Bouletic bias in German questions: Evidence from production and perception

Abstract

This thesis examines the syntactic, semantic, and pragmatic characteristics of questions with a bouletic bias and how they are prosodically realized by German speakers. Biased questions are not neutral towards the answer from the addressee but carry implicit information about the questioner's predispositions or attitudes towards the possible answers (e.g., Büring & Gunlogson, 2000; Krifka, 2011; Reese, 2007; Sudo, 2013; van Rooy & Safarova, amongst others). In questions with a bouletic bias this attitude is a desire towards the proposition (p) of the question or its negation (¬p). The theoretical part of this dissertation establishes that biased questions can be realized with the same syntactic surface structure as their neutral counterparts (e.g., Pheby, 1975; van Rooy & Šafárová, 2003) and that they are questions both in the semantic and pragmatic sense (e.g., Krifka, 2011; Reese, 2007). While bias can be marked through morpho-syntactic structures like certain question types that are inherently biased (e.g., negative polar questions, tag-questions) or negative polarity items (e.g., to bat an eye), we argue that the pragmatic function of a biased question is detached from its grammatical form and that prosody alone is sufficient to add the speaker bias on a pragmatic level. As such, biased questions provide a case study of the interaction of prosody and pragmatics (cf. Reese, 2007: ix). In a series of experimental investigations, we analysed the production, perception, interpretation, and utilization of biased questions, integrating both quantitative and qualitative data. The conducted production studies yielded results in two ways. Firstly, it became evident that alternative questions (e.g., Kommst du mit oder bleibst du zu Hause? ‘Are you coming along, or are you staying at home?’) appear to be more suitable to pose an unbiased, neutral question, while polar questions (e.g., Kommst du mit? Are you coming along?’) are typically used to formulate a biased question (confirming assumptions by Biezma & Rawlins, 2012); and secondly that in string-identical questions speakers use prosodic marking strategies to indirectly express their underlying intention or desire for one of the possible answers. To this end they employ categorical and gradual differences in the intonation contour, i.e., the produced tonal height throughout the utterance, as well as in the speech rate. Specifically, higher pitch and a larger pitch range in the final rise is associated with neutral questions while lower pitch, a shallower utterance final rise and a slower speech rate are associated with bouletic bias. Statements obtained in a qualitative interrogation showed that speakers intentionally devote these prosodic aspects to the marking of their desires. This was indirectly corroborated by the results of a decision task: Listeners are capable to distinguish between biased and neutral questions when hearing them in isolation, i.e., without any further contextual information, suggesting that prosody alone is a sufficient cue to speaker bias. Participants furthermore demonstrated that they are aware of the purpose and benefits of questions with a bouletic bias in natural communication and that they assign specific meanings (correlated with the connotative meaning and intended discourse function of biased questions) to the intonation contours that were identified as prototypical for the different pragmatic functions. Although the exact interpretive contribution of any prosodic cue is highly context dependent and we therefore restrain from postulating a one-to-one relation between prosodic form and pragmatic feature, the results still highlight the importance of prosody in conveying certain discourse functions. Overall, this work contributes to our understanding of how bias is expressed and perceived in spoken communication and on the role of prosody to signal pragmatic functions. Thereby, the empirical findings allow for important theoretical conclusions at the prosody-pragmatics interface. We argue that bouletic bias might be modeled as a multilayered pragmatic function consisting of a complex speech-act type (following Reese, 2007) in which they are simultaneously questions and requests, as well as an additional propositional attitude (towards p or ¬p), both being marked by a bundle of distinct prosodic cues.