun fully applies (e.g., a
true asset, a regular fool, a complete idiot, a perfect/total stranger, a real friend). This
meaning is impossible if the adjective occurs in predicative position. On the other hand,
adjectives such as the only, one, very, first, last and main establish the uniqueness of the
referent. Because they function like determiners, they are also impossible to move to
predicative position.
That an adjective can indeed demonstrate different uses can also be shown by the
meanings of the adjective old (Taylor 1992; Athanasiadou 2006: 214). A very old friend
in the sense „an aged friend‟ demonstrates a scalar use of old, in the sense „friend of
long-standing‟ it has a deadverbial use5, while in old girlfriend in the sense „former‟ it
demonstrates a determining use (Athanasiadou 2006: 214). Because adjectives
demonstrate great variability in use, Athanasiadou (ibid., p. 215) claims that adjectives
cannot be rigidly categorized to a particular type.

3.5.5 Iconicity and position

According to Radden and Dirven (2007: 154), the position of premodifying adjectives is
determined by the iconic principle of proximity, which they formulate as follows: “The
more essentially a property functions in further specifying a thing or an instance, the
closer the adjective is placed to its head noun.”
The importance of iconicity as a factor explaining word order had been previously
observed by Whorf (1956: 93), who wrote: “A group referring to „inherent‟ qualities -
including color, material, physical state (solid, liquid, porous, hard, etc.), provenience,
breed, nationality, function, use - has the reactance of being placed nearer the noun than
the other group, which we may call one of noninherent qualities, though it is rather the
residuum outside the first group - including adjectives of size, shape, position,
evaluation (ethical, esthetic, or economic). These come before the inherent group, e.g.
„large red house‟ (not „red large house‟), „steep rocky hill, nice smooth floor‟”. The
passage makes evident that the more inherent to the noun a property is, the closer it is
5
Athanasiadou (2006: 214) uses the terms event-like for deadverbial, thing-like for denominal, and
determiner-like for determining.
27
placed. Table 3 shows the order of adjectives according to the adjective types posited by
Radden and Dirven (p. 154).
According to Table 3, the adjectives closest to the noun are the denominal due to
their subcategorizing function. Scalar and deadverbial adjectives together belong to a
class of characterizers and thus are not as essential as the denominal. If found together,
the scalar is closer to the noun than the deadverbial, as an intelligent reliable expert
seems preferable to a reliable intelligent expert (Radden & Dirven 2007: 1546,
Athanasiadou 2006: 224-225). Last are the determining adjectives, which ground the
referent and are placed next to the determiner.

Functions Grounding Characterizing Categorizing


Adjective determining scalar/deadverbial denominal
type
(the) only reliable (scalar) economic expert
(the) first intelligent diplomatic solution
(deadverbial)

Table 3. The order of adjectives according to the iconic principle of proximity (adapted from Radden &
Dirven 2007: 154).

Radden and Dirven (ibid., p. 154) also claim that the more stable and relevant a property
is, the closer it is to the noun, while evaluative meanings are variable and tend to be
placed further from the noun, for example in stupid white man „white‟ is a more stable
property that „stupid‟. This statement supports the previous claims in 3.3 that adjectives
with evaluative meanings are placed at a distance from the noun.

3.5.6 Subjectivity and property stability

Employing the adjective categories by Radden and Dirven (2007), Athanasiadou (2006)
shows that expression of property stability by adjectives is related to subjectivity.
Specifically, she demonstrates that stability and subjectivity are inversely related, with

6
There seems to be a confusion, as the authors initially categorize reliable and intelligent as scalar and
deadverbial respectively, about which they write, “If scalar and deadverbial adjectives co-occur, they also
have a preferred order: an intelligent reliable expert sounds more natural than a reliable intelligent
expert”. But afterwards they say “… the deadverbial property „reliability‟ is more essential to expertise
than the scalar property „intelligence‟”, which contradicts what they have said and perhaps is a mistake.
See also Athanasiadou (2006: 225), who is based on Radden and Dirven‟s classification and seems to
clarify the picture by saying that event-like (deadverbial) uses precede scalar uses.
28
denominal and determining adjectives at the polar extremes: the former possess greater
stability and less subjectivity, while the latter show greater subjectivity and less stability
(pp. 224-227). In denominal adjectives, which express the more stable properties, the
speaker is placed externally to the viewing of the entity, the effect being its objective
construal (e.g., presidential candidate, linguistic papers), while in the other polarity, the
determining adjectives, which express the less stable properties, the experiencer is
situated at a privileged vantage point (e.g., the only solution to the problem). From the
left part of the NP to the right, Athanasiadou presents how the conceptualizer starts
from being external to the maximal scope of the predication and step-by-step moving
towards the inner part of the onstage region (pp. 226-227).
In contrast to premodifier position, predicative adjectives make no claim about the
stability of a property expressed by them (ibid., p. 232). Without further explication
they can designate either a temporary or a permanent property. Consider the adjective
navigable in the following sentences (adapted from Athanasiadou 2006: 232):
(19) The river is navigable. [ambiguous]
(20) The river is navigable all year round. [permanent]
(21) The river is navigable only in winter. [temporary]

