Corpus-Based Modeling of Second Language Acquisition Trajectories: A Multidimensional Analysis of English Grammar Development Among Chinese Learners

In the context of the growing global demand for English learning, it is crucial to understand the systematic patterns of grammar acquisition by Chinese learners. Traditional cross-sectional studies and teaching methods, focused on static rules, fail to capture the dynamic, nonlinear trajectories of development and the profound influence of first language (L1) typology. This study addresses these gaps by integrating Dynamic Systems Theory with machine learning algorithms to model the developmental pathways of three core grammatical domains: articles, prepositions, and verb collocations in Chinese learners of English. Utilizing a multi-source longitudinal corpus and applying mixed-effects models alongside random forests, we uncover stage-specific transitions, demonstrate how L1 typology differentially shapes the acquisition of grammatical structures, and establish a hierarchy of difficulty among grammatical items. Methodologically, this research introduces a novel analytical framework that effectively captures the heterogeneity and nonlinearity inherent in L2 grammatical development. Theoretically, it provides empirical evidence for the nuanced role of L1 typological transfer in shaping developmental trajectories. These findings offer a scientific basis for stratified pedagogy and the design of adaptive learning systems tailored to learners’ developmental stages and L1 backgrounds.