Effects of AI-generated feedback on self-regulated learning (SRL) strategies among xenoglossophobic English learners

This explanatory sequential mixed-methods study investigated the effects of an AI-supported intervention incorporating AI-generated feedback on self-regulated learning (SRL) strategies among 68 xenoglossophobic upper-intermediate English learners. Using a randomised pre-test–posttest control-group design, participants were assigned to an experimental group (n = 34), receiving AI-generated feedback supplemented by reflective prompts, progress dashboards, motivational badges, and human-in-the-loop (HITL) instructional support, or a control group (n = 34), receiving equivalent instructor feedback and materials across 36 sessions over 12 weeks. ANCOVA revealed higher posttest SRL scores for the experimental group (ηp² = .22, p < .001), with greater metacognitive and meta-affective gains. Follow-up focus groups provided insight into these findings, indicating that learners perceived AI feedback as personalised, immediate, structured, and non-judgemental, which they associated with enhanced autonomy, confidence, and reflective learning. Findings suggest that AI-supported instruction may enhance metacognitive and meta-affective self-regulation among xenoglossophobic English learners.