Out now in the Journal of Neuroscience, Junyuan Zhao develops an approach to decoding the category features of phrases from EEG data. He goes on to demonstrate that the neural representation of the phrase head is reactivated at phrasal boundaries and also that such reactivation is does not appear to be modulated by conceptual similarity…
Category: publications
Paper: Chi-Lin Yu on social and language processing during story-listening
Chi-Lin Yu leads a project, along with a team from the Kovelman developmental neuroscience lab, that uses fNIRS imaging to identify neural signatures of social and language processing in school-aged children during a naturalistic listening task. The results contribute to a growing body of research that is disentangling social theory-of-mind processing from other aspects of…
Paper: Jeonghwa Cho on shared morphosyntactic representations in L1/L2
In a new paper appearing in Bilingualism: Language and Cognition, Jeonghwa Cho (Michigan PhD 2024) demonstrates evidence for shared morphosyntactic representations across languages in multilinguals, but these representations are modulated by language dominance. One key innovation of this study is a focus on prefixes, unlike the majority of previous studies that focus on shared (or…
Jeonghwa Cho defends dissertation, publishes two new papers
Dr. Jeonghwa Cho successfully defended her dissertation on April 16th, titled Cross-linguistic Representation and Processing of Words,Grammatical features, and Sentences. The committee was Julie Boland, Jonathan Brennan (chair), Shota Momma from UMass, and Savi Namboodiripad. Shortly after the defense, Jeonghwa was on the road giving talks at the Central European University in Vienna, then on…
Paper: Neural decoding + Transformer + minimal pairs = new syntactic probing approach for LLMs
MA student Linyang He leads a team that advances probing methods for large language models by combining a linear decoder with the BLiMP large-scale benchmark of linguistic minimal pairs. The result is a probing method that isolate patterns of layer-wise activation that are sensitive to distinct linguistic phenomena. There is a lot to unpack in…
Paper: Tung on prediction and memory retrieval
Tzu-Yun Tung publishes the first paper from her dissertation! Combining ERPs with an experimental design using NP-ellipsis in Mandarin, this work tests how interference effects in memory retrieval can be ameliorated by predictability. The particular pattern of amelioration – most evident in cases of ungrammatical “facilitatory interference” – are consistent with the hypothesis that predictions…
Paper: Comparing parsing strategies for a head-final language with RNNG
Yushi Sugimoto (UMich PhD 2022) and Yohei Oseki lead this paper demonstrating an advantage for left-corner parsing, as implemented with their updated RNNG, to capture neural signals while participants read Japanese newspaper text. This is the first paper, to my knowledge, testing parsing strategies in this way on a head-final language! Sugimoto, Y., Yoshida, R.,…
Paper: Parsing, CCG, and large language models
Miloš Stanojevič led this tremendous effort to test alternative approaches to structure-building that vary across grammar formalism and eagerness and, simultaneously, tease apart structure-building from next-word predictability. We are so happy to share it! Stanojević, M., Brennan, J. R., Dunagan, D., Steedman, M., & Hale, J. T. (2023). Modeling Structure-Building in the Brain With CCG…
Two preprints: separating incremental composition from predictability and localizing dependency-processing across languages
Update: Both of these papers are now in print! Stanojevic et al, in Cognitive Science and Dunagan et al. in Neurobiology of Language I’m still catching up on some of our efforts from before the new year. These include two (2) pre-prints of papers that make use of the Little Prince datasets. First up: Miloš…
Paper and data: fMRI datasets and linguistic annotations from naturalistic listening in English, Mandarin, and French
Jixing Li let a large team to prepare these unique neurolinguistic datasets. Speakers of English (49), Mandarin (35) or French (28) listened to a 1.5 h audiobook of The Little Prince during fMRI scanning. The full MRI datasets are released along-side numerous linguistic annotations including prosodic contours, GloVe embeddings, POS tags, constituency parses, dependency parses,…