VietTESOL International Convention 2026

Vietnam's national plan to reposition English as a second language (2025–2035) places significant expectations on learner autonomy and technology integration, with AI tools promoted as vehicles for extending speaking practice beyond the classroom. Yet whether AI‑supported self‑regulation actually reduces the affective barriers constraining Vietnamese learners' oral performance remains uncertain. This quantitative cross‑sectional study examined the associations among metacognitive awareness of speaking strategies (MASQ), self‑regulated learning with AI tools (SRL‑AI), and public speaking anxiety (PSCAS) among 230 English‑major undergraduates at a Vietnamese university. Three validated instruments were administered, and data were analyzed using Pearson correlations and bootstrapped mediation analysis. Participants reported frequent metacognitive strategy use and active AI engagement, yet speaking anxiety remained moderately high. Metacognitive awareness showed no direct association with anxiety but was positively linked to AI‑supported self‑regulation, which was in turn associated with higher anxiety. Mediation analysis revealed a significant positive indirect effect, suggesting that strategic learners' intensive AI use may amplify rather than alleviate evaluative pressure. These findings carry direct implications for the EFL‑to‑ESL transition: if technology integration is to serve Vietnam's repositioning agenda, pedagogical designs must attend to the learner effect rather than assume that increased AI‑mediated practice automatically reduces anxiety. Combining metacognitive training with explicit anxiety management may be essential for realizing AI's full pedagogical potential in sustainable English language development.