Академический Документы
Профессиональный Документы
Культура Документы
ToappearinPierreLarriveandChungmin
Lee(Eds).NegationandPolarity.Experimentalandcognitiveper
spectives.Springer.
TheMarkednessofDoubleNegation
1
PierreLarrive
NormandieUniversit,Unicaen,CRISCO(EA4255)
0.Abstract
IsDoubleNegation(DN)amarkedinterpretationwithrespectto
Negative Concord (NC) as universally assumed? This is what is
verifiedinthisarticlethatreportsonactualusageinFrenchand
English. It investigates three configurations with multiple
clausemate negative expressions (clausal negator with a n
word,twonwords,andtwoclausalnegators).Thedataisana
lysedfortherelativeproportionofNCandDNreadings,andfor
identifiable triggers of DN. The predictions if DN is marked are
thatNCasadefaultshouldoccurevenincontextsbiasedforDN,
and that specific collocations and contextual factors trigger DN
readings. Current work leads to the expectation that the deter
minantfactorforDNisanInformationStructure(IS)configura
tion,inwhichtherejectednegativeclauseisoldinformationex
plicitly mentioned in the antecedent context and the rejecting
negatorisdiscoursenew.Bothpredictionsaresupportedbythe
data:contextsbiasedinfavourofDNstilldisplayNCinterpreta
tionsinuptohalfofthecorpusoccurrences;andDNisstrongly
correlatedtotheexpectedISstructure(upto84%),andinother
cases to recurrent collocations (up to 46%). The findings
demonstrate that DN is marked as the result not of a macro
Iwishtothankthecontributorstothisvolumeanditsrefereesfortheirconstructive
commentsonthischapter;anearlierversionalsobenefitedfromfeedbackbyPatrick
Duffley,VroniqueLenepveuandtheparticipantstoaseminarheldinNeuchtelonHallo
ween2013.Responsibilityforanyshortcomingshouldnotbeattributedtonobodyelsebut
me.
parameter,butofapsycholinguisticbiasthatfavoursNCasade
fault interpretation for negative dependencies due to greater
easeofprocessing,inlinewithrecentpsycholinguisticresults.
Keywords:negativedependency,DoubleNegation,negative
concord,informationstructure,corpusstudies,macro
parameter
1.Introduction
It is universally assumed that the Double Negation interpretation by
which two negative markers cancel their value to yield a positive interpretation is marked (i.a. Corblin and Tovena 2001, Espinal and Prieto
2011, Horn 2001, Huddlestone 2010, Moscati 2006, Puskas 2012,
Swart 2010, Zeijlstra 2004). This is allegedly the case in so-called Negative Concord languages such as French, Spanish and Italian, where the
concord relation would be the default interpretation for multiple negative expressions sharing the same structural clausal scope. Even where
concord relations are believed to be disallowed as in standard English,
Dutch and German, Double Negation interpretation of sequences such
as (1) and (2) would continue to be marked.
(1)
(2)
2.ThemarkednessofDoubleNegationreadings
All natural languages are found to express negation. Multiple negative
expressions sharing the same clausal scope entertain two possible types
of dependencies. One is Negative Concord, by which each of these expressions contributes to the overall negative interpretation of the
proposition. The other is Double Negation, where each negative cancels the other out to imply a positive proposition. Both are illustrated
by the following French example:
(3)
(3) can be interpreted with a Negative Concord reading, where no action is taking place. This can be paraphrased by the second negative
being replaced by a negative polarity item.
(4)
The example can also have a Double Negation interpretation, such that
some action is indeed taking place. This can be disambiguated by
strong focus on one of the negatives, cleft structures and question/answer pairs.
(5)
(6)
(7)
a.
The co-occurrence of clausal negator pas not with a n-word like rien
nothing and personne nobody is assumed to exclude Negative Concord interpretations in European French (inter alia Biberauer and Roberts 2011, Corblin and Tovena 2001, Dprez 2000: 261, Giannakidou
2007: 350, Muller 1984: 64-65, Rowlett 1998: 143ss, de Swart 2010:
156). No paraphrase could therefore be expressed by replacing the nword with a NPI.
