In a classroom, obtaining active, whole-focused, and engaging learning results from a design is often difficult. In this study, we propose a self-observation model that employs an instinctive interface for classroom active learning. Students can communicate with virtual avatars in the vertical screen and can react naturally according to the situation and tasks. Students can also immediately observe themselves mixed with a virtual environment and therefore reflect on the necessary improvements. With the designed system, potential advantages such as motivation, enjoyment, context for situated learning, engagement, social collaboration, and role-playing might arise. To examine the idea, we conducted a case study with 60 fourth grade elementary students to investigate learners' behavior as performers and peer audience and their perception of body movements and speech commands in the designed learning environment. The results show that the students' image in the vertical screen affected the students and the peer audience positively. Moreover, they positively perceived their competence and their enjoyment after they performed contextual learning activities through the body movement interface