After a detailed consideration of Saporta's (1966) “paradox of second‐language learning” as a point of departure, the nature of paradoxes in general is discussed, particularly the conclusions they force upon a researcher. The nature of Saporta's paradox is then dealt with in terms of the linguistic aspects of second‐language learning. This leads to the proposal of a research strategy involving profiles of idealized learners with specific types of individual differences. Detailed suggestions are provided for experiments relevant to the wayindividualsactually learn second la