Teacher Professional Reasoning for AI-Mediated Language Teaching Under Multi-Level AI Governance: An Embedded Multiple-Case Study

Abstract

Generative artificial intelligence is reshaping teaching, learning, assessment, and professional practice in English as a Foreign Language. However, little is known about how university teachers exercise professional reasoning within emerging AI governance environments. This embedded qualitative multiple-case study explored how university teachers of English as a Foreign Language in Vietnam reasoned professionally about generative artificial intelligence. The study combined semi-structured interviews with 13 teachers from 13 universities and an analysis of 88 policy and governance documents comprising 10 national, 7 ministerial, and 71 institutional documents. A documented search, screening, inventory, and staged coding procedure enabled the systematic identification, classification, and analysis of the document corpus. Interview and documentary data were analysed separately, first within cases and governance levels and then across cases using matrices and joint displays. Findings showed that teachers' professional reasoning was developed through continual negotiation between classroom realities and multi-level AI governance. Participants moved beyond their traditional roles as language experts to become learning designers, pedagogical facilitators, critical mentors, and ethical guides. They set context-appropriate boundaries for AI use, reframed ethics in terms of transparency and cognitive engagement, and adapted assessment to preserve evidence of student learning. The documents revealed a layered but uneven governance landscape. National strategies articulated broad directions; ministerial documents provided partial guidance; and institutional policies varied considerably. The integration of the two datasets showed that AI literacy, governance, assessment, ethics, and teacher agency were interconnected with teachers’ professional reasoning. The study proposed the governance-informed Identity-Agency-Ethics-Assessment Practices heuristic for understanding AI-mediated teacher professionalism.