The study of how people learn languages has a long history but it became an identifiable sub-area of applied linguistics in the 1960s driven by two seminal papers (Corder, 1967; Selinker, 1970), which motivated intensive empirical enquiry. Early research focused on investigating the order and sequence of L2 acquisition, leading into work on variability in learner language and rethinking the role of the L1. Subsequently, researchers turned to the role of input and interaction, implicit and explicit learning, and the importance of consciousness in language learning. At this stage, SLA was predominantly a cognitive-interactionist enterprise directed at explaining how learners acquire grammar. However, the 1990s saw a social turn in SLA. There was greater emphasis on the social context of learning, on learners’ social identity, and on different aspects of language. Sociocultural SLA became a major influence at this time. More recently, the cognitive and social sides of SLA have come closer together through the investigation of learners’ complex dynamic systems. As SLA has evolved, we can see a change in the reasons for investigating L2 acquisition. In the early phases SLA researchers were interested in improving language teaching. In the later phases. SLA has become less applied and more purely academic, directed at contributing to our understanding of language and the human mind. I will illustrate how this change has taken place through an analysis of the journals that publish SLA research and suggest that this is one reason why teachers have become increasingly skeptical of SLA as a useful source of information about pedagogy.