The ambiguity of the position is also seen with past participles. According to Bolinger
(1967: 3), the jewels are stolen is ambiguous between action and characteristic, but
stolen jewels expresses only characteristic while jewels stolen the action. Concerning
the subjectivity of the position, Athanasiadou argues that the conceptualizer construes
their presence in the situation “as a non-experienced reporter, external to the onstage
region” of the conceptualized entity (ibid., p. 233). The speaker is regarded as a
presenter/reporter of information with no direct involvement with the entity. This
explains why we find only scalar and perhaps deadverbial adjectives in this position but
not the other two extreme categories.

3.6 Comments on the functional approach

Investigating the order of the adjectives in the NP had the advantage of showing that
there are specific positions in the NP associated with certain functions and properties.
The functional approach showed that there is an association between the position of the
adjective in the NP and the way it modifies the head noun. We noticed that proximity to
29
the noun has an effect on the adjective‟s prototypical formal properties by reducing
them as well as its ability to appear in predicative position. Maximal distance from the
noun also leads to reduction of these properties. The middle ground is the place where
adjectives demonstrate their usual properties, and this concerns scalar and, perhaps,
deadverbial uses. As it is evident from the previous discussion, adjectives cannot belong
to fixed and closed semantic types, even though there are adjectives that are more easily
associated with a particular functional category, for instance only (determining), fiscal
(denominal), virtuous (scalar). Moreover, the order of the adjective uses is related to the
cline of property stability, with the more stable one to the right, which is inversely
related to the cline of subjectivity, with the more subjective one to the left. The
adjective order, which is accompanied by these characteristics, seems to be explained by
a scale of inherence. We should remember, however, that this is the „normal‟ order,
since there are also marked orders which are accompanied by marked phonological
patterns.

30
CHAPTER 4

PRENOMINAL AND POSTNOMINAL POSITION

4.1 Adjectives before and after the noun

Up to this point we have investigated how meaning and position of adjectives are
related with a focus on the premodifier zone as well as its relation to the predicative
position. Another position where adjectives occur is the postnominal position, which is
usually discussed in relation to the premodifier or prenominal position. The main view
in the bibliography is that these two positions are associated with specific meanings
(Bolinger 1967, Teyssier 1968, Dirven 1999, Athanasiadou 2006, Radden & Dirven
2007). When a modifier occurs in prenominal position, it tends to express permanent
qualities as well as qualities characteristic of the noun referent, whereas if it occurs
postnominally, it evokes temporary and occasional qualities (Radden & Dirven 2007:
144).
According to Teyssier (1968: 242-244), postnominal position is a hybrid, both an
explicit form of attributive and an implicit form of predicative.7 The adjective placed in
postnominal position as a form of explicit attributive is employed to emphasize the
prenominal adjective. Classifying adjectives can appear in this position with unaltered
normal stress pattern. The noun is weakly stressed compared to the strong stress of the

adjective, for instance, Teyssier‟s examples things ˈFrench, things ˈEnglish, matters

philoˈsophical. These phrases phonologically resemble fixed expressions (the princess

ˈroyal) that are a direct inheritance from French denominations and have been preserved
in English as a form of attributive group (e.g., attorney general, princess royal, court
martial, poet laureate, sheriff substitute, heir apparent) (pp. 236-237). Because
emphasis implies a total categorization, only generic nouns are appropriate to the
position, usually in the plural without determiners. As an implicit predicative, the
adjective is non-determining and descriptive, with stress being applied equally to all

