How to infer negation scope? Logical, semantic and pragmatic features on scope restriction Jacques Moeschler Department of linguistics University of Geneva [email protected]https://sites.google.com/site/moeschlerjacques/ EPICS V 14-16 MARCH 2012 SEVILLA 1
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
How to infer negation scope?Logical, semantic and pragmatic features on scope restriction
Jacques MoeschlerDepartment of linguisticsUniversity of Geneva
Introduction: three issues for a pragmatics of negation
The pragmatics of negation is a three-sided issue:The logical issue: Which type of entailments does a negative sentence trigger?The semantic issue: What is the semantic scope of negation?The pragmatic issue: What is the pragmatic relation between the negative utterance (NEG) and the corrective one (COR)?
2
1. The logical issueIt is the well-known that the sets of entailments of negative sentences are not the same:
In ordinary negation, the corrective sentence (COR) entails the negative one (NEG)In non-ordinary negation (metalinguistic negation), two situations arise:a. COR entails the corresponding positive sentence
(POS)b. COR entails NEG and some other entailed propositions,
as presuppositions (PP).
3
Some examples1. Abi n’est pas laide (NEG), elle est belle (COR)
Abi is not ugly, she is beautifula. COR ➞ NEGb. Abi est belle ➞ Abi n’est pas laide
Abi is beautiful ➞ Abi is not ugly2. Abi n’est pas belle (NEG), elle est extraordinaire (COR)
Abi is not beautiful, she is gorgeousa. COR ➞ POSb. Abi est extraordinaire ➞ Abi est belle
Abi is gorgeous ➞ Abi is beautiful3. Abi ne regrette pas d’avoir échoué (NEG), elle a réussi (cor)
Abi does not regret to have failed, she passeda. COR ➞ NEG & NEG-PPb. Abi a réussi ➞ Abi ne regrette pas d’avoir échoué & Abi n’a pas échoué
Abi passed ➞ Abi does not regret to have failed & Abi did not fail
4
Entailment relations at a glanceThe logical issue gives rise to three types of negation: one ordinary negation and two metalinguistic negations (Moeschler 2006, 2011):
Ordinary negation is truth-conditional: it scopes over a propositionMetalinguistic negation is not truth-conditional: it scopes over some more specific contents
The semantic issue is about what negation scopes over.Ordinary and metalinguistic negations display an asymmetrical behavior:a. Ordinary negation is downward orientedb. Metalinguistic negation is upward orientedHow to explain this difference in scope?Is negation oriented or neutral?
6
The denotation of negation
In a set-theoretical semantics, negation targets the complement set of the predicate.1. ⟦Abi is married⟧M = 02. ⟦ Abi is not married⟧M = 13. ⟦ Abi is single⟧M = 14. ⟦Nath is single⟧M = 05. ⟦Nath is engaged⟧M = 16. ⟦Nath is not married⟧M = 1The domain of negation is unspecified: it becomes specific because of the corrective sentence.6. Abi is not married, she is single.7. Nath is not single, he is engaged.
A
[[MARRIED]]
[[NOT-MARRIED]]
Abi
[[ENGAGED]]
[[SINGLE]]
7
Nath
Anne
Jacques
AlexAxel
Abi
Nath
The downward orientation of negationWith scalar predicates, negation triggers a downward entailment: not-P means less than P1. Abi n’est pas grande → Abi is less than tall2. Il n’y avait pas 20 étudiants → there were less than twenty
students3. Il n’est pas certain qu’Abi vienne →it is less than certain
that Abi will comeWhy is negation downward oriented?Why is negation not neutral relative to scales?
8
A neutral analysis of negation?If negation had no intrinsic semantic orientation, then the corrective clause should equally entail upward and downward propositions:1. Abi n’est pas grande, elle est
très grande → Abi est grandeAbi is not tall, she is very tall → Abi is tall
2. Abi n’est pas grande, elle est petite ↛ Abi est grandeAbi is not tall, she is small ↛ Abi is tall
But negation is not symmetrical.When no corrective clause appears, the entailment is downward, not upward:3. Abi n’est pas grande → Abi est moins que grande↛ Abi est grande Abi is not tall → abi is less than tall↛ Abi is tall
So negation entails a «less than» quantity.
