Skip to main content
Apply Give
Menu

Psycho Meeting - Sebastián Mancha / Syntactic boundaries and belief update in naturalistic MEG

Two young men, both wearing blue jeans and a grey sweater, standing next to each other in front of a book case.

Psycho Meeting - Sebastián Mancha / Syntactic boundaries and belief update in naturalistic MEG

Linguistics Tuesday, October 27, 2026 3:30 pm - 4:45 pm Marie Mount Hall, 1108B

Tuesday October 27, Sebastián leads the Psycholinguistics Meeting with discussion of his 896 for Philip and Ellen, described below.

In this talk I'll present a proposal for my 896, working with Philip and Ellen. I propose to use measurements of syntactic belief updating to examine how the brain deals with the need to predict, verify, and integrate syntactic boundaries during naturalistic listening. One measurement of interest here is entropy reduction, the degree to which the parser's belief about the underlying structure of a sentence changes with new lexical input (consider how clarifying fell is in the horse raced past the barn fell). These measurements will be derived from a Recurrent Neural Network Grammar, a symbolic neural network that jointly models language strings and structures. I'll be discussing clause and sentence wrap-up effects, prosodic boundaries, information theoretic measures, etc. I'll also be looking for options on some alternative models of difficulty at major syntactic boundaries.

Add to Calendar 10/27/26 15:30:00 10/27/26 16:45:00 America/New_York Psycho Meeting - Sebastián Mancha / Syntactic boundaries and belief update in naturalistic MEG

Tuesday October 27, Sebastián leads the Psycholinguistics Meeting with discussion of his 896 for Philip and Ellen, described below.

In this talk I'll present a proposal for my 896, working with Philip and Ellen. I propose to use measurements of syntactic belief updating to examine how the brain deals with the need to predict, verify, and integrate syntactic boundaries during naturalistic listening. One measurement of interest here is entropy reduction, the degree to which the parser's belief about the underlying structure of a sentence changes with new lexical input (consider how clarifying fell is in the horse raced past the barn fell). These measurements will be derived from a Recurrent Neural Network Grammar, a symbolic neural network that jointly models language strings and structures. I'll be discussing clause and sentence wrap-up effects, prosodic boundaries, information theoretic measures, etc. I'll also be looking for options on some alternative models of difficulty at major syntactic boundaries.

 Marie Mount Hall false