The use of generative AI tools by EFL university students for L2 oral presentations: a mixed methods study

The introduction of generative artificial intelligence into language education has presented new potential for supporting learners in second language oral presentations. While existing research has widely examined generative artificial intelligence use in second language writing and speaking, little is known about how university students autonomously use these tools across different stages of oral presentation. This mixed methods study reconceptualises the second language oral presentation as a multi-stage, process-oriented practice encompassing planning, practising, delivering and monitoring. It investigates the use of generative artificial intelligence by 14 English and foreign language undergraduates at FPT University, Vietnam, across these four pedagogical stages. Findings from semi-structured interviews and single Likert-type items indicate that students primarily used generative artificial intelligence, especially ChatGPT, for planning and monitoring, with greater emphasis on generating ideas and content than on language support. Because most widely used tools (for example, ChatGPT and Gemini) are text-based, students seldom use generative artificial intelligence for real-time speaking enhancement in practising and delivering. Although they actively evaluated and refined artificial intelligence outputs in planning, monitoring and questions and answers, they expressed concerns about over-reliance on artificial intelligence, reduced user confidence, and the authenticity and credibility of artificial intelligence-generated content, as well as cultural and contextual mismatches. The article proposes a pedagogical framework for integrating generative artificial intelligence into oral presentations and highlights the need for digital literacy training, pedagogical guidelines and ethical regulations.