(9)
Paraphrase is rather found in the diagnostics of DN that are strong focus on one of the negatives, cleft structures and question/answer pairs.
(5)
b.
(DN)
(6)
b.
(7)
b.
Double Negation is the interpretation that also obtains when the same
scope is not shared by negative expressions (Larrive 2004: 169 i.a.), as
in the following where a clausal negation pas and personne cancel out the
local morphological and constituent scope negative.
(10)
(11)
Justas(single)negationismarkedinrelationtoaffirmation,sotooistheoccurrence
ofdoublenegation(whethersyntacticorsemantic)inrelationtotheoccurrenceof
singlenegation.(Huddlestone2010:16)
Doublenegationreadingsareinsteadmoredifficulttocheck,sincetheyarealways
marked.(Moscati2006:136)
6
WestartthisdiscussionfromtheassumptionthatDNissemanticallymarkedwith
respecttosinglenegation,anditissemanticallymarkednotonlybecauseDNis
unexpectedinNClanguages(i.e.,languagesinwhichmultipleexpressionsof
negationgiverisetoasinglenegationreading),butalsobecauseDNisexpressed
lessfrequentlythansinglenegation.(EspinalandPrieto2011:2404)
However,DoubleNegationinnaturallanguageisextremelyrare.Notonlyisithard
togiverisetoDNreadingsinNegativeConcordlanguages,butalsoinlanguagesthat
lackNegativeConcord,suchasStandardDutch,StandardGerman,orthe
Scandinavianlanguages,constructionswithtwonegativeelementsarehardtofind.
(Zeijlstra2004:58)
Doublenegation,inturn,ismarkedinrelationtosinglenegationforsimilarreasons.
Firstly,doublenegation,whetherthesyntacticdoublenegationobservedinNC
languages,orthesemanticdoublenegationobservedinDNlanguages,involvesmore
morphologicalcomplexitythansinglenegation,asthereismultiplemarkingof
negation.Secondly,doublenegationinDNlanguagesismarkedintermsof
(pragmatic)meaning.(Huddlestone2010:8)
DNoccursinspecialcontexts(Puskas2012:612)
Thedoublenegationreadingitselfrequiresveryspecialconditionstoarise,which
leadsonetothinkthatonenegationperclauseisthedefaultcase.(Corblinand
Tovena2001:98)
clauses such as (1) and (2) with DN readings in her Afrikaans corpus
(2010: 140). This supports the claim in the citations above that DN is
indeed rare, and therefore more marked than NC interpretation.
Finally, and a possible explanation of the third point, DN requires explicit contextual triggers in a way that NC does not, maybe
because it is harder to process as suggested by Francis Corblin (i.a.
1996). A major trigger of DN is accentual focus (Corblin 1996, Dprez
2000, Espinal and Prieto 2011, Falaus 2007a,b, Larrive 2004: 162-165,
Puskas 2012, 2009, 2006). Focus may be realised by various prosodic
factors such as pause, duration, intonation contour and pitch range
(Krahmer, Swerts, Theune and Weegels 2002), and may be supported
by gestural elements (Prieto, Borrs-Comes, Tubau and Espinal 2013).
The contradiction contour is demonstrated through perception experiments to be a key factor in bringing about Double Negation interpretation by Espinal and Prieto (2011) and by Espinal, Tubau, BorrsComes and Prieto (this volume). This seems intuitively correct. As
mentioned, the DN reading of (5a) calls for focus on one of the nwords:
(5)
a.
Configurations such as (13) are however again attested with a NC interpretation (Larrive 2004: 168; see also the discussion by Muller
2005). Finally, the use of a n-word with a clausal negative as in (2) and
(8) would support DN (Corblin, Dprez, de Swart and Tovena 2004:
420; see also Jespersen 1917: 62ff, Payne 1985: 204). While that is true
in one register of European French, many stylistic and regional varieties allow NC for this configuration. The proposed structural triggers
of DN thus at best contribute to a DN reading rather than define it.
