Abstract : Online interactive activities are increasingly used, notably in the field of language teaching. The technical improvements enable synchronous long distance exchanges with a native speaker. The tutor engaged in a communication via a synchronous online platform must coordinate a complex set of operations for learning purposes. He can combine the different communication modes which are at his disposal (textual, visual and aural). We analyze, in the case of French lessons via Skype, which modes are used by each of the four tutors in our study. We postulate that the modes are differently used by each tutor and we seek to know in what proportion and with which “degree of utilization” (Develotte, Guichon, Vincent, 2011). We also try to understand what consequences the choice of the method has on the nature of the interaction. We present two hypotheses. First, we think that the initial profile of the tutors (individual skills, professional experience with face to face or distance teaching, comfort with the computer environment) and the context of the interactions (exterior perturbations, technical problems, type of task, needs expressed by the learners) have an influence on the way the tutor uses the modes. Secondly, we think that the choices influence the nature of the interaction and the relation between tutors and learners. We propose a grid of analysis of the multimodal online interactions in order to sustain our research hypotheses.