Abstract: After more than four decades of development, video-based analysis has become a major approach in educational research, in particular analysis of learning processes. Yet, with regard to the "why" and the "how-to," video researchers have not achieved a consensus. Based on a brief review of history and relevant theories, this article uses two case studies to illustrate three crucial aspects of video-based analysis:the individual body, the social bodies, and the bodies-things interaction. We argue that these aspects profoundly shape our analyses and understandings of classroom interaction, learning processes, and mechanism of teaching, and that video data play a critical role therein. Finally, we argue that educators and educational researchers need to avoid the bias of logocentrism, to empower teachers to evaluate their own classroom, and to respect the everyday workings of classroom life-worlds, if we are to fully appreciate the potential of video-based analysis.

Key words: classroom video, classroom discourse, classroom interaction, video-based analysis, interaction analysis