Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural Mapping of Language Knowledge Representations in Phrases
Poster Session A - Sandbox Series, Thursday, October 24, 10:00 - 11:30 am, Great Hall 3 and 4
This poster is part of the Sandbox Series.
Feizhen Cao1,2, Nan Li1,3, Wenjia Zhang4, Suiping Wang1,2; 1Philosophy and Social Science Laboratory of Reading and Development in Children and Adolescents (South China Normal University), Ministry of Education, Guangzhou, China, 2School of Psychology, South China Normal University, Guangzhou, Guangdong, China, 3School of Foreign Studies, South China Normal University, Guangzhou, China, 4Key Laboratory for Artificial Intelligence and Cognitive Neuroscience of Language, Xi’an International Studies University, Xi’an, China
The core of the human language function architecture involves knowledge representations, including comprehension of individual words/constructions/expressions and the combinatorial processes for forming compound words, phrases, and sentences. Investigating the neural correlates of these knowledge representations has long been a major challenge in the field of the neurobiology of language. In this study, we used functional magnetic resonance imaging (fMRI) to record brain activity of participants while they read Chinese phrases composed of two consecutive two-character words. Participants were asked to perform semantic judgment and composition relations judgment tasks on these phrases, with approximately one week between the tasks. We combined representational similarity analysis (RSA) techniques and a Bayesian hierarchical partial pooling model to explore the mapping of knowledge representations related to words and phrases in the human brain. Additionally, we used computational linguistics methods to calculate semantic vectors and attempted to quantify semantics using multiple models. This approach allowed us to further investigate the neural mapping of word- and phrase-level knowledge representations and their semantics. The results indicate that the neural representation strength of word grammatical classes and phrase composition relations in the brain can be flexibly adjusted according to different tasks. Under the composition relations judgment task, robust representations of word grammatical classes and phrase composition relations were observed. Specifically, at the word-level, only the grammatical classes of the second two-character word in a phrase could be represented in specific brain regions, whereas no such regions were found for the grammatical classes of the first two-character word. Similar results were observed during the semantic judgment task only using ROI analysis. Thus, evidence for the brain representation of the second two-character word's grammatical class aligns with the combinatorial view, which posits that word grammatical class information is represented only when words are combined into larger linguistic units. At the phrase-level, in the explicit composition relations judgment task, the representation of compositional relations involved the superior parietal lobule and extended to the angular gyrus. Similar results were observed in the semantic judgment task only when ROI analysis was used. Incorporating semantic vectors, the analysis revealed that even when we further controlled for semantic effects, evidence for the brain representation of both phrase composition relations and word grammatical classes was still observable. This indicates that these knowledge representations are mapped in the brain at least partially independent of semantics. Finally, results of the effective connectivity analysis indicated that brain regions representing composition relations receive information from regions representing word grammatical classes, and simultaneously influence the processing of word class information. Overall, these results partially reveal how the human brain supports language processing through the dynamic interaction among multiple brain regions.
Topic Areas: Syntax and Combinatorial Semantics, Meaning: Lexical Semantics