7
According to Teyssier‟s terminology (1968: 242-244), attributive adjectives are located in intraposition,
which contrasts with the extraposition of the predicative adjectives. The middle position, i.e. the
postmodifier of the NP, is called juxtaposition by Teyssier and it is the position of the appositive
adjective.
31
items (e.g., a ˈman ˈyoung and ˈhandsome). What seems to be implicit is a predicative
structure, a relative clause as an appositional clause (a man who is/was young and
handsome) (p. 244).
According to Radden and Dirven (2007: 144-145), the typical position of English
adjectives is the prenominal in which they express permanent and characteristic
properties, as in (22a), but if they occur after the noun, the position imposes a sense of
temporariness and occasion, as in (22b). Likewise, a qualifying relation, which typically
occurs in the postnominal position, as in (23b), adopts a permanent and characteristic
attribute if placed before the noun, as in (23a) (pp. 144-145).
(22) a. Taurus and Capella are the only visible stars. [prenominal]
b. Taurus and Capella are the only stars visible (tonight). [postnominal]
(23) a. The limousine comes with a five-line cellular phone system. [prenominal]
b. The limousine comes with cellular phone with fives lines. [postnominal]
In (23b), according to Radden and Dirven (ibid., p. 145), the phone is understood either
as a system permanently equipped with five lines or one connected to as many as five
lines. Thus, it is ambiguous between a permanent and a temporary interpretation. On the
other hand, (23a) can only evoke the permanent interpretation which implies a specific
kind of phone. Also consider the pairs:
(24) a. the visible stars/the stars visible
b. drinkable water/water drinkable
c. eatable game/game eatable
d. marketable products/products marketable
According to Radden and Dirven (ibid., p. 154), these pairs show not only a crucial
difference in the feature permanent/temporary but also a difference in relation to being
complete. The premodified NPs seem to be complete in themselves, in no need of
further explication. In contrast, postmodified NPs seem to be unable to stand on their
own and require further specification (adjuncts), for example, water drinkable when
boiled, products marketable in winter.
The difference of meaning induced by position is greatly demonstrated by
adjectives that have different meanings in different positions (Teyssier 1968: 237), for
example a proper behavior („decent behavior‟)/a behavior proper to… („behavior
characteristic of‟), the present king („the king who reigns‟)/the king present („the king
who happened to be there, or in the presence of the king‟), a peculiar habit („a strange

32
habit‟)/a habit peculiar to dogs („a habit specific to dogs‟). These examples also
demonstrate the need for further specification for the postnominal adjectives.
However, the discussion in Chapter 3 has shown that premodifier adjectives
display variability regarding how stable a property is. In particular, there are prenominal
adjectives with ephemeral properties, namely the determining uses. Permanent
properties can be expressed, normally by rightmost adjectives like denominal ones. At
this point I think it would be desirable to take an example and examine the extent of the
association prenominal-permanent and postnominal-temporary. Consider the pairs in
(25). If prenominal adjectives are associated with permanent qualities, then it should be,
in principle, impossible to occur with an item that evokes a sense of temporariness. This
is examined in pair (26). Also, the same semantic clash should be expected if a
postnominal adjective co-occurs with an item evoking a permanent situation in pair
(27).
(25) a. Arcturus and Capella are visible stars.
b. Arcturus and Capella are stars visible.
(26) a. Arcturus and Capella are visible stars tonight.
b. Arcturus and Capella are stars visible tonight.
(27) a. Arcturus and Capella are visible stars all year long.
b. Arcturus and Capella are stars visible all year long.

Without any further contextual and co-textual factors, (25a) seems to be a complete
sentence expressing a complete thought, namely the visibility of the stars Arcturus and
Capella. This thought is also expressed in (25b) but with a sense of incompleteness
which is provoked by the postnominal placement of the adjective. In short, it seems as if
something is missing after the adjective. In (26) a time adjunct specifies the temporal
aspect of the stars‟ visibility specifically for a short time period. Such brevity of the
stars‟ visibility would render the sentence in (26a) ungrammatical or unacceptable, as
prenominal adjectives expressing permanent properties should be incompatible with a
short time specification adjunct. As we observe, the sentence is grammatical and
semantically acceptable, just like (26b). In (27) the adjunct indicates a larger period of
time, i.e. the visibility of the stars holds all year long and as such it is permanent. This
permanence indicated by the adjunct would be expected to create a semantic clash in
(27b), as postnominal adjectives are said to express a sense of temporariness. However,
it is as grammatical and semantically acceptable as (27a).

