The purpose of the research reported in this monograph has been twofold. First, it aims at contributing to an inquiry of the ways in which language and context are intertwined. Second, it aims at giving a characterization of a specific communicative event, second language teaching. The study starts out from a broad social-theoretical perspective, inspired by language game theory and ethnomethodology, as well as Goffman's (1974) 'frame analysis' and the work of Ragnar Rommetveit (1974, 1987). Levinson's (1979) notion 'activity type' is used in exploring how relevance criteria and frames of interpretation vary with the context of the activity in which language is used. The empirical material for the study consists of eight dyadic lessons of Swedish as a second language in grades 4-6 of the Swedish comprehensive, compulsory school. As material for comparison, the pupils, 10-12 year old boys from the Middle East, also participate in two non-didactic conversations around tasks defined by the research team, one together with his teacher of Swedish, one together with a class-mate of his. The first of the three empirical studies is a qualitative, discursive analysis of salient dialogue processes in language teaching activities. Abrupt shifts and breaks in the dialogue, misunderstandings, and lack of tuning between the conversational parties are interpreted as results of a tension between language at two levels in the language lesson. The dialogue in the language lessons of the corpus is characterized by an ambivalence between two perspectives on language, the ordinary, everyday perspective on language as a means for constructing and conveying messages versus the 'level 2 perspective', where language is seen as anabstract system of decontextualized linguistic items. The two other empirical studies are quantitatively oriented. In the first of these, important differences in dialogue processes, concerning dynamics, coherence and fluency are found between the lessons and the non-didactic conversations, as well as between different activities within the confines of a lesson. One of the most important results is that the teacher's interactional dominance seems to be systematically related to the content of lesson activities. The results of the last study suggest that in lessons, and especially language lessons proper, the pupil is givenfewer opportunities for talking and, also, that he refrains from taking the opportunities actually given to him. The main significance of the research is the demonstration of the dynamic character of linguistic communication and of the way in which linguistic meaning is the product of utterances being embedded in activities on which activity-specific premisses for communication are brought to bear. Also, the second language teaching situation is characterized as connected with particular communicative practices that are imbued with a certain degree of ambivalence and ambiguity.