This thesis presents the results from two research projects on teachers’ assessment beliefs and practices in multilingual education. Study I involved teachers of biology, history or English as a foreign language (EFL) in Swedish upper secondary Content and Language Integrated Learning (CLIL) schools, grades 10-12, where English is used as the medium of instruction. In Study II teachers from schools with newly-arrived migrant students (NAS) grades 7 to 12 in the natural and social sciences participated. In both contexts, students are learning the language of instruction at the same time as they are expected to develop subject area knowledge why issues in relation to the role of language in assessment come to the fore. The aim is to contribute to the knowledge of an underexplored research area on subject matter assessment in multilingual schools and draw attention to the consequences varying language policies and pedagogies may have on fairness in access opportunities and validity in assessment outcomes.

In this thesis teachers’ language beliefs and practices as expressed in interviews, questionnaires and assessment samples were compared and analyzed in relation to the cognitive and linguistic requirements of language functions in syllabi and the assessment tasks. Whereas Study I was mainly qualitative in nature, involving 12 teachers, a mixed method approach was adopted in Study II where 196 teachers participated in a survey and 13 in follow-up interviews. The responses in the survey were analyzed using descriptive and inferential statistics. Thematic content analysis was used for the interview data, the assessment samples and open-ended survey questions.

The findings from the two studies indicate that although teachers state that language is not part of the assessment, they maintain that students need to use language to show proof of critical thinking and more advanced analytical skills. All teachers regardless of discipline shared the weight attached to covering course content as expressed in syllabi which points at a shared accountability culture. One of the main concerns expressed by the teachers is how to construct assignments where language does not represent a hindrance to show content knowledge. Teachers in both studies claimed to focus mainly on subject concepts, not general academic language, and the assessment beliefs and practices seemed to be closely related to the character of the subject. The non-parametric tests of association revealed that teachers with a dual language and subject content certification displayed significantly higher results in relation to all activities involving a visible language pedagogy, e.g. looking at useful sentence structures and providing model texts. Oral follow-up was used by all teachers to remedy poor written results. Although the use of the students’ strongest language is advocated in guidelines for the instruction of NAS, most teachers in Study II referred to a monolingual Swedish language norm. 

In a society where the educational discourse has become characterized by diversity, inequality and segregation, these two studies underline the need for a shared language policy and pedagogy across subjects and school contexts. They also suggest that an organization and teacher profession with an explicit responsibility for academic language is needed to provide equal access to subject content and validity and comparability in assessment in multilingual schools.