33
If each position carried an exact and definite meaning, this should not be the case.
There should be a clash between the prenominal adjective and tonight and between the
postnominal adjective and all year long. What we can conclude from this is that the
difference in meaning between prenominal and postnominal position cannot be located
in the grammaticality of sentences or in their „strict‟ semantics, as the pairs express
almost identical meaning. Almost in the previous sentence is crucial. As I discussed, in
(25a) the sentence appears to be complete, while in (25b), even though it expresses a
full thought, we feel that there should be something following the postnominal
adjective.
Let us consider now the nature of the adjective visible. It belongs to the group of -
able adjectives. This suffix constructs deverbal adjectives with the meaning „it is
possible to V‟ where V stands for the meaning of the base verb. Their verbal origin
means that it is possible to add items to them typically associated with verbs, for
instance a by-phrase expressing the agent:
(28) a. Arcturus and Capella are stars visible by people of the south hemisphere.
?
b. Arcturus and Capella are visible stars by people of the south
hemisphere.
In these sentences the PP specifies the agent, that is, who is able to look at the specific
stars, similar to passive constructions. Thus, it is connected to the adjective, not to the
noun. (28a) seems to be completely acceptable, but (28b), though not strictly
ungrammatical, seems odd. As observed, acceptability is enhanced if the adjective is put
postnominally. This can be explained by the fact that, given the verbal nature of the
adjective, the phrase needs to be placed immediately after the adjective, just as it
follows in the passive construction it is seen by people of the south hemisphere. Let us
now consider an adjunct PP:
(29) a. Arcturus and Capella are stars visible in the winter.
b. ?Arcturus and Capella are visible stars in the winter.
In both sentences, the PP is connected to the adjective, not to the noun. It is concerned
with the verbal nature of the adjective, namely the time of the year when one can see
these stars. As such, it can be said to be part of the structure of visible, like the agent by-
phrase. Perhaps, this explains the total acceptability of (29a) and only partial
acceptability of (29b). Consider now the following pair with an of-phrase:
(30) a. *Arcturus and Capella are stars visible of the winter.
b. Arcturus and Capella are visible stars of the winter.
34
In these sentences, the of-PP does not modify the adjective but the head noun of the NP.
As a modifier of the head, (30a) is totally unacceptable. The only way to make it
acceptable would involve a pause after visible, but even in this way it would again seem
as an addition to the head noun.
The conclusion that perhaps derives from this examination is that a postnominal
adjective opens a position to its right and which can be filled. This can be partially
acceptable if the adjective is placed prenominally. Perhaps what follows the
postnominal adjective can be treated inside an adjective phrase (AP), thus [ S [NP
Arcturus and Capella] [VP are [NP stars [AP visible [PP in the winter]]]]]. What follows the
adjective is, in essence, a contextual specification. Complements and adjuncts further
specify the meaning of the adjective, thus rendering it with a sense of temporariness and
malleability. If temporariness were a property inherent to the postnominal position, then
it should be observed in a simple contrastive set of sentences, as in (25). This means
that the association between temporariness and postnominal position is attributed to the
need for structural completeness by specifying elements. In contrast, no such element is
required structurally in the prenominal position, as this is the canonical position of the
English adjective.

4.2 Participles

A category which also exhibits the behavior found with prenominal/postnominal


adjectives is participles. Participles in English are of two kinds: present participles,
which denote progress in action, and past participles, which denote a completed action
(Teyssier 1968: 237). According to Radden and Dirven (2007: 155), participles
constitute a category between verbs and adjectives. On the one hand, they can function
as predicates receiving complements and adjuncts, but on the other, because they are
atemporal, they can also function as adjectives. Dirven (1999: 61), commenting on the
difference between past participle adjectives and descriptive/qualifying adjectives,
argues that past participles express a two-argument predication relation, while
adjectives express a one-argument predication. A carved table contains the implicit
predication „someone carved a table‟, but round table and nice table do not conceptually
involve a second argument.

35
According to Radden and Dirven (2007: 155), postnominal present participles can
be used only with complements, for instance men drinking whiskey (object), woman
lying on the floor (PP adjunct), people queueing for work (PP adjunct). Otherwise, they
are incomplete and thus the phrase *woman lying is ungrammatical, but in a contrastive
set they can be grammatical (e.g., the man shooting was caught, while the man driving
escaped). In contrast, prenominal participles occur without further specification (e.g., a
lying woman, drinking men, sleeping beauty). In fact, specification would make the
phrase ungrammatical, for example, *a lying on the floor woman. Because contextual
information is not allowed to be used in this position, the authors conclude that phrases
with such participles “do not refer to a specific situation but describe stable attributes
and, like premodifier adjectives, have subcategorising function” (p. 155). Thus, a lying
woman is contrasted with a sitting woman, sleeping beauty with dancing beauty, and
drinking men with non-drinking men.
Past participles typically describe the resultant state of an event and a situation
that is finished and complete (Radden & Dirven 2007: 155). When past participles are
used in prenominal position, they focus on a stable result, reflecting the
conceptualization of more permanent resultant state (ibid., p. 155; Dirven 1999: 61), for
instance a reviewed article, deposited money, reduced costs, broken window - breaking
the window leaves a lasting mark (Dirven 1999: 61). In postnominal position the focus
is on the temporary event which caused the result (Radden & Dirven 2007: 155-156),
for instance article reviewed, money withdrawn, profits gained. Moreover, the same
participle in different positions can show differences in meaning, for example reviewed
article (article accepted for publication, shows stability) but article reviewed (article
which may not have been accepted) (p. 156). Further, lasting states seem to be
unacceptable in postnominal position. For instance, deposited money describes money
in a stable state, but money withdrawn describes money no longer there. Notice that
money deposited is acceptable but not *withdrawn money, as the semantics of
withdrawn does not involve a lasting property (Dirven 1999: 62). In fact, Dirven (ibid.,
p. 62) claims that lasting states are also impossible in predicative position, for instance,
*the money is deposited (describing a lasting state)8. However, lasting states in this
position are possible with compounds as in hand-written letters/the letters are hand-
written and they also cannot receive an agent *letters (that are) hand-written by my

