The aim of our research is to describe and to analyze the process of mutual comprehension between a distant learner and a teacher, engaged in a computer-mediated didactic interaction. How does the multimodal mediation in this kind of interaction appear ? How the actors of the interaction, to be understood, combine the modalities of the knowledge aimed by the learning and the linguistic modalities classically studied in conversational analysis ? How to transcribe a multimodal filmed corpus to analyze the influence of the multimodal mediation on the learning process? We invest these questions here with a constructivist paradigm of situated linguistic interaction.