Dr Bingjiang Lyu and Professor Lorraine K. Tyler | Neural dynamics of incremental speech comprehension
Leipzig Lectures on Language
- Datum: 29.09.2021
- Uhrzeit: 15:00 - 16:30
- Vortragende(r): Dr Bingjiang Lyu and Professor Lorraine K. Tyler
- Department of Psychology, University of Cambridge, United Kingdom
- Ort: MPI für Kognitions- und Neurowissenschaften
- Raum: Zoom Meeting
- Gastgeber: Abteilung Neuropsychologie
Bingjiang Lyu is Postdoctoral Researcher at the University of Cambridge. His research interests lie in the cortical mechanisms and dynamics involved in language comprehension using EEG/MEG and fMRI. He is currently working on how the incremental processing of speech is affected by contextual information embedded in sentences. Lorraine Tyler is Professor the Department of Psychology, University of Cambridge. Her research combines cognitive models with multi-modal imaging to understand the neurobiological substrate for language functions.
Human speech comprehension is remarkable for its immediacy and rapidity. The listener interprets an incrementally delivered auditory input, millisecond by millisecond as it is heard, in terms of multilevel representations of relevant linguistic and nonlinguistic knowledge. However, the spatiotemporal dynamics underlying the complex set of neural processes subserving incremental speech comprehension are still unclear. In this talk, we will present our studies focusing on two key aspects of incremental speech comprehension, (1) meaning composition of two structurally proximate words, i.e., a verb and its direct object noun and (2) building a structured interpretation from an unfolding sentence under multifaceted probabilistic constraints. We use electro-/magnetoencephalography (E/MEG) combined with cognitive models of language functions to reveal the millisecond-by-millisecond multilayer incremental processing of spoken language. Specifically, we derived quantitative measures of both linguistic and nonlinguistic properties from computational models including artificial neural networks and tested them against spatiotemporally resolved brain activity in source-localised E/MEG. This unique combination of methods enables us to probe directly the neural representation of relevant lexical properties that are activated as each word is heard and the incremental processes through which they are evaluated and integrated into coherent interpretations, providing a plausible neural substrate for incremental speech comprehension in real time.