A better trigger than structural factors therefore seems to be
found in prosodic focus. However, prosodic factors are not categorical,
as shown by results in Espinal and Prieto (2011) and Espinal, Tubau,
Borrs-Comes and Prieto (this volume), who submit a perception test
to Catalan speakers who are asked to identify the reading of question/answer configurations of the type Who didnt eat dessert? No
one.. Unlike English and French, the configuration is ambiguous in
Catalan between NC and DN, but is expected to be DN when the answer n-word is bearing a contradictory contour. Even in the maximally
favourable DN context of a negative answer bearing a contradictory
contour to a negative question, there are still 32% of subjects who
identified a NC reading. These maximally favourable environments of
question-answer and contradictory contour both relate to Information
Structure (IS). IS concerns the status of the information conveyed by
linguistic material in relation to being accessible to the hearer at that
point of the discourse (for essential references on this, see Prince 1981,
Gundel and Fretheim 1993, Krifka 2007). Accessible information is
discourse-old, it has been mentioned before, and information that has
not been mentioned or cannot be accommodated or inferred as such is
discourse-new. The answer to a question as in (7) thus has discoursenew status with the rest of the proposition being discourse-old.
(7)
a.
(7)
b.
The partition between a discourse-new negative answer to a discourseold negative question yielding DN suggest the determining role of IS.
Further indications as to the role of IS are also found by Moscati
(2006: 136ff). His study of negative scope in Italian reports that a nword in a preverbal focus position brings about DN in a language
where normally only one preverbal marker is needed for clausal negation.
(14)
10
es such as Personne ne fait rien (3) and Jean ne fait pas rien (8). (8) yields a
Double Negation interpretation as a matter of course in normative registers of European French. This associates to strong accentual focus on
pas in the contexts where it is denied that Jean does nothing (the case
where it is the n-word that is focused is not considered). (3) may yield a
Double Negation interpretation, although this need not be the case.
Again, strong accentual focus comes to bear on the negative that rejects the negative proposition, and focus on personne in (3) communicates the denial that someone is doing nothing. This would follow
from the focused item in (3) and (8) being raised in a Verum Focus position to take scope over the whole negative proposition, leading to
negative values cancelling out. (The fact that the n-word is not negative
in itself according to Puskas forces her in cases like (3) to make the
Verum Focus the recipient of the second negative value to ensure cancellation, these theory-internal considerations being orthogonal to our
concerns.) The structure is reminiscent of metalinguistic negation, rather than recursive quantification, as indicated by the H*L prosody also
found with presupposition denial. A second mechanism relating to a
weak DN is proposed (Puskas 2012) on the basis of the Hungarian data
where the clausal negative like pas or the n-word is a Contrastive Focus. The difference would be expressed by the type of focus and intonation, the syntactic position of the second negative which would be
covert in most cases and languages, and the type of alternatives involved (polar reading for strong DN induced by Verum Focus vs. nonexclusive alternatives induced by contrastive topic for weak DN).
Whatever the necessity of weak and strong Double Negation might be,
the analysis does articulate the relation between DN and IS.
Converging indications support the view that Double Negation
is the result of Information Structure configurations, where the rejected negative proposition is discourse-old and the rejecting negator is
discourse-new. To illustrate, the sequence John is not doing nothing with a
DN reading paraphrased by It is not the case that John is doing nothing
would divide into a discourse-old negative proposition John is doing nothing rejected by a discourse-new not. The fact that it would rely on a specific IS configuration would make DN a marked negative dependency
with respect to NC that does not require such a configuration. Clearly,
these proposals need to be put to the empirical test. More must be
known about the relations between IS and DN in actual usage, and
whether indeed DN is marked. A protocol for the study of DN in real
11
usage is proposed in the next section, which also presents the research
results before their significance is discussed in the final section.
