The Short Answer: Implications for Direct Compositionality

Monday, 9 November 2009, Colloquium

Pauline Jacobson, Brown University.

Abstract

This talk will consider the analysis of “short” answers to questions (as in the dialogue in (1a-b) and their consequences for the hypothesis of direct compositionality:

(1a) Who left the party at midnight?
(1b) Claribel.

The hypothesis of direct compositionality maintains that the syntax and semantics work in tandem, with the consequence that each local expression has an interpretation. It will be shown that such a view would lead us to doubt an analysis of (1b) whereby it is elliptical for a full sentence (as has been proposed in, e.g., Morgan 1973, Merchant 2003, and many others). Rather I will defend (with modifications) the proposal in Groenendijk and Stokhof (1984) (see also, among others, Ginzburg and Sag 2000) according to which short answers are not syntactically sentences nor, by themselves, semantically propositions. There is, however, a compositional semantic rule relating question-answer pairs, and hence the proposition that Claribel left the party at midnight is derived from combining the meanings of the question with that of the answer. Unlike Groenendijk and Stokhof, however, I will argue that only (1b) is a proper linguistic answer. The full reply Claribel left the party at midnight serves the purpose of priving a response, but it is not a true answer in the sense of a lingusitic question/answer pair. First, I discuss some past argument for the ellipsis account, showing that these do not go through. Classic arguments for the ellipsis account involve various syntactic matching requirements - such as, for example, case matching. It will be shown, however, that ellipsis actually cannot be the account of this requirement, whereas it can easily be accounted for without ellipsis (which was known already in G&S 1984).. Second, I provide some new evidence against ellipsis. Consider for example (2a) and the possible responses in (2b) and (2c);

(2a) Which mathematics professor left the party at midnight.
(2b) Jill
(2c) Jill left the party at midnight.

(2b) is a proper answer and presupposes that Jill is a mathematics professor; (2c) seems like a “best I can do” answer (and does not commit to whether Jill is a mathematics professor). Similar contrasts emerge in questions which explicitly ask for an exhaustive answer:

(3a) Who all left the party at midnight?
(3b) Jill and Tom
(3c) Jill and Tom left the party at midnight.

(3b) commits to the answer being an exhaustive listing; (3c) does not. It will be shown that this follows from the analysis defended here. However, these contrasts will prove problematic for the ellipsis account. Such an account will need a condition that ellipsis is possible only if the reply is an actual “answer” to the question, but it will be shown that no reasonable definition of “answerhood” will get all of the facts. Finally, the evidence that short answers are not elliptical in turn provides new evidence for a purely local treatment of a variety of phenomena, thus supporting direct compositionality.