How to explain these facts?There seems to be a downward default orientation when negation applies to scalar predicates.The downward orientation explains partially why no explicit correction is required to obtain the interpretation for a descriptive negation:1. Anne does not have three
children In this case the scope of negation corresponds to the set of its entailments.
The entailments indicate the domain in which the appropriate quantity has to be searched.
For an upward reading of negation, the entailments are distinct from negation scope, and they are restricted by the corrective clause.2. Anne does not have three
children, she has four.The corrective clause is necessaryNegation does not scope over a quantitative domain, but over one specific quantity.
10
Entailment and negative orientation
11
2. Anne does not have three children, she has four.
➞ Anne has four children➞ Anne has three children➞ Anne has two children➞ Anne has one child
1. Anne does not have three children➞ Anne has two children➞ Anne has one child
ENTAILMENTSSCOPE OF NEGATION
CHILDRENCHILDREN
32
1
4
We have now an explanation why COR is necessary only with an
upward negation
3. The pragmatic issue
What is the pragmatic relation between NEG and COR?Do we have pragmatic evidence for distinguishing between three types of negation?Two criteriaa. Discourse relationsb. Connectives
12
Discourse relationsThe relation between NEG and COR is not of the same type:
1. Descriptive negation: CORRECTIONa CORRECTION relation between α and β holds if α is a false description of the world, β a true description, and β entails α.
2. Upward negation: CONTRASTa CONTRAST relation holds between α and β if α implicates not-β and if β entails α
3. Presuppositional negation: EXPLANATIONAn EXPLANATION relation holds between α and β if β explains why α is the case and β entails not-α and the negation of its entailments
13
Examples 1. Abi is not beautiful, she is ordinary
a. ⟦Abi is beautiful ⟧ = 0b. ⟦Abi is ordinary⟧ =1c. Abi is ordinary ➞ not (Abi is beautiful)
2. Abi is not beautiful, she is gorgeousa. Abi is gorgeous ➞ Abi is beautifulb. Abi is not beautiful +> Abi is not gorgeous
3. Abi does not regret to have failed; she passeda. Abi passed ➞ not(Abi failed) & not(Abi regret to have failed)
14
Connectives
These three types of discourse relations and negation are made explicit by three connectives:
Ordinary negation - CORRECTION - au contraire (on the contrary)Upward negation - CONTRAST - mais (but)Presuppositional negation - EXPLANATION - parce que/puisque (because/since)
15
Examples
1. Abi n’est pas belle, au contraire elle est quelconqueAbi is not beautiful, on the contrary she is ordinary
2. Abi n’est pas belle, mais extraordinaire.Abi is not beautiful, but gorgeous
3. Abi ne regrette pas d’avoir échoué, parce que/puisqu’elle a réussiAbi does not regret to have failed, because/since she passed
16
Is the semantics of connectives consistent with negation?The semantics of au contraire implies a contrary relation between proposition1. Abi n’est pas belle, au contraire elle est quelconque.Abi est belle and Abi est quelconque are contrary propositions:
They cannot be true together, but false togetherOne of them can be true, the other false
17
ABI EST BELLE ABI EST QUELCONQUE P ∣ Q1 1 01 0 10 1 10 0 1
maismais has as its semantics the logical meaning of a conjunction (et) and as its pragmatics a contrast between the implicature of its first discourse segment and its second one1. Abi n’est pas belle, mais extraordinaire2. Abi n’est pas belle +> Abi n’est pas extraordinaireBoth discourse segments have to describe true propositions, as the semantics of the conjunction requires it.
18
ABI N’EST PAS BELLE ABI EST EXTRAORDINAIRE P ∧ Q1 1 11 0 00 1 00 0 0
parce que/puisqueparce que and puisque are connectives introducing causal relations.1. P parce que Q ➞ [Q CAUSE P] ∧ [P ∧ Q]2. Abi ne regrette pas d’avoir échoué, parce qu’elle a réussi
a. Abi a réussi CAUSE (I inform you that) Abi does not regret to have failed
b. Abi does not regret to have failed ∧ not(Abi failed)c. Abi a réussi ➞ Abi did not fail
The use of parce que is pragmatic (speech act) - Sweetser (1990), Zufferey (2010) & (2012), Blochowiak (2010)
19
Summary We have discussed three issues concerning the scope of negation:a. The logical issueb. The semantic issuec. The pragmatic issueAs a result,we have a convergence of properties and criteria to distinguish between three types of negation:i. Ordinary negationii. Upward negationiii. Presuppositional negation
COR ➞ NEG Set of entailments CORRECTION au contraire
COR ➞ POS Restricted CONTRAST mais
COR ➞ NEG (P & PP) Wide EXPLANATION parce que
puisque
4. How to infer negation scope?
Classical semantic vs. pragmatic answers to scope attribution.The classical analysis makes a difference between the semantics and the pragmatics of negation: the default semantics is narrow scope, and the potential pragmatics is wide scope:The king of France is not wise.
