Isn't there more than one way to bias a polar question?

摘要

I show that speaker bias in polarity focus questions (PFQs) is context sensitive, while speaker bias in high negation questions (HNQs) is context insensitive. This leads me to develop separate accounts of speaker bias in each of these kinds of polar questions. I argue that PFQ bias derives from the fact that they are frequently used in conversational contexts in which an answer to the question has already been asserted by an interlocutor, thus expressing doubt about the prior assertion. This derivation explains their context sensitivity, and the fact that similar bias arises from polar questions that lack polarity focus. I also provide novel evidence that the prejacents of HNQs lack negation, and thus only have an outer negation reading (see, e.g., Ladd in Papers from the seventeenth regional meeting of the Chicago Linguistic Society, vol. 17, pp. 164-171, 1981; Romero and Han in Linguistics and Philosophy 27(5):609-658,2004; Krifka in Contrastiveness in information structure, alternatives and scalar implicatures, pp. 359-398, 2017; AnderBois in Questions in discourse, pp. 118-171, 2019; Frana and Rawlins in Semantics and Pragmatics 12(16): 1-48, 2019; Jeong in Journal of Semantics 38(1):49-94, 2020). Based on a treatment of HNQs as denoting unbalanced partitions (Romero and Han in Linguistics and Philosophy 27(5):609-658, 2004), and competition with their positive polar question alternatives, I propose a novel derivation of speaker bias in HNQs as a conversational implicature. Roughly, if the speaker is ignorant, then a positive polar question will be more useful because it is more informative, so the use of an HNQ conveys that the speaker is not ignorant. The denotation of the HNQ then makes clear which way the speaker is biased. The result separates high negation from verum focus, and I argue that it is more parsimonious and has better empirical coverage than prior accounts.