In this article the focus is on methodology for analysing learner-learner oral conversations mediated by computers. With the increasing availability of synchronous voice-based groupware and the additional facilities offered by audio-graphic tools, language learners have opportunities for collaborating on oral tasks, supported by visual and textual stimuli via computer-conferencing. Used synchronously with real-time voice-based work, these tools present learners with the challenge of learning a new type of oral interaction, and researchers with the need for developing methodologies for redefining L2 oral competence in these environments. In this paper we address the latter. We examine approaches from the interactionist branch of Second Language Acquisition research, and we question the ability of this model of language learning to fully account for the processes that take place when learners are interacting with machines while talking to each other. To complement the socio-cognitive insights of that school, we look to interactional linguistics and to social semiotics. Building on findings from these fields, we offer a qualitative discussion of the discourses evidenced in conversational data from two distance-learning projects that use synchronous voice in conjunction with other stimuli, in an intermediate French programme at the UK Open University. We then present detailed conclusions about the methodological challenges involved in analysing the oral competence of students who use these tools.