This article proposes a methodology to create a multimodal corpus that can be shared with a group of researchers in order to analyze synchronous online pedagogical interactions. Epistemological aspects involved in studying online interactions from a multimodal and semiotic perspective are addressed. Then, issues and challenges raised by corpus creation and sharing are examined with a particular focus on ethics. Basing my discussion and analysis around a particular research project, the steps involved in the creation of a multimodal and shareable corpus are described and the scientific benefits of a collective exploration of data and their subsequent multimodal representations are discussed and illustrated. It is finally claimed that corpus sharing can contribute vastly to the field of computer-assisted language learning by enhancing its scientific robustness as it favors a multidisciplinary, systematic, and in-depth analysis of multimodal data.