Modelling and mechanism analysis of English vocabulary associative learning process of L2 learners

Although the associative learning strategy (ALS) is widely employed in vocabulary teaching, its mechanisms have rarely been quantified from a complex dynamic systems perspective. This study addresses this gap by constructing a word association corpus from Chinese university students and developing a stepwise diffusion modelling framework to simulate vocabulary network formation. Three computational diffusion strategies are proposed and compared through 750 computational simulations: RADS (random association diffusion strategy), MDDS (maximum degree diffusion strategy), and IDASDS (integration of degree and association strength diffusion strategy). Quantitative analysis reveals that IDASDS achieves the fastest network convergence and produces stronger semantic connectivity in terms of clustering coefficient and path length. The initial word count significantly affects the size of the first fully connected network but has minimal impact on ultimate convergence steps, confirming the emergent nature of associative learning. These findings provide quantifiable principles for vocabulary instruction: prioritising high-connectivity hub words and optimising initial vocabulary load can enhance learning efficiency by leveraging the small-world properties of lexical networks.