After characterizing the study of second language acquisition (SLA) from three viewpoints, I try to answer the question, raised by DeKeyser (2010), of whether the SLA field is disintegrating. In answering this question, I first propose a distinction between SLA as the relatively fundamental academic discipline and SLA as the relatively applied field of language education. Instead of portraying the field in terms of quantitative or laboratory studies on the one hand, and qualitative or anthropological studies on the other, I will look at SLA in terms of theories that differ in their empirical basis. All scientific disciplines must create room for ideas or theories that do not yet lend themselves to empirical testing, but for a discipline to develop fruitfully it is crucial that nonempirical ideas do not outnumber the empirical. The fact that the number of empirical SLA theories is large is not in itself a problem: through the practices of rational "normal science" (Kuhn 1962), the best theories (in terms of coherence, testability and scope) will rightfully come out on top.