Schwarz & Simonenko
Wh-restrictor plurality
Wh-restrictor plurality and question pragmatics∗ The grammar and pragmatics of interrogatives and their (special) uses, GLOW 41, Budapest
Bernhard Schwarz1 and Alexandra Simonenko2 1
2
1
McGill University Research Foundation Flanders/Ghent University
Introduction
A conspicuous, yet seemingly unexplored, feature of wh-questions is that they often support a restrictor plurality inference that the wh-restrictor applies to more than one entity. For example, (1) suggests that Group A contains more than one girl. (1)
Which [girl in Group A] complained?
restrictor plurality
Aligned with restrictor plurality is the restrictor non-uniqueness effect exemplified by (2): the infelicity of wh-questions with uniquely denoting restrictors, i.e., restrictors that cannot hold of more than one individual. (2) #Which [oldest member of the team] resigned?
restrictor non-uniqueness
In this talk we: • (i) propose that restrictor plurality and restrictor non-uniqueness are not grammatically encoded, but follow from natural felicity conditions on the use of information seeking questions (including some previously invoked in the analysis of certain island effects; see Schwarz and Simonenko (2016)) • (ii) support this by identifying two types of correctly predicted exceptions to restrictor plurality/restrictor non-uniqueness, where a relevant felicity condition is independently observed to be suspended.
2
The pragmatics of restrictor non-uniqueness
In a H(amblin)/K(arttunen) semantics for questions (Hamblin 1973, Karttunen 1977): ∗ The first author acknowledges support for this research from the Social Sciences and Humanities Research Council of Canada (SSHRC), grants #435-2016-1448 and #435-2013-0592.
1
Schwarz & Simonenko
(3)
[wh R(estrictor)] S(cope)
(4)
λw.{S(x)|R(x)(w)}
(5)
[Which [girl in Group A]R ] complainedS ?
Wh-restrictor plurality
Question Hamblin/Karttunen semantics
If Ann, Meg, and Cat are in Group A in w, then question’s extension in w is: (6)
{λw0 . Ann complained in w0 ; λw0 . Meg complained in w0 ; λw0 . Cat complained in w0 }
Felicity conditions Constraints on the permissible relations between context set c (Stalnaker 1978) and possible sentence denotations. (7)
answerability condition (cf. Schwarz and Simonenko 2016) Question Q is felicitous wrt c only if ∃p[∀w[w∈c → p∈Q(w)] & p is felicitous wrt c] i.e. iff in every world of the context set there is a felicitous answer.
(8)
informativity condition Answer p is felicitous wrt c only if c*p & c∩p6=∅ i.e. iff it is neither entailed by nor is incompatible with the context set.
Given (4) and (8), (7) can be shown to entail (9). (9)
consequence of answerability Condition and informativity Condition Question is felicitous wrt c only if ∃x[ c⊆R(x) & c*S(x)]
So, a felicitous wh-question requires existence of an individual that common knowledge entails to have the restrictor property but that the speaker does not know to have the scope property.
3
Deriving Restrictor Non-uniqueness
In cases like (10), the content of the wh-restrictor guarantees restrictor uniqueness for any context set c. (10) #Which [oldest member of the team] resigned? (11)
(repeated from (2))
restrictor uniqueness c ⊆ {w: |{x: R(x)(w)}| ≤ 1} i.e. the set of individuals with the restrictor property is not larger than a singleton
Suppose now (with, e.g., Horn (1972), Abusch (2010)) that questions carry an existence presupposition.
2
Schwarz & Simonenko
Wh-restrictor plurality
This requires the existence of an individual who has both R and S, as in (12). (12)
existence presupposition Question is felicitious wrt c only if c ⊆ {w: ∃x[R(x)(w) & S(x)(w)]}
Restriction uniqueness, (11), the consequence of answerability & informativity conditions in (9), and existence presupposition, (12), are logically inconsistent. In conjunction with existence presupposition, (12), restrictor uniqueness, (11), entails (13), but in conjunction with answerability condition, (9), restrictor uniqueness, (11), entails (14): the unique individual described by the restrictor must yield a H/K answer that is entailed by the context set c (to satisfy existence presupposition) and not entailed (to satisfy informativity condition). (13)
consequence of restrictor uniqueness and existence presupposition Question is felicitous wrt c only if c ⊆ {w: S(ιy.R(y)(w))(w)}
(14)
consequence of restrictor uniqueness, answerability condition, and informativity condition Question is felicitous wrt c only if c * {w: S(ιy.R(y)(w))(w)}
We propose that the restrictor non-uniqueness effect arises due to an irreconcilable conflict between answerability condition and some other felicity condition (in this case existence presupposition) (cf. Oshima 2007). (15)
4
restrictor non-uniqueness (derived) Question is felicitous wrt c only if c * {w: |{x: R(x)(w)}|≤1} i.e. the context should not entail that the set of individuals in the restrictor is equal to or smaller than a singleton.