a. Semantics: the K is not-Wb. Pragmatics: not[the K is W]The alternative (Gricean) analysis makes the opposite claim: the default semantics is wide scope and the pragmatics is narrow scope (Carston 2002): a. Semantics: not[the K is W]b. Pragmatics: the K is not-W
22
Arguments in favor of the pragmatic analysisWide scope reading is not always obtained as second processing, but can be prepared by the first discourse segment (Carston 2002):1. a. I’ve never smoked in my life, so I
have not given up smoking.b. I have not given up smoking; I’ve never smoked in my life.
2. a. Mary is patriotic and fanatic; not patriotic or fanatic.b. Mary is not patriotic or fanatic; she is patriotic and fanatic.
Scope restriction as a pragmatic reading is a general case, obtained with quantifiers and logical connectives too.
3. Some students passed +> some students didn’t pass
4. Some students didn’t pass +> some students passed
4. Nath turned the key and the engine started +> Nath turned the key and then the engine started
5. If you mow the lawn, I give you 10 €+> if you don’t mow the lawn, I don’t give you 10 €
6. Mum is in the kitchen or in the bathroom +> Mum is in the kitchen and not in the bathroom or in the bathroom and not in the kitchen
23
An illustration1. The default reading is wide scope
1. Anne n’a pas trois enfantsAnne does not have three children
2. [not [Anne has 3 children]]2. Wide scope reading is either
confirmed or cancelleda. It is confirmed by a COR clause
(CONTRAST):3. Anne n’a pas trois enfants,
elle en a quatreAnne does not have three children, she has four
4. [not [Anne has 3 children]] & [Anne has 4 children]
b. Negation scopes over the implicature, not the proposition
5. [not [Anne has exactly 3 children] & [Anne has 4 children]
c. Wide scope reading is cancelled if no information is given or new information requires it:6. Anne n’a pas trois enfants
(elle en a deux)Anne does not have three children (she has two)
3. Negation is attracted for reason of relevance by a local domain, that is, a predicate:
7. Anne n’a pas trois enfantsAnne does not have three children
8. [Anne not [has 3 children]]
24
5. Negative utterances and their contexts
Are the contexts for the different negation interpretation the same?Our hypothesis is that negative utterances are pragmatically processed against contexts in which some assumptions and some previous utterances are required.What are these contexts?
25
Ordinary downward negation
a. The relevant context contains POS.b. COR strengthens NEG (COR ➞ NEG)c. NEG + COR defeats POS.d. The main positive cognitive effect of
NEG + COR is to suppress POS from the actual context.
1. A: Paul croit qu’Anne a trois enfants.B: Non, Anne n’a pas trois enfants, elle en a deux.A: Paul believes that Anne has three children B: No, Anne does not have three children, she has two
2. A: Est-ce qu’Abi regrette d’avoir échoué?B: Non, elle ne s’intéresse pas à son avenir.A: Does Abi regret to have failed?B: No, she has no interest in her future
3. A: Je trouve qu’Abi est belle.B: Non, elle n’est pas belle, elle est juste quelconque.A: I think Abi is beautiful.B: No, she is not beautiful, she is just ordinary
Context a. Contextual assumption: POSb. Utterance: NEG + CORc. Contextual effect: POSA non-assertive context triggers POS as a contextual assumption4. A: Combien d’enfants a Anne?
B: Elle n’a pas trois enfants.A: How many children does Anne have? B: She does not have three childrenB implicates that POS belongs to the context (Anne has three children)
26
Upward negationWithin upward negation, the context is different.POS is maintained, and NEG + COR does not cancel it.The main positive cognitive effect of NEG + COR is the strengthening of POS and the addition of a stronger value to POS.