3.Corpusevidence
This study is concerned with empirically testing the assumptions that
DN is a marked interpretation, and that markedness arises from a specific IS configuration. These assumptions can be tested in a variety of
ways, and are here verified by a study of actual usage. This choice is
motivated by the fact that usage study provides us with an understanding of the patterns of phenomena. Such an understanding is particularly important for a phenomenon on the triggers of which so little is actually known, as evidenced by incompletely accurate claims made by
existing studies evoked above. The importance is furthered by the fact
that it is usage that is the input for acquisition and that it is therefore
crucial for the cognitive status of a linguistic phenomenon.
The study of DN is conducted by looking at sequences in
French, a Negative Concord language. DN should therefore be a
marked option correlating to particular triggers, hypothesised to be an
IS configuration on the basis of existing studies and available observations. IS is here defined as the status of information conveyed by linguistic material as to whether it is accessible to the hearer at that point
of the discourse (Dryer 1996, Schwenter 2005, Larrive 2012). Because
the impact of IS configuration on DN is what is tested, I select sequences believed to be strongly biased in favour of DN. One such sequence is (8), repeated below.
(8)
12
terpreted as He didnt do anything some of the time. The other expectation from DN markedness is that it should correlate to identifiable contextual triggers, with a dominant involvement of IS. More specifically, the underlying negative proposition should be discourse-old in
having been explicitly mentioned in the antecedent context, and the rejecting negative to be discourse-new. To repeat, in (8), the propositional negative pas would be new information rejecting the discourse-old
negative proposition Jean ne fait rien John is doing nothing, which
would be expected to have been mentioned explicitly before. While
previously explicit usage is not the only diagnostic of discourse-old status, it is the most tangible manifestation, and therefore, given the notorious difficulties in identifying discourse status of propositional material in actual usage, it is the one that we retain here.
The search for the string na pas rien in March 2013 yielded
445 hits. Each was extracted, and 74 irrelevant cases were excluded.
They concerned manifestly ungrammatical sentences (from e.g. automatic translation), productions from second-language speakers, repeated pages, sequences with items belonging to different sentences or
clauses, and cases of mention as opposed to usage. The 371 relevant
occurrences were analysed for their reading (NC or DN) and for the
explicit presence of the rejected negative proposition in the antecedent
context.
Just over half the occurrences have a DN interpretation, with
186, versus 185 with a NC reading. The NC reading relates to explicit
activation in 39 cases out of 185 (21%), as illustrated below.
(15)
a.
13
b.
a.
b.
14
- non () seulement le SAV na pas rien fait, il a bloqu ma xbox avec ce fameu bug. (http://www.logicsunrise.com/forums/topic/40737-aide-sav-xbox-360/)
- Yes, that would maybe explain why the SAV wanted
to do nothing
- No, except the SAV didnt do nothing, it blocked my
Xbox with the well-known bug
Explicit previous mention is often signalled by quotation marks and
reference to an antecedent statement typically by another speaker.
(18)
a.
b.
15
Explicitly
activated
Other
Totals
DN
122
NC
39
146
185
161
210
371
Table1.DistributionofreadingofnapasrieninGoogleFrance
16
are extremely difficult to find in French (only one attestation being cited by Larrive 2004: 166). Google is a poor search engine for such rare
configurations, and there is as of yet no extensive collection of searchable French non-literary texts. As a DN language, English should have
strongly preponderant DN interpretation for examples such as (1) Nobody is doing nothing, which would allow to check for the intervention of
IS factors. The configuration was searched for through the cooccurrence of nothing and nobody at the maximal distance of 9 words allowed by the search engine of the COCA corpus made available at
Brigham Young University. Of the 480 examples that were returned,
110 were relevant in having the two n-words in the same clause. The
higher proportion of noise in the English (77%) than in the French
(17%) search follows form an exact string not being set and from the
greater distance between the items. Also, cases of coordination were
excluded from the count, as this is a context previously unnoticed in
existing studies that exclusively allows NC readings2. The following can
only mean that the person in question was not going to be disturbed,
and not that she was going to be disturbed by everything and everyone.