8
It is important here to note that the participle is understood on its own, in a fashion similar to the
window is broken.
36
mother (Dirven 1999: 62). This means that the status of these compounds is closer to a
full adjective.
Furthermore, when a postnominal past participle focuses on the completed event,
it can occur without any complement (the gifts bought, the points mentioned, the issue
brought up) but sometimes they need further specification about the situation (the
conference held last month, a campaign begun in March, a friend met long ago)
(Radden & Dirven 2007: 156). Further specification can only be possible with
postnominal past participles, since according to Dirven (1999: 61) they bear all the
characteristics of a verb predication. Thus, they are capable of being further modified by
an agent or any adjunct, as in a table carved by a Chinese carpenter, a table carved in
the 17th century. In contrast, prenominal past participles cannot explicitly express agents
or patients, even though they contain an implicit role configuration (ibid., p. 62).
If anything, the discussion about participles shows that there is indeed a division
of labor between the prenominal and the postnominal position. Participles in the
postnominal position behave like verbs in that they can receive complements and
adjuncts that provide further information, thus making explicit the predication they
contain. It seems, then, the postnominal position can be a type of position that makes
possible the realization of a full predication. As I argued in 4.1, co-textual information
makes a situation rather specific and thus the predicational character of the position is
associated with a temporary character. Furthermore, a large constituent with the
participle as its head can only occur in postnominal position. Observe an essay written
with orange ink by a former student/*a written with orange ink by a former student
essay. The event expressed by the predication can only be mentioned after the noun.
There is also a feeling that the event is given an importance as a comment about the
essay. That postnominal position pays specific attention to the event shows an affinity
with clauses. Observe that relative clauses can be placed in the same position (e.g., a
man old in years/a man who is old in years, a pyramid built by the Egyptians/a pyramid
which was built by the Egyptians, stars visible with naked eyes/stars that are visible
with naked eyes, children playing hide-and-seek/children who play hide-and-seek).
In the prenominal position, such a detailed comment in an event-like behavior is
not possible, as it is shown by *a written with orange ink by a former student essay. But
if forced to be put in the prenominal position, it will be more like a written-with-
orange-ink-by-a-former-student essay. In this way, the status of the event is erased and
it is given the status of a type of essay. In short, it is treated like a subcategorizing
37
feature, reminding such phrases like an I-don’t-know look. This shows a move from
talking about an event to transforming it to a category of a thing. In Chapter 3
subcategorization of a noun was discussed as Teyssier‟s (1968) classifying function and
Radden and Dirven‟s (2007) denominal adjectives, both of which involved a
premodifier position very close to the noun with the ability to further categorize it
(economic program). Therefore, it may be that, when such large constituents are forced
to the premodifier zone, they are located in this specific position, which is associated
with maximal property stability as well as maximal objectivity. As a consequence, the
position provides them with a „permanent‟ sense.
These comments can perhaps lead to a tentative conclusion that the permanence
typically associated with the prenominal position results from the placement of the
modifier in the position of denominal adjectives. On the other hand, the temporariness
associated with the postnominal placement of a modifier seems to result from the
predicational character of this position.

