The study presented here is interested in understanding the ways in which social interaction in technologymediated institutional settings is constrained and afforded by what Pennycook defines as “critical moments” in the educational experience. Drawing on Social Learning Analytics and on the concepts of heteroglossia, contingency and chaining, this paper critically discusses a methodology that allows the analysts, and ultimately learners and educators, to follow and visually represent the mobility of the learners-in-concert-with-tools across space, time and language varieties and modalities in technology-mediated communication. The empirical data focused on here is drawn from a large project which includes 40 hours of naturally occurring interactional materials, generated through screen recordings of online sessions of an Italian for Beginners course offered by a Swedish university in the videoconferencing platform Adobe Connect. Preliminary findings suggest that the environment, both in terms of what happens inside and outside the virtual learning site, is of primary importance when it comes to the organization of the interaction among individuals in terms of what becomes the participants’ focus during the encounters. A visual representation through a multivocal approach of this mobility of learners, topics and tools will, it is suggested, support learners, educators and designers in locating where and when in the interaction “critical moments” have occurred, in order to understand how such shifts in focus support or hinder the learning experience.