Reducing listening anxiety in blended learning: The role of AI-driven feedback and strategy in EFL education

This mixed-methods study examined listening strategies, anxiety, and achievement among 436 vocational college EFL students in an AI-augmented blended learning program. Quantitative analysis revealed that strategies positively predicted achievement whereas anxiety negatively predicted achievement. Notably, higher anxiety correlated with increased strategy use, suggesting compensatory engagement. Mediation analysis confirmed that strategies significantly partially mediated the anxiety-achievement relationship. Qualitative findings from 15 semi-structured interviews documented adaptive AI tool use (Duolingo, FIF) and identified novel anxiety triggers specific to blended contexts: environmental constraints (ambient noise, absence of visual cues) and technological barriers (poor audio quality, connectivity issues). These findings establish listening strategies as dual-purpose cognitive-affective resources that simultaneously enhance proficiency and mitigate anxiety in technology-rich environments. The study extends Information Processing Theory, Metacognitive Theory, and the Affective Filter Hypothesis by demonstrating how strategic competence dynamically mediates affective-cognitive interactions in AI-mediated vocational settings, advocating for integrated pedagogical designs combining explicit strategy instruction with equitable technological implementation. Practically, the findings advocate for integrated pedagogical designs that systematically combine explicit strategy instruction, affective awareness training, and equitable, pedagogically aligned technological implementation-particularly for learners in vocational education, where pragmatic language outcomes and digital access disparities demand context-responsive solutions.