The neural mechanism underlying statistical learning’s contribution to L2 grammar learning
Yao Chen, Li Li, Jing Yang, Mengxing Wang, Ruimin Wang, Ruimin Wang · Brain and Language · 2026
Grammar sets human language apart from animal communication. While first language (L1) grammar relies on language-specific mechanisms, second language (L2) grammar predominantly engages domain-general mechanisms, such as statistical learning (SL), which helps detect input patterns. However, there is limited empirical evidence on how SL contributes to L2 grammar learning and its neural underpinnings. This study employed fMRI protocol of SL task and artificial grammar learning task to explore the role of SL in L2 grammar learning, examining its distinction between auditory and visual SL, and how its contribution varies across different learning stages, particularly the first day and last day of learnings. We found that auditory SL initially engages the left postcentral gyrus, facilitating the bottom-up extraction of structural regularities. As learning progresses to proficiency, this reliance transitions to the left precentral gyrus, accompanied by reduced involvement of right parietal attentional regions. This neural shift reflects a progression from perceptual processing toward automated motor representation and more efficient retrieval of grammatical patterns. These results provide neurobiological evidence for the role of domain-general SL in L2 grammar learning, underscoring its importance over innate grammatical knowledge and advancing a more nuanced understanding of the learning process.