Acoustic and Semantic Auditory Processing of Continuous Speech: a Time Response Function Analysis for MEG Data

Speech perception is a complex process involving multilevel neurophysiological processing of various speech components. In the presented work, we used temporal response function (TRF), a novel method of processing the magnetoencephalogram (MEG) recorded while listening to continuous speech to analyse the neural response to auditory and semantic components of speech during its perception in vivo. The temporal response to dynamic changes in the sound envelope demonstrated an early neurophysiological response: from 20 ms with an amplitude peak at 100 ms, and the response to the perception of word onset had a peak latency at 120 ms. The semantic component of speech showed a later temporal response: from 200 ms with a peak latency of 300 ms bilaterally in temporal magnetometers. Thus, TRF showed a later response to semantic changes in speech than to changes in acoustic features in the MEG study.