(19)
In any event, the relevant 110 occurrences only yielded 6 cases of DN.
Each case involved constituent scope of n-word nothing in the expression for nothing rejected by the other negative expression, as illustrated
below.
Areviewerhowevernotesthataparallelreading,involvingneithercon
cordnordoublenegation,seemspossiblewithconjunction.WhereasNobody
didnothingmightbeseenasequivalenttonobodydidanythinginaNCread
ing,andtoEverybodydidsomethinginaDNreading,thesentenceIwantno
dogsandnocatscanneithermeanIwantnodogsandanycats,norIwanteve
rydogandsomecat(s).Astandardinterpretationinvolvingintersectionor
unionofgeneralizedquantifiers(alaKeenan&Faltz1985,oranyotherMon
tagovianapproach)correctlypredictsthatthetwonegativequantifiersdo
notcancelout,anddonotrequireaspecialreadingforoneofthenegative
quantifiers(asinNC)either.
17
(20)
a.
b.
a.
b.
a.
b.
18
a.
b.
c.
d.
(25)
a.
The results from the search are summarised in the following table.
19
Explicitly
activated
Other
Totals
DN
16
3 (2 with 19
recurrent
expressions)
NC
14
16
30
30
19
49
Table2.DistributionofreadingofcooccurrenceofnobodyandnotinCOCA
This confirms what was found for the same configuration in French.
Despite English being a DN language, and despite the configuration
being strongly biased in favour of DN, DN is a minority reading.
While explicit activation plays a role for a little under half of NC cases,
it associates to 84% of English DN, which a standard statistical test
shows to be significant. When not explicitly discourse-old, DN cases
relate to recurrent expressions in 2 cases out of 3. Both triggers account for 95% of English DN data, thus supporting the markedness of
DN as well as the French data do.
A final confirmation of DN markedness even in DN-biased
configurations and of the role of contextual cues and IS determinants
for the interpretation is sought by looking at adjacent uses of clausemate not in the Coca corpus. The 204 395 single occurrences of not
yield 102 hits, of which 89 are clause-mate. There are 34 cases that do
not have a DN interpretation: they are found in spoken and written
press and may or may not relate to performance errors (as described in
Horn 2010: 124-128), although some of these appear genuine cases of
negative spread.
20
(26)
a.
b.
with a possible emphatic value for the examples above. Explicit activation plays a role for 13 occurrences (38% of NC), non-activated cases
co-occurring with modals in 8 further cases (24% of NC).
The 55 DN cases relate to explicit activation in a little under
half of the occurrences (49%).
(27)
a.
b.
in my book, it's not about the weight. But it's not not
about weight if you're uncomfortable in your body.
Aren't those sex toys? " # " Not exactly, I don't think.
But they're not not sex toys, either.
In 1 case, it is the positive proposition rather than the negative one that
was present in the antecedent context.
(28)
The 28 cases of DN that do not involve explicit activation are massively co-occuring with modal verbs, typically can, with 26 examples (93%
of non-explicitly activated DN instances).
(29)
a.
b.
She stops. They will all think she is crazy if she doesn't
stop. You can not not expect people to understand
such things as your whole body coming to the strangest,
most vivid kind of life when the sight of something in
the present sparks a memory of the past.
On the other hand, even in the churches you can only
trust up to a certain degree. Everyone knows that there
are some people in the churches whose job it is to report things to the police. And, in fact, they can not not
do that because they don't know who else is an informer in the community.
21
Explicitly
activated
Other
Totals
DN
27
28 (26 with 55
recurrent
expressions)
NC
13
21
34
40
49
89
Table3.DistributionofreadingofadjacentnotnotinCOCA
4.Concludingdiscussion
This paper addresses the issue of negative dependencies between multiple clause-mate negative items. Conclusive results on Negative Concord and Double Negation interpretations are provided as to the following assumptions:
22
PatrickDuffleycorrectlyidentifies(21)and(22)asdisplayingnonprescriptivefeatures.
