Fostering multidimensional learning engagement through generative AI in Chinese L2 writing: A quasi-experimental investigation grounded in self-determination theory

Abstract The expanding availability of generative artificial intelligence (GenAI) tools, most notably ChatGPT, has created new opportunities for second language (L2) writing pedagogy. However, empirical evidence on how these tools influence different facets of learner engagement remains limited, particularly within Chinese as a second language (CSL) settings. Drawing on self-determination theory (SDT; Ryan & Deci, 2000), this quasi-experimental study involved 82 CSL learners at HSK Level 4 proficiency who participated in either a treatment condition (n = 41) incorporating teacher-facilitated ChatGPT writing workshops or a comparison condition (n = 41) following conventional instruction. Engagement was operationalized along three dimensions: behavioral, cognitive, and emotional engagement. Mixed ANOVAs revealed significant Group × Time interactions for behavioral engagement (F = 4.35, p = .040, ηp² = .052), cognitive engagement (F = 4.33, p = .041, ηp² = .051), and emotional engagement (F = 11.63, p = .001, ηp² = .127). Simple-effect analyses indicated that behavioral engagement increased significantly in the treatment group, whereas cognitive and emotional engagement were maintained in the treatment group but declined, or approached decline, in the comparison group. Interpreted through the SDT lens of autonomy, competence, and relatedness, the findings suggest that teacher-mediated GenAI integration may support behavioral activation and help stabilize cognitive and emotional engagement during CSL writing, although direct within-treatment gains were not observed across all dimensions. Practical implications for constructing AI-augmented L2 writing curricula that combine technological scaffolding with learner agency are discussed.