Search Papers | Poster Sessions | All Posters
Poster B126 in Poster Session B - Thursday, August 8, 2024, 1:30 – 3:30 pm, Johnson Ice Rink
Correlates of Syntax in EEG Responses to Naturalistic Speech
Ole Bialas1 (), Sophea Biswas1, Edmund Lalor1; 1University of Rochester
Research on the cortical processing of continuous naturalistic speech has shown that the brain represents speech on different levels of abstraction. Speech is represented as a continuous acoustic signal, in discrete phoneme categories or as semantic information on the level of words. However, it remains unclear if and how the brain represents the syntax to combine words into larger meaningful structures. Here, we use multivariable models to predict the EEG recordings of participants listening to naturalistic speech. We show that syntactic features, derived from automated dependency parsing, improve the prediction accuracy of a baseline model using acoustic and semantic speech features. While our findings suggests that correlates of syntax can be found in EEG responses to continuous speech, they also highlight the need for further research to disentangle syntax from possible correlated acoustic speech cues.
Keywords: EEG naturalistic speech syntax temporal response function