Neurophysiological correlates of producing and perceiving natural connected speech

Spoken language is an essential part of our every-day lives. Despite being one of the most prominent human behaviors, little is known about the cortical substrates supporting the perception and production of continuous, connected speech. This doctoral dissertation aimed to fill this gap through an extensive experimental paradigm consisting of both natural speech production and speech perception tasks. There was a special emphasis on speech rhythm which is tied to speaking rate. Cortical signals were measured with magnetoencephalography. Electromyographic and acoustic signals were recorded during the speech production tasks. Study I aimed to quantify speech rhythm using a multimodal spectral approach in the form of coherence between electromyographic and acoustic signals. This multimodal approach was compared to a unimodal approach, which involves spectral analysis of electromyographic and acoustic signals in isolation. As opposed to a unimodal analysis, a multimodal analysis was shown to successfully reveal the shared periodic components in the two signals. This suggests that coherence metrics are especially useful in quantifying rhythm in the inherently complex natural speech. Study II investigated the cortical correlates of natural speech production and perception. Modulations in band-limited cortical signal power were examined in response to variations in three fundamental speech-related features: the amount of linguistic content present in an utterance, speaking rate and social relevance. The resulting spatiospectral patterns revealed that the right hemisphere is markedly involved in natural speech processing and, particularly, that processing of socially relevant speech engages the right temporo-parietal junction. Notably, natural speech production and perception were found to extensively overlap in both hemispheres. Study III examined, using audio-MEG coherence, the cortical tracking of global rhythmic structure and local, transient variations in speaking rate in perceived natural, connected speech. It was found that cortical tracking of perceived natural, connected speech extends beyond the previously reported emphasis on the temporal regions and also engages higher-order cortical regions. Furthermore, the observed audio-MEG coherence patterns revealed two spatially and functionally distinct components of cortical tuning: evolutionary tuning to global rhythmic structure, and predictive tuning that is driven by local changes in speaking rate. This doctoral dissertation presents novel methodological tools for characterizing rhythm in natural speech. It also contributes to an emerging view on cortical speech processing that is not confined to traditional left-hemispheric cortical regions and offers an insight into the functional role of speech rhythm in speech comprehension.