During collaborative writing, two or more learners construct and write a text together, participating equally in planning, reviewing, and revising the text; and being equally responsible for the quality of the final text. In collaborative writing the partners have to exchange ideas, plans and suggestions for the topics and formulations of the text. Furthermore, they have to reach a common solution to the problems that arise. (Giroud, 1999.) Collaborative writing can be used as a kind of didactic solution which removes focus from teaching to learning. When collaborative writing is used as a learning task it is based on the idea of collaborative learning that can be defined as a situation in which two or more people are working together to learn (Dillenbourg, 1999). The main advantage of collaborative writing, when compared to individual writing, is that writers can receive and give immediate feedback. The other advantages are that collaborative writers need to test their hypotheses, clarify their goals, and justify their propositions by verbalizing and negotiating for their meanings (Erkens, Jaspers, Prangsma, & Kanselaar, 2005). This kind of joint construction of knowledge both at metacognitive and at a specific content level, is an essential part of collaborative writing (Erkens, Prangsma, & Jaspers, 2006). The studies of collaborative writing have recently focused on computer-mediated communication. The present study concerns the quality of discussions when students are concentrating on a collaborative writing task face-to-face. This study investigates how university students (n=20) discuss in small groups (six groups, 3–4 persons per group) when they engage in collaborative writing during a course on educational psychology. The students had prepared themselves for the group writing by reading six chapters of a course book on developmental theories and by writing summaries of them. The summaries had to contain at least three essential ideas of each theory. In collaborative writing situation each group was, firstly, asked to discuss the various theories and, secondly, to write an essay of one of those theories. The data consist of the students’ discussions (8177 speech turns in total) when they were writing their essays together. The discussions consisted of 564 conceptual episodes which were divided into five categories: 1) directive discussion, 2) writing, 3) topic related discussion, 4) evaluation, and 5) off-task discussion. The results showed that the writing episodes were the most common (50%) and that there were only few evaluation episodes (6%). The proportion of episodes that concentrated on the content area of writing, i.e. on theoretical concepts or assumptions, varied between the groups (range 19 %–31%). The results suggest that some groups were mainly oriented to get over the writing task while other groups discussed concepts more. This can be associated with the quality of the students’ essays. The results also suggest that there were only few evaluation episodes which can be associated with the students’ metacognitive skills.

The results showed that the writing episodes were the most common (50%) and that there were only few evaluation episodes (6%). The proportion of episodes that concentrated on the content area of writing, i.e. on theoretical concepts or assumptions, varied between the groups (range 19 %–31%). The results suggest that some groups were mainly oriented to get over the writing task while other groups discussed concepts more. This can be associated with the quality of the students’ essays. The results also suggest that there were only few evaluation episodes which can be associated with the students’ metacognitive skills.