38
CHAPTER 5

CONCLUSION

The discussion of constructions with adjectives seems to have yielded some significant
facts. In particular, I attempted to discuss adjectives in three constructions, the
premodifier zone, the predicative and the postnominal position, and investigate whether
there are semantic patterns related to them. According to Goldberg (2006: 5), a
construction is any linguistic pattern “as long as some aspect of its form or function is
not strictly predictable from its component parts or from other constructions recognized
to exist”. A key feature of a construction is that it carries meaning “independently of the
words in the sentence” (Goldberg 1995: 1). Such a constructional view of grammar
seems to be based on the view that “a difference in syntactic form always spells a
difference in meaning” (Bolinger 1968: 127). Following this perspective, it seems to me
that this discussion has indeed found some significant relations between positions and
meanings and thus these relations can truly be regarded as constructions.
As I have discussed so far, the premodifier zone of an NP contains an
arrangement of positions that are associated with a specific semantic interpretation. This
is what has been called „function‟ (Teyssier 1968) in Chapter 3 in order to differentiate
this treatment from an approach which addressed the arrangement of adjectives by
classifying them in fixed semantic categories (Dixon 1982). The association between
position and function can be schematically shown in Table 4. The Table shows that the
construction contains five positions between the determiner and the noun. These
positions are given in the second row and they include Radden and Dirven‟s (2007) four
adjective types. I have also added a position to the right of the denominal adjectives in
order to take into account adjectives that seem to be parts of a compound with nouns
and, as such, they appear as even closer to them than denominal adjectives. The third
row shows how the functions proposed by Teyssier relate to these types. Thus, I have
given them the term General Function, while Radden and Dirven‟s types are given the
term Specific Adjective Types. In this way, the Table captures the observation that there
are more general functions associated with more specific adjective types which may
possess specific attributes, even though they have the same general function. It is
interesting to notice that scalar adjectives, which are the prototypical adjectives, occupy
39
the middle position in this construction. The Table also shows how the sequence of
adjectives is determined by the iconic principle of proximity. Furthermore, the attributes
of property stability and subjectivity are presented in Table 5 to show how they relate to
the positions of the construction. It indicates that these attributes are inversely related
and form two continua with two extreme poles. In this way, Table 5 takes into account
the subjective-objective split, with subjective meanings being placed in the far left of
the NP and objective ones close to the noun.

Part of Determi- Adjectives Noun


Speech ner
Specific Determining Deadverbial Scalar Denomi- Compound
Adjective nal Adj.
Type
General Identify- Identifying Characteriz- Charac- Classify- Classifying
Function ing ing terizing ing
the only beautiful long short story
the first intelligent magnifi- linguistic paper
cent

Table 4. The premodifier construction in the NP with positions of adjectives from left to right. At the left
of the NP we have the more grounding elements, those that anchor the referent of the noun, at the center
the characterizing elements and at the right the more classifying elements. Two NPs are given as
instances.

Specific Adjective Type Determining Deadverbial Scalar Denominal Compound adj.


Property Stability
continuum Minimal ----------------------------------------- Maximal
Subjectivity
continuum Maximal ---------------------------------------- Minimal

Table 5. The continua of property stability and subjectivity and their relation to the positions of the
premodifier construction.

The constructional view adopted here also takes into account the observation that a shift
in position can encode a shift in meaning and vice versa. Moreover, it is important to
note that this specific arrangement depicts the unmarked or default way of arrangement,
as there are also marked sequences that do not follow this exact order. When such an
order occurs, it is also accompanied by a marked phonological pattern, which means

