The present study explores the educational use of metaverse for middle school students’ English speaking activities. As an online learning space, we chose Gather.town, one of the most popular metaverse platforms used in Korea, to investigate student perceptions of their learning experience and examine how they interacted with the teachers and peers on metaverse by analyzing the screen recording of the activities. The results showed that the students (N=36) were highly satisfied with the experience and actively initiated opportunities for interaction during the English speaking sessions. Several factors may have contributed to their active engagement during the sessions. First, the affordances of metaverse enabled the learners to have control over his/her own avatar and gave them the freedom to choose who to talk to and when to approach them. Second, providing them with less controlled activities seemed to have positively affected their levels of interest and motivation for learning English speaking. Third, designing speaking tasks that make them move around different locations on the metaverse made learning more dynamic and engaging. Pedagogical implications are discussed based on the findings of the study.