The Weekly Newsletter of MIT Linguistics

LFRG 2/10 - Wataru Uegaki

Speaker: Wataru Uegaki
Title: Emotive factives and the semantics of question-embedding
Date/Time: Monday, Feb 10, 12:00pm
Room: 66-148

At least since Karttunen (1977), it has been observed that emotive factives such as “surprise”, “amaze” and “annoy” exhibit puzzling embedding behavior. As in (1) below, they embed declaratives and constituent wh-complements, but don’t embed polar questions (PolQs) and alternative questions (AltQs).

(1) a. It is surprising that they served coffee for breakfast. (declarative)
b. It is surprising what they served for breakfast. (constituent question)
c. *It is surprising whether they serve breakfast. (PolQ)
d. *It is surprising whether they served coffee or tea for breakfast. (AltQ)

This fact poses an interesting challenge for the semantics of question-embedding. First of all, since this embedding behavior holds across predicates of similar intuitive semantic class cross-linguistically (e.g., German, French and Japanese), it is desirable if it can be derived from the semantics of these predicates, rather than from idiosyncratic selection restrictions. However, in the standard treatment of question-embedding, where embedded questions are converted to some form of their answer, it is not clear why emotive factives are incompatible with PolQs and AltQs. This is so because there is no semantic anomaly in the predicted truth-conditions of surprise + PolQ/AltQ sentences: ‘x is surprised by the answer of whether p’ or ‘x is surprised by the answer of whether p or q’.

In this talk, I propose a solution to this puzzle employing the independently established distinction between strongly exhaustive and weakly exhaustive readings of questions (Heim 1984; Beck & Rullmann 1999). According to the proposal, a strongly exhaustive reading is ruled out in questions embedded under emotive factives as it would violate a principle similar to Strongest Meaning Hypothesis (Dalrymple et al. 1998). On the other hand, AltQs and PolQs are inherently strongly exhaustive (George 2011, Nicolae 2013), which conflicts with the requirement against strong exhaustivity under the relevant predicates. After giving the detailed account of the embedding behavior of emotive factives, I lay out the general typology of attitude predicates that the proposed view entails, and discuss some open issues.