Abstract EN:

This thesis investigates variability at the listener level in the mapping between intonation and question bias. The language investigated is Salerno Italian, for which previous investigations have unveiled a variable mapping between intonation and meaning. The general hypothesis behind the thesis is that variability can be explained by investigating the way phonological and phonetic elements are used to specify epistemic bias and by considering listener-specific variability. The thesis starts from a production analysis. Results confirm the presence of high levels of variation, mainly depending on speaker-specific choices. Then, two perception experiments were designed to investigate intonation-bias mapping and individual variability. The first experiment tested the mapping between phonological tunes and bias. Results showed that intonation is used to extract bias and identified two specific cues to positive bias: L+H* and H%. The second experiment tested the effect of pitch range. Results showed that a greater excursion in L+H* and H% is linked to positive bias. The experiments also tested for specific listener variability. Results showed that exposure to other languages and empathy are informative with respect to the cues exploited to decode bias. Exposure predicts whether listeners will use pitch accents or tonal edges to interpret bias. As for empathy, it was found that low empathy listeners extract bias based on pitch accent alone and high empathy listeners can process both cues (accent and boundaries) at the same time. Finally, specific considerations have been made with reference to how research should integrate variability into an intonational meaning model.