I argue that abstraction is a critical feature of SLA research, although it is not relevant to every domain of that research. For those interested in the underlying mental representation of L2 learners, abstraction in the form of linguistic features (e.g. subject, Infl, ±strong agreement) is part and parcel of the tools needed to describe acquisition. For those interested in something like skill theory, such abstraction is not necessary to talk about the development of concepts such as fluency and accuracy. In the end, the importance of abstraction in SLA research is dependent on what aspect of SLA one focuses on.