This study examined how 86 listeners perceived the talker attitude from 714 tokens produced with neutral, doubting/disbelieving, or trusting/believing prosody. Within-subject acoustic analysis of these tokens revealed that, compared to neutral prosody, doubting prosody featured a slower speech rate and trusting prosody featured a higher utterance mean F0. Listeners responded to tokens by selecting at least two attitudes/emotions from a grid of 20 labels to describe the talker's affect. Relationships between recordings and selected attitudes/emotions were analyzed with nonparametric multivariate methods. Although the labels “disbelieving” and “trusting” were selected infrequently, distinct clusters of labels were used in response to tokens produced with each attitude. Doubting tokens were often described with negative-valence labels (e.g., “annoyed”), while trusting tokens were assigned positive-valence labels (“kind”). In contrast, neutral tokens were paired with low-arousal terms like “calm,” “sad,” or “neutral.” Positive-valence labels were attached to tokens featuring a higher F0 and faster speech rate, while negative-valence labels described tokens with a slower speech rate and a rising F0 slope. These results suggest that rather than maintaining a one-to-one mapping of categorically distinct pitch contours or speaking rates to affects, listeners may use prosodic cues to sort tokens along basic affective dimensions, like valence and arousal.
Abbey L. Thomas (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: