This is a qualitative study designed to investigate how professional support group leaders talk about their role as supervisor of support groups for children and adolescents with problematic home situations. We also examine how these supervisors talk about the method they use and how they talk about the individual child. We have chosen critical discourse analysis to help us explore if there are different ways to talk about these territories. Our questions are:

• How does the support group leader talk about his role?
• How does the support group leader talk about the method?
• How does the support group leader talk about the individual child?

The theories we have chosen to use in the analysis of our material is Aaron Antonovsky's Sense of Coherence and social constructivism. We have, as mentioned, chosen to do a qualitative study using semi-structured interviews conducted with seven active support group leaders at three different support group organizations. We transcribed the interviews and coded them separately and then went through the coded material together. The analytical method we have chosen is discourse analysis focusing on Norman Fairclough´s critical discourse analysis. Along with Fairclough´s three-dimensional model, we analyzed the interviews at the following levels: text practice, discursive practice and social practice. Based on our purpose and our questions we managed to distinguish six different discourses, within each territory we found two discourses that stood in contrast to each other. In the territory "How the support group leader talks about his own role," we found the knowledge discourse and the playful discourse. In the territory "How the support group leader talks about the method" we were able to discern a discourse view of the method as complete and a view of the method as adaptable. In the territory "How the support group leader talks about the individual child" we found the discourse about the perception of the child as a subject and the perception of the child as an object. These discourses were put together to constitute two discourse chains which showed two different orientations of the support group leader. We then analyzed these discourse chains our theories and questioned them based on how the two different types of leader’s we located could affect the method’s design and the children participating in support groups. The study concludes with a part in which we compare our results with earlier research within this field followed by suggestions for further research.