Speech Rate Perception & Matching3/1/2020 What factors affect how listeners perceive and mirror speech rate? People often unconsciously mirror aspects of the speech and gestures of the people they're in conversation with, for example by talking quieter or faster when their partner does -- but not necessarily if they perceive their partner's speech as unusual. This project began by studying the relationship between the speech rates of interlocutors in laboratory tasks. First, preschoolers with cochlear implants (CI) who repeated sentences after an adult did not match the adult's speech rate as well as older children and those with typical hearing (TH). But communication is a two-way street, and CI users can't do all the work. Was this even a realistic thing to expect them to do (without instructions, no less)? Although we don't really know if TH adults naturally match speech rates sentence-by-sentence, one study (Borrie & Liss 2014) showed that TH adults adapted their speech rates in response to recordings of long sentences read by adults with dysarthria (trouble articulating, with very fast or very slow speech rates). Would TH college students match recorded speech rates of TH speakers, those with CIs, or those with CIs and atypical speech rates (either fast or slow)? Turns out, no. They didn't change their speech rates no matter how fast, slow, or unintelligible the previous sentence was. Maybe they weren't paying attention to speech rate. Maybe the speech rates in the recordings weren't varied enough to notice, or maybe they needed more than a short sentence to get the feel for the speech rate. So, in the next experiments, the recordings were longer and listeners were asked to assess speech rate (and then hopefully match it naturally, now that it was brought to their attention). TH adults did fairly well at gauging the speech rates of recorded passages, but they didn't change their speech rates at all -- not from baseline before listening, not after listening, and not after speech rate was pointed out as relevant. All this leaves more questions than answers about whether speech rate-matching is possible in the lab. Maybe people need to be in dynamic conversation for the mirroring effect to work. Maybe the dysarthria study worked because the speakers (who had Parkinson's) sounded older and possibly sick, making listeners conscious that they should act differently. Maybe the CI users in our studies didn't sound unusual, or those who did were too unintelligible to adapt to. These questions will take more complicated study. In the meantime, we're looking at a curious result: listeners rated the least intelligible CI users as talking faster, even though they spoke with a typical rate. Maybe understanding their speech took extra effort and cognitive resources, which made the speech rate feel faster than it was. See related news here. Papers & Presentations
People & Related Work
Collaborators:
Related work:
Student Corner
Data and materials:
Comments are closed.
|