Hearing what is being said: The distributed neural substrate for early speech interpretation

Abstract

AbstractHuman speech comprehension is remarkable for the immediacy with which the listener hears what is being said. Here, we focus on the neural underpinnings of this process in isolated spoken words, to define the intrinsic properties of these core mechanisms independent of contextual constraints or predictions. We analysed source-localised MEG data for a set of 218 spoken nouns heard in isolation, using Representational Similarity Analysis to probe the spatiotemporal coordinates of processes sensitive to speech phonology, lexical form, and the semantics of emerging word candidates. Phonological model fit was detectable within 40-50 ms from onset, rapidly engaging a bilateral network including superior and middle temporal cortex and extending into anterior temporal and inferior parietal regions. Model fit to lexical form emerged within 60-70 ms, and model fit to cohort semantics from 100-110 ms, both overlapping strongly with the spatial pattern of phonological model fit in bilateral temporo-parietal cortices. Strikingly, the majority of vertices in a central core of brain regions showed significant model fit to all three dimensions tested, consistent with a fully distributed neural substrate for early speech analysis. The early interpretation of speech seems to be conducted in a unified integrative representational space, in conflict with the conventional view that speech analysis is underpinned by a linguistically stratified representational hierarchy.