Acoustic and linguistic processing of speech-specific temporal structure

Tobias Overath

Research Seminar

When?

Tuesday, 17 July 2018, 13:00

Abstract

Speech perception entails the mapping of the acoustic waveform to its linguistic representation. For this transformation to succeed, the speech signal needs to be tracked across a large temporal range at high temporal precision in order to decode linguistic units ranging from phonemes (10s of milliseconds) to sentences (seconds). The nature of this analysis - where and how temporal speech properties are processed in the human auditory system, where and how they interface with linguistic representations - is still not fully understood. We have recently developed a novel ‘speech quilting’ algorithm (Overath et al., 2015) that controls the temporal structure of speech at various temporal scales (30 to 960 ms), and have shown (1) that superior temporal sulcus is sensitive to the analysis of such acoustic speech structure, and 2) that this is independent of linguistic analysis (speech stimuli were in a foreign language). I will present new fMRI and EEG data from our lab that explicitly allows the dissociation of acoustic and linguistic analyses by comparing the same acoustic manipulation (speech quilting) in two languages (familiar, foreign).