AI-driven adaptive feedback and EFL writing performance: The roles of engagement, metacognition, and epistemic agency in a cross-linguistic context.

This study investigates the role of AI-driven adaptive feedback systems in enhancing feedback engagement, metacognitive strategies, epistemic agency, and writing performance among Saudi EFL learners. Adopting a cross-linguistic perspective, this study examined the impact of feedback provided in both Arabic (L1) and English as a Foreign Language (EFL) contexts. A quantitative predictive research design was employed utilising Structural Equation Modelling to test the hypothesised relationships among the study variables. The sample comprised 1049 Saudi university EFL learners from various institutions. Five instruments were developed and validated for this study: the Feedback Engagement Scale, Metacognitive Strategy Scale, Epistemic Agency Scale, Adaptive Feedback Scale, and the writing task. The findings revealed that AI-driven feedback in the L1 significantly predicted feedback engagement, metacognitive strategies, epistemic agency, and EFL writing performance. Conversely, AI-driven feedback in EFL significantly predicted metacognitive strategies, epistemic agency, and EFL writing performance but not feedback engagement. Furthermore, feedback engagement, metacognitive strategies, and epistemic agency were significant predictors of EFL writing performance. English language proficiency moderated the relationship between AI-driven feedback in EFL and writing performance, but not between AI-driven feedback in L1 and writing performance. These findings reveal the importance of integrating L1 and EFL feedback to address diverse learner needs and support skill development across proficiency levels. This study contributes to the growing body of literature on AI-enhanced language education. It offers valuable insights for teachers, policymakers, and technology developers seeking to optimise EFL instruction in the Saudi context and beyond.