MypointisthatNCreadingsofclausematenegativesineitherEnglishorFrenchcannotbe
explained away by relegating them exclusively to regional dialectsknown only to some
speakers: the data suggests that speakers of these languages are able to produce and un
derstand clausemate negatives with a NC reading (see the evidence in Blanchette 2013).
ThisofcourseiswhatisexpectedifNCisunmarkedforpsycholinguisticreasonsasdiscus
sed below. The vehement denial of this ability by some speaker is a testimony to the
strenght of prescriptive exhortations in communities with a normative tradition (Haspel
math1997:section8.2).
23
24
Configuration
DN
Explicitly
Not explicitly
Percentage of
interpretation
activated DN
activated DN
DN not tied to
as % of total
as % of total
with recurrent
an
corpora
DN
expressions as
trigger as % of
% of total DN
total DN
occurrences
84%
11%
5%
66%
30%
5%
John
is 69%
not
not
working
49%
47%
4%
identified
Table3.DistributionofDNreadingsintermsofInformationStructure
25
b.
It is the predicate that is discourse-old, and neither negative is mentioned before, although nothing is accommodated as activated as suggested by the paraphrase I hope its not the case that nobody is listening. There
is a similar case cited in (28) among the 27 occurrences of activated
DN found in the not not configuration.
Information Structure configurations are also realised through
recurrent expressions. 42% of English DN cases for examples such as
(1) are found in correction environments of the type not nobody; 72% of
the French cases relating to (8) are found with the four most frequent
contexts; 47% of DN with two clausal negators in English relate to
modal verb environments.4 That is not to say that DN cannot occur
outside these environments, but some collocational environments definitely favour a DN reading. This further supports assumption 2 above
that DN markedness relates to specific contextual triggers. (It was
pointed out that NC has one categorical trigger, and that is coordination, illustrated by (19), for which an explanation is yet to be offered by
theoretical models, although see footnote 2.) This contradicts a macroparameter analysis in which French would be NC across the board and
English DN. The presence of NC to significant degrees in two DNbiased configurations in both languages does not support a macroparametric analysis where one language would favour NC and the other DN. The fact that configurations with two clausal negators are more
related to DN that the other two configurations further illustrate the
relation of negative dependencies to specified environments, against a
macro-parametric approach. The absence of contextual cues in a lan
4
Onewouldwantthisobservationtoreceiveanexplanation.Ispeculatethattherearetwo
probablereasonsforEnglishpreferenceforDNinmodalenvironments:theymakeparti
tionofthediscoursenewnegatorandthediscourseoldnegativepropositioneasierto
computeasthediscourseoldnegativecanberelatedtotheinfinitiveverb;suchapartition
canalsobehelpedbythetendencyofmodalstorangeoveralternativeevent,including
nonrealisation,thusfacilitatingparsingofdoublenegatives.Thedemonstrationofeither
thesespeculationssadlygoesbeyondthescopeofthischapter.
26
guage may account for the difficulties in eliciting DN, Romanian being
a case in point (Falaus 2006), calling for further examination.
If parameters are not responsible for negative dependencies,
then what is? The comparable situation in English and French suggest
that we may be dealing with a psycholinguistic bias rather than some
categorical parameter. An indication of this is the general assumption
that all Creoles allow NC (Dprez 1999), and given that Creoles are believed to provide with unmarked options (e.g. Bickerton 1981), NC
would therefore represent a universally unmarked option, forcing DN
to be signalled by contextual factors. Such a conclusion would need
support from acquisitional studies. The prediction if there is a cognitive bias in favour of NC is that the acquisition and processing of DN
would appear later or be more demanding. Fortunately, such studies do
exist, and the results they document tally with expectations. Zhou,
Thornton and Crain (2013) establishes that whereas NC was unproblematic to Mandarin-speaking children, these did not achieve adult-like
results in comprehension and production of DN until age 6, before
which they interpreted and produced DN configurations as NC. This
suggests that the non-adult performance of children before age 6 might
be due to working memory limitations, echoing the claim by Corblin
that DN is more difficult to process than NC. There are reasonable
grounds to propose that DN markedness is due to a bias in favour of
NC because the latter is easier to process. We are back to Corblins
proposals, although with rather a better empirical basis to support the
claim.
