Abstract

Predictive processing may differ across one’s languages – for instance, the ability to predict an upcoming noun based on a gendered determiner cue may be greater when processing in one’s first/dominant language relative a second/less dominant. We are principally interested in explicating how and to what bilinguals are generating predictions during naturalistic language processing, and are furthermore curious as to how bilinguals weigh cue information in their L1 and L2, especially during syntactic dependency parsing. Using EEG in a naturalistic listening paradigm, we plan to collect responses from ~ 60 L1 English/L2 Spanish participants to spoken English & Spanish chapters from the children’s storybook “The Little Prince.” These data will be annotated and analyzed using Natural Language Processing (NLP) with Transformer-based Large Language Models. Additionally, we will compare data within participants as (potentially) modulated by experiential factors such as exposure and usage.

Our work at UF is done in collaboration with Dr. Jonathan Brennan at the University of Michigan, whose lab is using similar materials with a L1 Mandarin/L2 English population, and Dr. Teresa Bajo at the University of Granada with a L1 Spanish/L2 English population.

Current Status

We are in the midst of data collection at UF and are currently working to develop materials for use at the University of Granada.

Funding

This work is supported by NSF BCS-2518248.

External Links

Forthcoming!