L2 Verb Use in English Verb‐Argument Constructions: A Network Science Approach

ABSTRACT Motivated by the long‐standing goal in cognitive science to quantify the organization of semantic memory, this study examined the knowledge structure of verb‐argument constructions (VACs) in second language (L2) learners by integrating the verb generation task with computational methods grounded in network science. Two hundred and six university students (CEFR B2+) and high‐school students (CEFR B1–) completed a verb generation task using 20 frequently‐occurring VAC frames (e.g., V+about ) as prompts. Results showed that both groups’ networks exhibited small‐world properties. However, university students tended to form VAC networks with more efficient connectivity and stronger local clusters compared to high‐school students. Additionally, high‐school students’ networks contained significantly more unattested verbs. Both groups’ network structures exhibited a significant correlation with VAC‐verb contingency. These findings support a usage‐based perspective on L2 development, suggesting that L2 learners’ VAC networks gradually become broader and more interconnected as they are shaped by L2 proficiency. This study offers the first systematic application of network analysis to characterize the development of VAC representations in L2 learners, thereby proposing an empirically testable framework for modeling the semantic organization of VACs.