Scholars and practitioners have increasingly come to recognise the centrality of learner agency in generating learning opportunities. Such agency is most clearly observed in the initiatives that learners take in the language classroom. Despite the importance of learner initiatives, however, a precise and comprehensive understanding of what such initiatives entail remains lacking. Based on a conversation analytic treatment of 160 relevant cases culled from 14 hours of ESL (English as a second language) classroom interactions, I propose an empirically based "typology" of "learner initiative." Drawing upon various theoretical assumptions of what promotes learning, I also give preliminary considerations to the kinds of learning opportunities such initiatives generate.