Negation and prosody as markers of pragmatic bias in English polar interrogatives

Polar interrogatives are often considered the unmarked way of seeking information in English. Yet, several cues can contribute to changing the unmarked, unbiased interpretation into a biased one. For instance, adding negation as in the negative question in (1a) expresses the speaker’s bias toward the positive propositional alternative (p), i.e., that the addressee has work to do. Accenting the auxiliary in the positive question in (1b) may also be interpreted as expressing a speaker bias, but now toward the negative alternative (¬p) (Romero & Han, 2004). Another ingredient for the expression of bias is the nuclear contour (Arnhold et al., 2021).

(1) a. Don’tH* you have WORKL* to do H-H% 
     b. DOL*(+H) you have work to do H-H%

We present two perception experiments which explore the effects of negation, auxiliary accenting, and variation in the nuclear contour in English polar interrogatives. The tested polar interrogatives (154 in total, one half negative and the other half positive) were semi-natural examples from a US TV soap (The Bold and the Beautiful) that were presented auditorily together with an avatar picture of the speaker. We employed a novel rating paradigm developed to tap into the different pragmatic meaning components associated with biased questions. In Experiment 1 (testing for basic speech act meaning), we assessed the extent to which the speaker is (i) giving out information, (ii) seeking information, and (iii) is suggesting an action. In Experiment 2 (testing for the illocutionary meaning of question bias and expressiveness), we assessed the extent to which the speaker sounds as if they are (i) checking a previous assumption, (ii) something they just noticed, and (iii) if the speaker has strong feelings in this situation.

We find that our rating paradigm provides a nuanced picture of the effects of negation and prosody in English polar interrogatives. For instance, negative questions with accented auxiliaries and rising contours seem more information giving, suggesting an action, monitoring previous beliefs and expressing strong feelings than PQs with the same prosodic profile. Those PQs still seem the preferred choice for seeking information and checking something the speaker just noticed (i.e., contextual evidence).