40
that in this case a change in syntactic behavior is accompanied by a change in
phonological behavior.
A consequence that follows from the comments on property stability concerns the
distinction permanent/temporary. The premodifier zone starts with minimal stability to
the left and ends with maximal stability to the right, as it is observed by Athanasiadou
(2006) and Halliday (1985). This means that the association between permanent sense
and prenominal position is partially correct. As we noted, the more stable adjective
types are the denominal as well as those participating in compounds. However, the
determining adjectives possess minimal stability, while scalar and deadverbial types are
found between the two extremes. The stability cline follows the cline from identification
to classification as well as the cline from the extrinsic property of the object to the more
intrinsic one. Last, it follows the subjectivity cline from the more speaker-oriented
meanings to referent-oriented ones and from context-dependent reference to context-
independent denotation.
The facts above align with each other and further support the claim that what we
are concerned with is a single construction. There is, however, an approach that
considers the existence of two separate adjectival constructions, the predicating and
domain adjective constructions (Sullivan 2013: 7). The distinction is based on the
adjective‟s ability to appear after the copula, with predicating adjectives appearing in
this position (e.g., a beautiful girl can also be a girl who is beautiful) but domain
adjectives cannot (e.g., an electrical engineer cannot be *an engineer who is electrical).
I think that the basis for recognizing two different constructions is Bolinger‟s (1967)
distinction, referent- and reference-modification, with predicating construction being
referent-modification and domain adjective construction being reference-modification.
However, as I have mentioned, Adamson (2000: 63) observed that reference-
modification conflates identifiers and classifiers and thus it is too general. Also, it seems
to me that rather than treating them as two different constructions, we can regard them
in a single construction with adjectives occupying different positions in it, as I have
already explained, thus being able to say a beautiful electrical engineer. Therefore, I
believe that it is more economical and theoretically elegant to posit one construction
that can account for the phenomena we study. Most crucially, it can also be feasible, as
Table 4 demonstrates.
Positing a premodifier construction, as I have presented it, also takes into
consideration Taylor‟s (1992) remarks that a premodifier adjective can express various
41
unprofiled relations in association with the noun it modifies. On the contrary, the
predicative position can only express extrinsic meanings and not any unprofiled
relations. These meanings are concerned with the noun‟s extension without any
indication as to its temporariness or permanence. As such, the predicative position can
also be regarded as another construction whose meanings are referent-oriented.
Evaluative meanings can also be expressed by scalar adjectives, since scalar adjectives
and, to a lesser extent, deadverbial ones can occur in this construction. That this is
another construction is supported by the fact that in attempting to turn a NP with a
synthetic reading into a sentence with a predicative adjective, the latter will be
interpreted mostly in an extensional fashion, for instance, old friend („aged friend‟,
„friend of long standing‟)  my friend is old („aged‟).
Last, I have also discussed the postnominal position, which has been typically
associated with temporary meanings. This appears to be a correct statement. But, as I
have argued in Chapter 4, this might occur because the adjective in this non-canonical
position seems to appear „incomplete‟ without further specification; *the man old
cannot stand on its own but the man old in years is acceptable. This structure appears to
be very similar in meaning to a relative clause (cf. the man who is old in years) and this
is also corroborated by participles the road (which was) not taken, a woman (who is)
lying. It is also true that relative clauses as well as participles can receive further
specification with adjuncts. This observation is perhaps indicative of the predicational
character of this position. Because co-textual information renders the thing talked about
rather specific, this is what may induce a sense of temporariness. Therefore,
temporariness is not per se inherent in the position but arises because the adjective is
structurally incomplete in this non-canonical position. However, there seem to be
degrees of incompleteness, for instance, a river navigable does not evoke such an
incompleteness as a man old. Consider again the fact that in this position we can also
find relative clauses. A clause expresses a full proposition built around a verb. Thus, it
expresses a predication which is attached to a noun (e.g., the man who is old). As
Teyssier (1968) noted as well, there seems to be an implicit copula in a postnominal
adjective and which is explicit in a relative clause. However, it seems that the copula is
not sufficiently evoked in the man old, but needs something else, namely old in years or
old enough. Instead, adjectives with an -able suffix are different. The suffix encodes the
very specific meaning „able to V‟, where V stands for the meaning of the verb which
serves as the base of the constructed adjective. Thus, the derivation produces an
42
adjective that encloses a whole predicate, which in turn is possible to add further
specification.
Teyssier (1968) described the postnominal position as a hybrid, being
simultaneously a form of explicit attributive and an implicit predicative. In the previous
paragraph I discussed the adjective as a form of implicit predicative, in which we
understand the presence of an implicit copula. However, classifying adjectives can
appear in this position as a form of explicit attributive and this is seen in expressions of
French origin (e.g., attorney general, princess royal, court martial) as well as others
like things French, things English, matters philosophical. Observe, however, that these
phrases do not evoke a temporary sense, as the adjectives are classifiers. Thus, they are
different from adjectives that evoke the predicational character of the position.

Furthermore, these phrases have a different stress pattern (the princess ˈroyal) from the

those that are a form of implicit predicative (a ˈman ˈyoung and ˈhandsome).
It is possible that these different patterns, implicit predicative and explicit
attributive, can be parts of a single construction in the postmodifier zone. After the
noun, what follows is the position of the explicit attributive, where classifying
adjectives can occur evoking a permanent sense due to their strong property stability.
After this position follows the implicit predicative associated with the sense of
temporariness, in the manner I have described it previously. It is interesting that in the
postmodifier zone as well, the farther an item moves from the noun, the less stable the
property it expresses, whereas the closer to the noun the adjective is, the more stable its
property. It appears, then, that the postmodifier zone is the mirror image of the
premodifier zone with one exception. In the postmodifier zone both classifiers and
characterizers can occur but not identifiers or determining adjectives. This is expected
given that identifiers have a similar function to determiners. Since determiners in
English cannot occur after the noun, then it makes sense that neither do determining
adjectives. This might also be the reason why these adjectives cannot occur in the
predicative position.
In conclusion, I have examined three adjectival constructions: the premodifier
zone construction, the postmodifier zone construction and the predicative construction.
The premodifier construction is the canonical position of English adjectives and it is
here where the full range of adjectives in all sorts of unprofiled relations with the noun
can be realized. Rather, the other two, the predicative and the postmodifier zone, are

