In drawing an ‘outline sketch’ of the field of research on interactional learning situations with computers, this paper attempts to provide a general framework for the contributions which follow. After a brief overview of the theoretical grounding of studies in the field, we examine the extent to which learner-computer interaction and learner-learner interaction can interfere with or support one another.

The need for a change of theoretical perspective on the role of social interaction in learning is highlighted. It is argued that viewing collaborative learning in terms of joint negotiation of a common problem space might help in defining the optimal characteristics of educational software.

Finally, the critical role of social dimensions such as social comparisons between partners is pointed out. In conclusion, the need for more local theories, taking particular account of the learning domain, is stressed.