Learning English conversation in a metaverse: The possibilities and challenges of world heritage tours

Abstract

[EN] This study explored the effects of voice quality and chatbot-based pre-task practice on learners’ foreign language anxiety (FLA) and speaking behavior in a metaverse-based English conversation environment. Thirteen Japanese university students participated in a ten-session virtual tour program featuring guided conversations conducted entirely in English. Quantitative data were collected via pre- and post-session questionnaires, while qualitative data were obtained through open-ended responses and researcher observations. Findings revealed that, although many participants initially felt anxious and lacked confidence in speaking English, repeated engagement in the metaverse reduced resistance to speaking for several learners. In particular, the clarity and tone of the instructors’ voices significantly influenced learners’ sense of psychological comfort and willingness to participate. While chatbot practice received generally positive feedback, some learners reported usability issues, and anxiety about spontaneous speech remained. Additionally, learners showed limited interest in avatar customization, and no evidence of the Proteus effect was observed. These results highlight the importance of voice design in virtual language learning and suggest that immersive environments can support anxiety reduction through repeated, low-stress interactions. Further research is needed to examine how vocal characteristics and avatar identity influence learner psychology in metaverse-based English education.