Comparing the roles of f0, speech rate, and timbre in expressing and perceiving politeness in Mandarin speech

Abstract

Previous research on vocal politeness has primarily focused on the acoustic correlates of polite expression and has yielded divergent findings, leaving unclear how multiple acoustic cues contribute to politeness perception. To address this gap, we conducted both speech production and perception experiments. In the production experiment, multivariate analyses showed that polite and rude utterances conveying identical linguistic content could be reliably distinguished using 11 acoustic features, with timbre contributing most prominently, followed by speech rate, whereas f0 contributed least. In the perception experiment, speech stimuli were generated by morphing between polite and rude utterances, with f0, speech rate, and timbre independently manipulated while other features were held constant. Linear mixed-effects models fitted to politeness ratings revealed timbre as the most robust cue, speech rate as the least influential, and f0 contributions varying with speaker gender and morphing direction. Latent class mixed-effects models further indicated largely consistent cue-use patterns across listeners. Although based on a limited dataset, particularly in the perception experiment, these findings provide preliminary insights into the relative roles of acoustic features in the expression and perception of vocal politeness in Mandarin and support analysing politeness as a form of prosodic mitigation.