AI-assisted instruction: an empirical study of an objective, observation, and organization-based SVVR approach to promote students’ English impromptu speaking performance, speaking anxiety, and cognitive network structure

The development of EFL students’ improvisational speaking skills remains a challenge in English language teaching. Limited exeperience and lack of authentic contexts often hinder their impromptu performance. To address this, immersive virtual reality (SVVR) has been introduced, yet existing SVVR-based methods face issues such as distracting materials, excessive information filtering, and difficulties in structuring speech. Advances in artificial intelligence offer new opportunities for improvement. Therefore, this study proposes an AI-supported SVVR approach, focusing on its effects on students’ impromptu speaking performance, speaking anxiety and cognitive network structure. 60 undergraduate students were recruited into the study: 30 in the experimental group using the OOO-SVVR approach and 30 in the control group using the T-SVVR approach. The results showed that students using the OOO-SVVR approach outperformed the control group regarding impromptu speaking performance and speaking anxiety. In addition, epistemic network analysis showed that the OOO-SVVR approach helped students generate more creative ideas, break through the traditional linear mode of expression, and realize more depth and breadth of speech content. The results of this study validate the effectiveness of AI and SVVR in teaching impromptu speaking and provide valuable insights and guidance for exploring more efficient English impromptu speaking teaching in the future.