This thesis reports and discusses results from a qualitative study of Swedish teachers' writing assessment in upper secondary school. Based on teacher group discussions, the study investigates teachers' interactions, expressions of norms and decisions when assessing and grading students' writing in the subject of Swedish.

The aim of the study is to describe and analyze how teachers interact and what they pay attention to when discussing writing assessment. In relation to this aim the following three research questions were posed: (1) What characterizes the teachers' conversations about writing assessment? (2) What assessment norms do the teachers express? (3) What decisions do they make?

The theoretical perspectives are dialogical and pedagogical. From a dialogical point of view, the assessment conversations are considered as social acts where teachers in interaction collectively create meaning. Furthermore, the assessment conversations are considered as pedagogical, as a part of teachers' reflective work to discuss how to assess students' work and how to make decisions about further instructions and grading. The methods used to describe and analyze interactional patterns, expressions of norms and decisions in these assessment conversations are topic analysis and initiative-response-analysis.

The empirical data consists of audio and video recordings from three teacher groups' conversations about writing assessment, comprising a total of 17 Swedish teachers from three upper secondary schools. Data was also collected via a questionnaire with the individual teacher's grading of the students' writing and evaluation of the assessment conversations.

The results show that teachers' assessments of students' writing focus on much more, and partly other, criteria than their pedagogical decisions. The quality standards that teachers express in the discussions about student texts focus mostly on communicative quality, language style and text structure but also on content and the use of sources. The pedagogical decisions, on the other hand, almost exclusively focus on text structure and the use of sources.

The results also show that shortcomings in the students' texts are crucial for teachers' summative assessments. Meanwhile, the teachers also express that students' age, their writing development in the course and the national test must be considered. These assessment norms can be compensatory and have a substantial impact on these teachers' decisions on summative assessments. The teacher groups show considerable variation in the basis for their decisions regarding summative assessment of students' writing.

Finally, the teachers demonstrate a high degree of consistency within each teacher group when discussing summative assessment of students' writing. However, the assessments that individual teachers make after the discussions comply only to some extent with the group's decisions. The least degree of consistency was shown by the teacher group who, in the discussion, seemed the most consistent. Decision-making processes in these discussions are characterized by low intensity, where few assessment alternatives are being considered. The overall assessment consistency between the teacher groups was found to be low.