Modeling Chinese EFL learners’ agency in AI-assisted language learning: a control-value theory perspective

Purpose This study extends the application of control-value theory (CVT) to the emergent context of AI-assisted language learning (AILL), empirically examining how AI-specific factors – namely perceived advantage and risk – influence English-as-a-foreign-language (EFL) learners' cognitive appraisals and agency.Design/methodology/approach Drawing on survey data from 718 Chinese EFL learners, the study adopts a quantitative design and utilizes partial least squares structural equation modeling (PLS-SEM) to test the proposed relationships.Findings The results indicate that (1) perceived advantage has a positive direct effect across all three dimensions of learner agency (key abilities, active actions, and mental characteristics), whereas perceived risk exhibits a negative influence; (2) perceived advantage exerts a positive influence on both control and value appraisal, while perceived risk is negatively related to these appraisals; (3) both control and value appraisal positively predict all dimensions of learner agency; and (4) control and value appraisal serve as significant mediators in the relationships between AI factors and learner agency.Originality/value The results could guide teachers and instructional designers in developing and evaluating more agentic and learner-centric AILL innovations.