43
more confined as to the adjective types that can occur in these positions. The
postmodifier zone construction has two positions, the explicit attributive, occupied by
classifying adjectives, followed by the implicit predicative, where adjectives seem to
require a complement or adjunct to felicitously occur. The sense of the permanence of a
property seems to be gradient in the NP. Both in the premodifier and in the postmodifier
zone, the modifiers‟ sense of temporariness or permanence is related to the proximity of
the position from the noun. The farther the position from the noun it is, the more
temporary its meaning and vice versa. Last, in the predicative construction the
adjectives that occur describe or evaluate some aspect of the extension of the noun. This
position does not evoke any temporariness or permanence per se.
As a final remark, it appears that the comments that have been made for adjectives
could perhaps also be useful for the investigation of modification in general. The
positions identified in specific constructions for adjectives could be in use for other
modifiers apart from adjectives, as it was also noticed in the discussion of participles.
Consider, for instance, the examples mentioned in Chapter 1: it was an inside joke
(adverb), ‘the dog’ is a noun phrase (noun), don’t give me an I-don’t-know look
(clause), the People’s Republic of China (noun in genitive case). In these cases, the
different items that occur as premodifiers seem to be interpreted as classifiers or a
denominal type of adjective. In short, although they are not adjectives, they seem to be
used as if they were. Perhaps this thesis may also have given some insights as to the
general issue of noun modification. The extent of such an implication is, of course, not
evaluated here and it is left for further investigation.

44
References

Adamson, S. (2000). A lovely little example. In O. Fischer, A. Rosenbach & D. Stein


(Eds.), Pathways of change (pp. 39-66). Amsterdam/Philadelphia: John Benjamins.

Athanasiadou, A. (2006). Adjectives and subjectivity. In A. Athanasiadou, C. Canakis


& B. Cornille (Eds.), Subjectification: Various paths to subjectivity (pp. 209-239).
Berlin: Mouton de Gruyter.

Bache, C. (2000). Essentials of mastering English: A concise grammar. Berlin/New


York: Mouton de Gruyter.

Bolinger, D. (1967). Adjectives in English: Attribution and predication. Lingua, 18, 1-


34.

Bolinger, D. (1968). Entailment and the meaning of structures. Glossa, 2, 119-127.

Carnap, R. (1947/1956). Meaning and necessity. A study in semantics and modal logic.
Chicago and London: The University of Chicago Press.

Chomsky, N. (1957). Syntactic Structures. The Hague: Mouton.

Dirven, R. (1999). The cognitive motivation for adjective sequences in attribution.


Journal of English Studies, 1, 57-67.

Dixon, R. M. W. (1982). Where have all the adjectives gone? Berlin: Mouton de
Gruyter.

Goldberg, A. (1995). Constructions: A construction grammar approach to argument


structure. Chicago: The University of Chicago Press.

Goldberg, A. (2006). Constructions at work: The nature of generalization in language.


New York: Oxford University Press.

45
Halliday, M. A. K. (1985). An introduction to functional grammar. London: Edward
Arnold.

Hetzron, R. (1978). On the relative order of adjectives. In H. Seiler (Ed.), Language


Universals (pp. 165-184). Tübingen: Narr.

Kitis, E. (2012). Semantics: Meaning in language. Thessaloniki: University Studio


Press.

Langacker, R. (1990). Subjectification. Cognitive Linguistics, 1, 5-38.

Langacker, R. (1991). Foundations of cognitive grammar. Volume II. Descriptive


application. Stanford, California: Stanford University Press.

Lyons, J. (1977). Semantics, vol. 1. New York: Cambridge University Press.

McGregor, W. B. (1997). Semiotic grammar. Oxford: Clarendon.

Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1972). A grammar of
contemporary English. Harlow: Longman.

Radden, G., & Dirven, R. (2007). Cognitive English grammar.


Amsterdam/Philadelphia: John Benjamins.

Sullivan, K. (2013). Frames and constructions in metaphoric language.


Amsterdam/Philadelphia: John Benjamins.

Sweetser, E. (1999). Compositionality and blending: semantic composition in a


cognitively realistic framework. In Th. Janssen & G. Redeker (Eds.), Cognitive
Linguistics: Foundations, scope and methodology (pp. 129-162). Berlin: Mouton de
Gruyter.

46
Taylor, J. (1992). Old problems: Adjectives in Cognitive Grammar. Cognitive
Linguistics, 3(1), 1-36.

Taylor, J. (1995). Linguistic categorization. New York: Oxford University Press.

Teyssier, J. (1968). Notes on the syntax of the adjective in Modern English. Lingua, 20,
225-249.

Vendler, Z. (1967). Linguistics in philosophy. Ithaca: Cornell University Press.

Vendler, Z. (1968). Adjectives and nominalizations. The Hague: Mouton.

Whorf, B. L. (1956). Language, thought and reality: Selected writings of Benjamin Lee
Whorf, J. B. Carroll (Ed.). Cambridge, Mass.: M.I.T. Press.

47

Вам также может понравиться