[Paper Review] Multifaceted neural representation of words in naturalistic language

Xuan Yang, Chuanji Gao+3 more|arXiv (Cornell University)|Jan 19, 2026

Neurobiology of Language and Bilingualism0 citations

TL;DR

The paper integrates large-scale psycholinguistic modeling with naturalistic fMRI to uncover eight latent word-property dimensions and their distributed cortical representations during narrative comprehension.

ABSTRACT

Understanding how the brain represents the multifaceted properties of words in context is essential for explaining the neural architecture of human language. Here, we combine large-scale psycholinguistic modeling with naturalistic fMRI to uncover the latent structure of word properties and their neural representations during narrative comprehension. By analyzing 106 psycholinguistic variables across 13,850 English words, we identified eight interpretable latent dimensions spanning lexical usage, word form, phonology orthography mapping, sublexical regularity, and semantic organization. These factors robustly predicted behavioral performance across lexical decision, naming, recognition, and semantic judgment tasks, demonstrating their cognitive relevance. Parcel-based and multivariate fMRI analyses of narrative listening revealed that these latent dimensions are encoded in overlapping yet functionally differentiated cortical systems. Multidimensional scaling and hierarchical clustering analyses further identified four interacting subsystems supporting sensorimotor grounding, controlled semantic retrieval, resolution of lexical competition, and contextual episodic integration. Together, these findings provide a unified neurocognitive framework linking fundamental lexical psycholinguistic dimensions to distributed cortical systems engaged during naturalistic language comprehension.

Motivation & Objective

• Identify latent dimensions underlying word properties across a large set of psycholinguistic variables.
• Link these latent dimensions to behavioral performance on lexical tasks.
• Map the neural representations of these dimensions during naturalistic language comprehension.

Proposed method

• Analyze 106 psycholinguistic variables across 13,850 English words to extract latent dimensions.
• Use parcel-based and multivariate fMRI analyses during narrative listening to relate latent factors to brain activity.
• Apply multidimensional scaling and hierarchical clustering to identify interacting neural subsystems.

Experimental results

Research questions

• RQ1 What latent dimensions capture the multifaceted properties of words in naturalistic language?
• RQ2 How do these latent dimensions map onto distributed cortical systems during narrative comprehension?
• RQ3 What cognitive processes and neural subsystems support sensorimotor grounding, semantic retrieval, lexical competition resolution, and episodic integration in language?
• RQ4 Do the latent factors predict behavioral performance on lexical decision, naming, recognition, and semantic judgment tasks?

Key findings

• Eight interpretable latent dimensions emerge, spanning lexical usage, word form, phonology-orthography mapping, sublexical regularity, and semantic organization.
• These latent dimensions robustly predict behavioral performance on lexical decision, naming, recognition, and semantic judgment tasks.
• Neural representations of these dimensions are encoded in overlapping but functionally differentiated cortical systems during naturalistic language processing.
• Multidimensional scaling and clustering reveal four interacting subsystems supporting sensorimotor grounding, controlled semantic retrieval, lexical competition resolution, and contextual episodic integration.