Neural correlates of dependency length processing during natural language comprehension

Although extensive research has explored the neural mechanisms underlying syntactic processing, the whole-brain neural correlates of the processing of dependency length-a core quantitative syntactic feature-in the context of natural language comprehension have not yet been systematically investigated. To address this gap, we proposed a two-stage progressive syntactic parsing framework and applied it to functional magnetic resonance imaging (fMRI) data from subjects listening to natural narratives. Results showed: 1) The neural correlates of dependency length processing are robustly and consistently identified across individuals; 2) The sensitive voxels were widely distributed in the frontal-temporal-parietal language network, indicating dependency length processing relies on multi-region functional collaboration; 3) Neural activation similarity between specific syntactic structures and dependency length differed significantly, possibly due to variations in dependency length variability and functional necessity of the integration across structures. Additionally, the identified functional regions align with the dependency length minimization hypothesis, supporting its neurobiological plausibility. This study advances the understanding of syntactic processing to the sub-property level, providing a reliable method for localizing the neural correlates of fine-grained syntactic processing, revealing dependency length-syntactic structure neural associations, and offering key evidence for elucidating the neural mechanisms underlying abstract syntactic processing.