Deriving Restrictor Plurality
The proposed account of restrictor non-uniqueness effectively delivers (15) as a derived felicity condition. However, an account of the restrictor plurality inferences illustrated in (1) – as accommodated presuppositions – would require the stronger derived felicity condition in (16). (16)
restrictor plurality (desideratum) Question is felicitous wrt c only if c ⊆ {w: |{x: R(x)(w)}|>1}
Strengthening Restrictor Non-uniqueness to Restrictor Plurality An additional felicity condition as an auxiliary premise, viz. the restrictor homogeneity condition in (17). restrictor homogeneity obtains in virtue of the speaker and hearer agreeing on the restrictor’s extension, thereby agreeing on the set of individuals that the question is about. This appears to be a natural condition
3
Schwarz & Simonenko
Wh-restrictor plurality
on felicitous use of an information seeking question, and in fact one that might help explain the need for tacit domain restriction in many cases (cf. Pesetsky (1987)). (17)
restrictor homogeneity Question is felicitous wrt c only if ∀w,w’∈c[R(w) = R(w’)]
• If the restrictor’s extension has more than one member in some context set world (Restrictor Nonuniqueness) • and the restrictor’s extension is the same in all context set worlds (Restrictor Homogeneity) • then it follows that the restrictor’s extension has more than one member in all context worlds • restrictor plurality is derived.
5
Predictions borne out
Questions of course have a broad range of uses in discourse (Hudson 1975), some of which arguably fail to respect all of the felicity conditions posited above. For any question use that can independently be established to not respect one of those felicity conditions (restrictor homogeneity, answerability condition), it is predicted that the relevant inferences are absent as well.
5.1
Quiz questions
As a quiz question, (18) surely does not require that the interlocutors agree on the members of any given set of Japanese mathematicians. Hence (18) is in violation of the restrictor homogeneity. (18)
quiz question Which [Japanese mathematician] died yesterday at age 81?
The suspension of the restrictor homogeneity predicts that the restrictor plurality won’t be derived. Borne out: (19)
quiz question Which [Japanese mathematician who won the Fields Medal in 1987] died yesterday at age 81?
Used in a quiz show setting, (19) clearly does not invite the inference that more than one Japanese mathematician won the Fields Medal in 1987.
4
Schwarz & Simonenko
5.2 (20)
Wh-restrictor plurality
Rhetorical question rhetorical question (So you think I’m not doing my share?) Which [member of the family] did all of the dishes?
(20) suggests that the speaker considers herself the one and only family member who did the dishes. So it violates the answerability condition. This predicts the absence of the restrictor non-uniqueness. Borne out: (21)
rhetorical question (So you think I’m not doing my share? After all,) which [tired female member of the family] did all of the dishes last night?
(21), asked rhetorically in a context where it is common knowledge the speaker is the sole tired female in the family and did the dishes, does not give rise to the restrictor non-uniqueness effect.
5.3
Condition suspendability vs. permanent felicity condition clash
While individual felicity conditions (answerability condition, restrictor homegeneity) seem to be subspendable in appropriate contexts (i.e. quizzes and rhetorical settings), what comes across as strongly ungrammatical are questions which give rise to a permanent conflict of two or more felicity conditions. E.g. (22) in any context violates either answerability or existence presupposition (see section 3). (22) #Which oldest member of the team.
6
Conclusions
The proposed account of restrictor non-uniqueness and plurality (i) adds to the broadly Gricean program of understanding how speakers draw inferences on the basis of pragmatic premises, and (ii) adds to the growing inventory of observed correlations between necessary violations of felicity conditions and judgments of unacceptability (e.g. Fox and Hackl 2006, Oshima 2007, Abrus´an 2014, Schwarz and Simonenko 2016).
References Abrus´ an, M´ arta. 2014. Weak Island Semantics. Oxford University Press. Abusch, Dorit. 2010. Presupposition triggering from alternatives. Journal of Semantics 27:37–80. Fox, Danny, and Martin Hackl. 2006. The universal density of measurement. Linguistics and Philosophy 29:537–586. Hamblin, Charles L. 1973. Questions in Montague English. Foundations of Language 10:41–53. Horn, Laurence. 1972. On the semantic properties of logical operators in english. Doctoral Dissertation, University of California, Los Angeles.
5
Schwarz & Simonenko
Wh-restrictor plurality
Hudson, Richard A. 1975. The meaning of questions. Language 51:1–31. Karttunen, Lauri. 1977. Syntax and semantics of questions. Linguistics and Philosophy 1:3–44. Oshima, David Y. 2007. On Factive Islands: pragmatic anomaly vs. pragmatic infelicity. In New Frontiers in Artificial Intelligence, 147–161. Springer. Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In The Representation of (In)definiteness, ed. Eric J. Reuland and Alice G. B. ter Meulen, volume 98 of Current Studies in Linguistics, 98–129. Cambridge, MA: MIT Press. Schwarz, Bernhard, and Alexandra Simonenko. 2016. Two pragmatic accounts of factive islands. In Proceedings of 46th annual meeting of the North East Linguistic Society (NELS 46), ed. Brandon Prickett and Christopher Hammerly. GLSA (Graduate Linguistics Student Association), Department of Linguistics, University of Massachusetts. URL https://www.createspace.com/6604179. Stalnaker, Robert. 1978. Assertion. In Syntax and semantics, ed. Peter Cole, volume 9, 78–95. New York: Academic Press, New York.
6