The work presented here establishes the following:
1. that DN is a marked reading with respect to NC, the latter
being found even in DN-biased environments;
2. that markedness is found in DN being associated to collocational environments that vary according to languages and to the
specific configuration involved (recurrent expressions for
French Jean ne fait pas rien, corrections and modals for He is not
doing nothing, modals for John is not not working)
3. that DN is strongly correlated to an IS partition such that
the rejected negative clause is discourse-old and the rejecting
negative is discourse-new.
It has been put forward that DN markedness is due to a psycholinguistic bias in favour of NC. That would be why DN is correlated to avail-
27
Noncategoricalassociationisexpectedaccordingtoareviewerbecausewearedealing
withpragmaticmatters,whichwouldonlyberelatedtogrammaticaltendencies.Icould
notdisagreemore.IhaveestablishedinLarrive(2011)thatsomemarkednegativesare
categoricallyusedwithdiscourseoldpropositions,andthatoncethiscategoricalassocia
tionislost,thenegativeeitherbecomesadefaultordisappears.
28
a.
b.
5.References
Amaral, Patrcia, and Kirby Varnadoe-Russ. 2013. An experimental approach to
Levinson's M-based implicatures. Talk at CIL, Geneva, July 22-27 2013. Powerpoint presentation, University of North Carolina at Chapel Hill. 42
slides.
Biberauer, Theresa, and Ian Roberts. 2011. Negative words and related expressions: A new perspective on some familiar puzzles. In The Evolution of
29
negation. Beyond the Jespersen cycle, ed. Pierre Larrive and Richard Ingham,
23-59. Berlin: Mouton de Gruyter.
Bickerton, Derek. 1981. Roots of Language. Ann Arbor: Karoma.
Blanchette, Frances. 2013. Negative Concord in English. Linguistic Variation 13,1, 147.
Chen, Lijing, Xingshan Li, and Yufang Yang. 2012. Focus, Newness and
Their Combination: Processing of Information Structure in Discourse.
PLoS ONE 7(8): e42533. doi:10.1371/journal.pone.0042533.
Corblin, Francis. 1996. Multiple negation processing in natural language. Theoria 62,3, 214-259.
Corblin, Francis. 1994a. Smantique des ngatifs multiples. In La ngation : Actes du colloque de Paris X Nanterre, 12-13-14 novembre 1992, ed. Pierre Attal,
279-298. Linx, special issue 29.
Corblin, Francis. 1994b. Le traitement des complexes ngatifs en DRT. In
Actes du Colloque ECCOS 92, 157-175.
Corblin, Francis. 1994c. Multiple negation processing. Human Communication
Research Centre, University of Edingburgh. Report 62.
Corblin, Francis, Viviane Dprez, Henritte de Swart, and Lucia Tovena.
2004. Negative concord. In Handbook of French Semantics, ed. Francis
Corblin and Henritte de Swart, 417-452. Stanford: CSLI.
Corblin, Francis, and Ivan Derzhanski. 1997. Multiple negation, optional arguments and the reification of eventualities. In Empirical issues in formal syntax and semantics: selected papers from the Colloque de syntaxe et de smantique de
Paris 1995, ed. Francis Corblin, Danile Godard and Jean-Marie Marandin.
219-242. Berne : Peter Lang.
Corblin, Francis and Lucia Tovena. 2001. On the multiple expression of negation in Romance. In Romance Languages and Linguistic Theory 1999, ed.
