Influence of Self-Regulated Learning Strategies on L2 Writing Performance in AI-Assisted Writing: The Serial Mediating Roles of Writing Self-Efficacy and Writing Engagement

Abstract

: Generative AI tools are increasingly integrated into EFL writing instruction. However, within AI-assisted writing contexts, the relationship between self-regulated learning (SRL) strategies and L2 writing performance, and its underlying mechanisms, has received limited empirical attention. To address this gap, the present quantitative study, grounded in social cognitive theory, investigated the relationship between SRL strategies and L2 writing performance and tested whether writing self-efficacy and writing engagement mediated it. A sample of 304 second-year Business English majors who used AI tools in their English writing course participated in the study. Structural equation modeling (SEM) analysis indicated that writing self-efficacy and writing engagement mediated the relationship between SRL strategies and L2 writing performance. Specifically, neither the direct effect of SRL strategies on L2 writing performance nor the indirect effect through writing self-efficacy alone was significant. The indirect effect through writing engagement was marginally significant. Writing self-efficacy and writing engagement had a serial mediating effect on the relationship between SRL strategies and L2 writing performance. The study also discusses the practical implications of these findings and its limitations.