Speaker-specific processing and local context information: The case of speaking rate

To deal with variation in the speech signal, listeners rely on local context, such as speaking rate in a carrier sentence directly preceding a target, as well as more global properties of the speech signal, such as speaker-specific pronunciation variants. The present study addressed whether, despite its variability even within one speaker, habitual speaking rate can be tracked as a speaker-specific property and how such speaker-specific tracking of habitual rate would interact with effects of local-rate normalization. In two experiments, listeners were exposed to a 2-min dialogue between a fast and a slow speaker. At test, listeners categorized minimal word pair continua differing in the German /a/-/a:/ duration contrast spoken by the same two speakers. The results showed that listeners responded with /a:/ more often for the fast speaker but only when words were presented in isolation and not when presented with additional local-rate information. That is, despite the general assumption that duration cues and speaking rate are too variable to be used in a speaker-specific fashion, tracking habitual speaking rate may help speech perception. The results are discussed in relation to a belief-updating model of perceptual adaptation and exemplar models.