This article has two purposes. At the theoretical level it is intended to contribute to debates about how learning happens over time. At the methodological level it is intended to suggest how a complex understanding of time can enrich both ethnographic methods and linguistic analysis. It uses the concept of timescapes to interpret ethnographic and linguistic data drawn from a study of identity and learning in adult literacy education in England. It makes three interconnected arguments. Firstly, learning and research do not just happen "in" time, but involve the social production "of" time. Secondly, learning and research do not happen in isolatable "moments", but each moment is temporally extended to include the past and the future. Thirdly, analytical approaches that treat time as a static container or decontextualised measure of experience cannot fully account for the complexity of learning, which involves the creation of multi-dimensional timescapes. (Contains 1 table, 1 figure, and 1 note.)