UNIVERSITY STUDENTS’ PERCEPTIONS OF GENERATIVE AI AS A PARTICIPANT IN ENGLISH CLASSROOM INTERACTION: A MIXED-METHODS STUDY AT A VIETNAMESE UNIVERSITY

Abstract

Generative Artificial Intelligence (GenAI) has entered university English as a Foreign Language (EFL) settings at remarkable speed, yet how learners perceive it as a participant in classroom interaction, alongside their teachers and peers, remains underexplored, particularly in Vietnamese higher education. Adopting a sequential explanatory mixed-methods design, this study surveyed 50 English-major undergraduates at a public university in Ho Chi Minh City, Vietnam, with a bilingual instrument of 29 Likert items and two open-ended prompts, and then interviewed four focal students to interpret the survey patterns. Descriptive statistics, reliability analysis, non-parametric comparisons, a Spearman correlation, and qualitative content analysis were used to characterise students’ patterns of GenAI use and their construal of GenAI as an interactional partner. Students used GenAI most for explaining grammar and vocabulary, generating ideas, and obtaining writing feedback, and least for speaking practice. They construed it primarily as a preparatory tutor and peer that scaffolds what they will say around class, reporting heightened willingness to communicate; yet they did not trust its English over their teacher’s, ranking their trust as teacher over AI over peers. Frequency of use was moderately associated with perceiving GenAI as an interactional participant. Both strands converged on one tension: the preparatory function that supported interaction also constrained it when GenAI produced over-complex language or fostered dependency that slowed real-time turn-taking. The study offers early Vietnam-based evidence that learners position GenAI as a backstage rehearsal partner rather than a replacement interlocutor, and points to a pedagogical need for prompt literacy and register-appropriate editing.