This paper examines research in the field of collaborative learning, concentrating on aspects suitable for use in an intelligent computer assisted language learning (ICALL) system. Current work on the project focuses on the construction of a student model, based on the notion that a model constructed and repaired through collaboration between the system and student will lead to a more accurate representation of the learner's beliefs, and is less likely to be regarded with suspicion by the student. Further, the act of collaboration should promote reflection on both the language to be learnt and the student's approaches to learning.