To explore knowledge co-construction patterns and learning motivation within virtual EFL co-creation environments, this study examined behavioral patterns and motivation in three different co-creation environments (paper-based, 2D digital, and 3D VR co-creation) through sequential behavioral analysis and ANCOVA. The study utilized a quasi-experimental research design with a total of 66 tenth-grade students from two English classes at a public senior high school in northern Taiwan. Based on the visualized behavior transition diagrams, the task-switching behaviors between dissonance identification and knowledge negotiation as well as the isolated behaviors of applying newly-constructed knowledge are the core of knowledge co-construction. Particularly, 3D VR co-creation was characterized by the highest number of higher level isolated acts and lower level circular continuity, both of which reflect VR co-creators' efforts to gain familiarity with advanced technology as well as the intention to exchange information and reach community consensus to overcome task complexity, form community consensus, and lower anxiety. Such behavioral patterns echoed the results of ANCOVA on intrinsic and extrinsic motivation; that is, on either intrinsic or extrinsic motivation, the influence of 3D VR co-creation was the greatest, followed by that of 2D digital co-creation and paper-based cocreation. For future co-creation instruction and research, it is suggested that the instruction of VR co-creation be invested with abundant time to allow mature higher level knowledge co-construction dialogues to occur. Moreover, to gain an even deeper understanding of the social structure embedded in knowledge co-creation, it is suggested that social network analysis (SNA) be employed in future research.