1. A: Abi est belle, tu ne trouves pas?B: Non, elle n’est pas belle, elle est extraordinaire.A: Abi is beautiful, isn’t she?B: No, she is not beautiful, she is gorgeous.Contexta. Contextual assumption:
In presuppositional negation, the context is either NEG or POS with a presupposition PP.NEG + COR defeat both POS + PP and NEG + PP.
1. A: J’ai vu Abi. Apparemment, elle regrette d’avoir échoué à ses examens.B: Non, elle ne regrette pas d’avoir échoué ses examens, parce qu’elle les a réussis.A: I met Abi. Apparently, she regrets to have failed.B: No, she doesn’t regret to have failed, since she passed.
2. A: J’ai vu Abi. Apparemment, elle ne regrette pas d’avoir échoué à ses examens.B: En effet, elle ne regrette pas d’avoir échoué ses examens, parce qu’elle les a réussi.A: I met Abi. Apparently, she doesn’t regret to have failed.B: In effect, she doesn’t regret to have failed, since she passed.Context1 (POS)a. Contextual assumptions: POS & PPb. Utterance: NEG + CORc. Contextual effect: POS & PP
The different contexts are consistent with the logical, semantic and pragmatic features of negation.
The scope is the result of contextual assumptions and entailments.Discourse relations and connectives are the result of the cognitive effects of negation.
29
Entailments Scope Discourse relation Connectives
Contextual assum-ptions
Utterances Cognitive effects
Ordinary negation
Upward negation
Presuppo-sitional
negation
COR ➞ NEG Set of entailments CORRECTION au contraire POS NEG+COR POS
COR ➞ POS Restricted CONTRAST mais POS NEG+COR POS+
COR ➞ NEG (P & PP) Wide EXPLANATION parce que
puisquea. POS & PPb. NEG & PP NEG+COR a. POS + PP
b. NEG + PP
Perspectives Have we made substantial progresses relative to other pragmatic approaches to negation?Not really, but we have a framework allowing further empirical studies:a. Corpus-driven research: Is the specification of the context precise enough
to be tested in large corpora?b. Cross-linguistic studies: What is variant and what is invariant across
languages regarding the 6 features described?c. Prosodic studies: Are there differences in pitch and stress within the three
types of negation relative to their contexts? d. Experimental pragmatic research: What is the interplay between contexts
and NEG in acquisition and processing, by subjects with and without an impairment, by children and adults? (Vender & Delfitto 2010 on dyslexic children).
This is what LogPrag is about.
30
References Blochowiak J. (2010), Some formal properties of causal and inferential because in different embdding contexts, Generative Grammar at Geneva 6, 191-202.Carston R. (2002), Thoughts and Utterances, Oxford, Blackwell.Carston R. (1996), Metalinguistic negation and echoic use, Journal of Pragmatics 25, 309-30.Carston R. (1999), Negation, ‘presupposition’ and the semantics-pragmatics distinction, Journal of Linguistics 34, 309-50.Moeschler J. (1997), La négation comme expression procédurale, in Forget D., Hirschbühler P., Martineau F. & Rivero M-L. (eds), Negation and Polarity. Syntax and Semantics, Amsterdam, John Benjamins, 231-249.Moeschler J. (2006), Négation, polarité, asymétrie et événements, Langages 162, 90-106.Moeschler J. (2010), Negation, scope and the descriptive/metalinguistic distinction, Generative Grammar at Geneva 6, 29-48.
Moeschler J. (2011), Pourquoi le sens est-il structuré? Une approche vériconditionnelle de la signification linguistique et du sens pragmatique, Nouveaux cahiers de linguistique française 30, to appear.Moeschler J. (in progress), Intention, commitment and propositional attitudes in linguistic communication, paper presented at the IPrA Conference, Manchester, 2011.Sweetser E. (1990), From Etymology to Pragmatics, Cambridge, CUP.Vender M. & D. Delfitto (2011), Towards a Pragmatics of Negation: The Interpretation of Negative Sentences in Developmental Dyslexia, Generative Grammar at Geneva 6, 1-28. Zufferey S. (2010), Lexical Pragmatics and Theory of Mind, Amsterdam, Benjamins.Zufferey S. (2012), Car, parce que, puisque Revisited: Three empirical studies on French causal connectives, Journal of Pragmatics 44/2, 138-54
31
Thanks for your attentionSee you in July 2013 in Geneva at the 19 ICL