Jeonghwa Cho (PhD 2024) tests the temporal dynamics and spatial stability of neural representations for lexical items and grammatical features by applying decoding methods to EEG signals recorded from bilingual Korean/English speakers. Take-aways include: (1) replicates stable decoding of both lexical items in two languages, (2) first demonstration of decoding of multiple grammatical features within a given language, but also (3) evidence for cross-language decoding is extremely narrow. Read on for more!
Cho, J., & Brennan, J. R. (2025). Decoding of lexical items and grammatical features in EEG: A cross-linguistic study. Neuropsychologia, 213, 109150.
https://doi.org/10.1016/j.neuropsychologia.2025.109150
Abstract
Diverse evidence supports the theory that bilingual language users have language-invariant representations of concepts and grammatical forms such as argument structure. Here we extend that work to test the representation of morphosyntactic features and lexical concepts in typologically different languages. Specifically, we deploy machine learning techniques with EEG data collected from eighteen Korean-English bilinguals while they read singular and plural nouns and present and past tense verbs in English and Korean. Whereas event-related potentials (ERPs) analyses show limited sensitivity to discriminate lexical, number, and tense information, neural decoding revealed robust within-language classification of lexical and morphosyntactic information in both languages. In contrast, between-languages decoding was possible only for number information; decoding of lexical items and tense did not generalize between the two languages, even when accounting for temporal differences. These results indicate stable within-language EEG representations for lexical items and morphosyntactic features but suggest that only the number feature show evidence for shared EEG response patterns between the two languages studied.