Generalized mathematics self-efficacy and student engagement: the role of self-concept and mathematics anxiety

IntroductionThe development of artificial intelligence (AI) has created new opportunities for AI-supported foreign language teaching and applications. This study investigates the interrelationships among foreign language enjoyment (FLE), writing self-efficacy, self-regulated learning (SRL) strategies, and learner engagement within the context of AI-assisted English as a foreign language (EFL) writing.MethodsA cross-sectional survey design was employed, involving 535 Chinese university students with prior experience in AI-assisted writing. Participants completed adapted and validated scales measuring FLE, writing self-efficacy, SRL strategies, and learner engagement. Structural equation modeling (SEM) was used to test a hypothesized partial mediation model linking these constructs.ResultsThe results supported a well-fitting partial mediation model. FLE was found to be a direct and significant positive factor of learner engagement. Furthermore, the analysis confirmed that self-efficacy and SRL strategies act as sequential mediators in the relationship between FLE and engagement. Specifically, FLE is positively associated with learners' self-efficacy, which in turn promotes the use of SRL strategies, ultimately leading to deeper cognitive, behavioral, emotional, and agentic engagement in the writing task.DiscussionThe study extends Control-Value Theory, Broaden-and-Build Theory, Social Cognitive Theory, and Self-Regulated Learning Theory to the novel domain of AI-assisted language learning, highlighting the critical affective and cognitive pathways that foster engaged learning. Pedagogical implications are discussed, emphasizing the importance of designing enjoyable, confidence-building, and strategy-rich AI-assisted writing environments.