This paper presents a project that recently received funding from the Swedish Research Council. Data collection will begin in autumn 2017, focusing on language practices in multilingual English classroom spaces in Swedish secondary schools. While English is the normal medium of instruction in lessons, research in bilingual educational settings shows that judicious switches between pupils’ mother tongues (L1s) and the target language (L2) facilitates communication and enhances L2 learning and motivation. Psycholinguistic research points in the same direction: L1 activation during L2 processing is unstoppable in low-proficiency L2 learners. Drawing on this prior research, on the translanguaging framework and the theory of language mode, the project examines a) whether and for what purposes teachers and pupils use their complete language repertoires in English lessons; b) whether other semiotic means are used to include L1s represented in the classroom of which the teacher has no knowledge; c) what mechanisms may underpin the language practices observed; and d) whether support in all the L1s represented facilitates pupils’ learning of L2 vocabulary and affects pupils’ participation in classroom interaction. Using a large survey to map self-reported teacher practices, ethnographic methodology and conversation analysis to study language practices in English classrooms, and experimental intervention and quantitative methods to measure L2 vocabulary learning over time, the project aims to contribute to the development of evidence-based language practices in English teaching in multilingual schools.