AI-Mediated Revision Practices in Arabic-Language Learning: Learner Engagement, Uncertainty and Meaning Negotiation

This qualitative case study examined how eight undergraduate learners of Arabic as a foreign language engaged with AI-mediated feedback across three revision sessions. Data comprised learner–AI interaction transcripts, written reflections and successive text revisions. Analysis focused on observable practices including correction-seeking, comparison of alternatives, delayed decision-making, selective uptake and meaning negotiation. Findings showed variable engagement across tasks: correction-oriented exchanges were more visible in sentence-level revision, whereas comparison, negotiation and delayed decisions appeared more frequently in paragraph-level tasks. However, these patterns were not uniform, and immediate acceptance, superficial revision and unresolved uncertainty remained evident. The study contributes process-level evidence from an underrepresented Arabic-language context, showing how AI feedback can function as a resource for revision-related evaluation rather than for simply automated correction while highlighting the importance of preserving learners’ responsibility for linguistic decisions.