This paper reports on the prosody of rhetorical questions (RQs) and information-seeking questions (ISQs) in German for two question types—polar questions and constituent questions (henceforth “ wh-questions”). The results are as follows: Phonologically, polar RQs were mainly realized with H-% (high plateau), while polar ISQs mostly ended in H-^H% (high-rise). Wh-RQs almost exclusively terminated in a low edge tone, whereas wh-ISQs allowed for more tonal variation (L-%, L-H%, H-^H%). Irrespective of question type, RQs were mainly produced with L*+H accents. Phonetically, RQs were more often realized with breathy voice quality than ISQs, in particular in the beginning of the interrogative. Furthermore, they were produced with longer constituent durations than ISQs, in particular at the end of the interrogative. While the difference between RQs and ISQs is reflected in the intonational terminus of the utterance, this does not happen in the way suggested in the semantic literature, and in addition, accent type and phonetic parameters also play a role. Crucially, a simple distinction between rising and falling intonation is insufficient to capture the realization of the different illocution types (RQs, ISQs), against frequent claims in the semantic and pragmatic literature. We suggest alternative ways to interpret the findings.

Questions can be marked as rhetorical by their prosodic realisation. In two eye-tracking experiments, we tested whether wh-questions can be interpreted as rhetorical (RQ) or information-seeking (ISQ) based on prosody. We manipulated nuclear pitch accent type (rise-fall with a late-peak L*+H vs. falling with an early-peak H+!H*) and voice quality (breathy vs. modal) and investigated the contribution of the modal particle denn. Participants had to decide whether they heard an RQ or ISQ by clicking on one of two labels. Experiment 1 presented listeners with wh-questions containing the modal particle denn. Experiment 2 replicated Experiment 1 without the particle. Results showed that late-peak accent and breathy voice quality led to a rhetorical interpretation, while earlypeak accent with modal voice quality was interpreted as information-seeking. The presence of the particle slightly strengthened these interpretations. Listeners decided faster when presented with late-peak/breathy and early-peak/modal compared to the other conditions. Fixation data showed different sensitivity to the prosodic cues depending on the presence of denn. In sum, listeners can use the prosodic realisation of wh-questions to interpret them as rhetorical or not, i.e. contextual linguistic information and other means (e.g., syntactic or lexical) are not strictly necessary.

The present study deals with a key factor of speaker charisma: prosody or, to put it less technically, speech melody. In a contrastive analysis, we investigate the extent to which computer-aided real-time visualizations of acoustic-melodic parameters in a self-guided training task help speakers use these parameters more charismatically. Fifty-two speakers took part in the experiment, subdivided into four equally large, gender-balanced groups. Each group received the same introduction and instruction, but then trained their melodic charisma with a different software tool. One toolthe "Pitcher"was specifically developed for visualizing melodic parameters and providing feedback on their proper use in charismatic speech. The other software tools serve language-or singer-training purposes. Results show that three out of the four tools in fact support speakers in adopting a more charismatic speech melody after training, whereby the "Pitcher" outperforms all other tools by more than 25%. We discuss the implications of our findings for refining speech-melody visualization strategies (also for L2 training) and for digitizing rhetorical speaker coaching.

No abstract

No abstract