Yves D'Hulst, Johan Rooryck and Jan Schroten, 87-115. Amsterdam: Benjamins.
Dprez, Viviane. 2000. Parallel (A)symmetries and the internal structure of
negative expressions. Natural Language and Linguistic Theory 18,2, 253-342.
Dprez, Viviane. 1999. The Roots of Negative Concord in French and
French Based Creoles. In Language Creation and Language Change: Creole, Diachrony and Development, ed. Michel DeGraff (ed.), 375-428. Cambridge: MIT
Press.
Dryer, Matthew S. 1996. Focus, Pragmatic Presupposition, and Activated
Propositions. Journal of Pragmatics 26,4, 475-523.
Espinal, Maria Teresa, Susagna Tubau, Joan Borrs-Comes and Pilar Prieto.
2015. Double Negation in Catalan and Spanish. Interaction between Syntax and Prosody. This volume.
Espinal, Maria Teresa and Pilar Prieto. 2011. Toward constraining double negation. Journal of Pragmatics 43,1, 2392-2410.
30
Falaus, Anamaria. 2007a. Double negation and negative concord: the Romanian puzzle. In Romance linguistics 2006, ed. Jos Camacho and Viviane Dprez, 135-148. Amsterdam: Benjamins.
Falaus, Anamaria. 2007b. Le paradoxe de la double ngation dans une langue
concordance ngative stricte. La ngation dans les langues romanes, ed. Franck
Floricic, 7597. Amsterdam and Philadelphia: Benjamins.
Giannakidou, Anastasia. 2007. N-Words and Negative Concord. The Blackwell
Companion to Syntax, ed. M. Everaert and H. van Riemsdijk, 327-391. London: Blackwell.
Gundel, Jeanette K. and Thorstein Fretheim. 1993. Topic and focus. Handbook of Pragmatics, ed. Gregory Ward and Laurence R. Horn (eds.), 175196. London: Blackwell..
Haspelmath, Martin. 1997. Indefinite pronouns. New York: Oxford University Press.
Horn, Laurence R. 2010. Multiple negation in English and other languages.
The expression of negation, ed. Laurence R. Horn. Berlin: Mouton de Gruyter.
Horn, Laurence R. 2001. A natural history of negation. Stanford: CSLI.
Horn, Laurence R. 1991. Duplex negatio affirmat...: The economy of double negation. Papers from the 27th Regional Meeting of the Chicago Linguistic Society. Part
two: The Parasession on negation, ed. Lise M. Dobrin, Lynn Nichols and Rosa
M. Rodriguez, 80106. Chicago: Chicago Linguistic Society.
Huddlestone, Kate. 2010. Negative Indefinites in Afrikaans. Netherlands Graduate School of Linguistics
Jespersen, Otto. 1917. Negation in English and other languages. Copenhagen: A. F.
Hst.
Krahmer, Emiel, Marc Swerts, Marit Theune and Mieke Weegels. 2002. The
dual of denial: two uses of disconfirmations in dialogue and their prosodic
correlates. Speech Communication 36,1-2, 133-145.
Krifka, Manfred. 2007. Basic notions of information structure. Interdisciplinary
Studies of Information Structure 6, ed. Caroline Fry and Manfred Krifka, 1356. Potsdam.
Larrive, Pierre. In press. The continuity of the vernacular. The case of the
evolution of negative doubling in French. Diachrony of Negation, ed. MajBritt Mosegaard Hansen and Jacqueline Visconti. Amsterdam: Benjamins.
Larrive, Pierre. 2012. Positive polarity, negation, activated propositions. Linguistics 50,4, 869-900.
Larrive, Pierre. 2011. The Role of Pragmatics for Grammatical Change : The
Case of French Preverbal Non. Journal of Pragmatics, 43,7, 1987-1996
Larrive, Pierre. 2004. Lassociation ngative. Depuis la syntaxe jusqu
linterprtation. Geneva: Droz.
Levinson, Stephen C. 2000. Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge: MIT Press.
31