This paper argues for a cognitive psychological approach to second language phenomena that emphasizes the importance of the development of automaticity and the process of restructuring. It is argued that practice can leadimprovement in performance as sub-skills become automated, but it is also possible for increased practice to create conditions for restructuring, with attendant decrements in performance as learners reorganize their internal representational framework. In the second case, performance may follow a U-shaped curve, declining as more complex internal representations replace less complex ones, and increasing again as skill becomes expertise. Examples are drawn from first and second language research, and from research on expert systems. The cognitive approach is not seen as competitive to, but as complementary with, linguistic approaches to second language development.