There are many theoretical and practical problems that attend the study of transfer. It is beyond the scope of this chapter to review them all, but problems in four areas have an especially important bearing on the discussion in subsequent chapters: definition, comparison, prediction, and generalization.

Problems of definition

The terminology used to study language reflects – and sometimes creates – vexing problems, and in the terminology of second language research, the term transfer is as problematic as any. The issue of cross-linguistic influence is controversial with or without the term, but the long-standing use of transfer has itself led to differences of opinion. Some scholars have advocated abandoning the term or using it only in highly restricted ways (e.g., Corder 1983; Kellerman and Sharwood Smith 1986), yet many others continue to use it without restriction. In this section, a definition of the term transfer will be presented, along with a critique of that definition. However, before any observations are made about what transfer is (or at least seems to be), some observations about what transfer is not are appropriate.

Transfer is not simply a consequence of habit formation. A discussion of contrastive analysis and behaviorism by Carroll (1968) makes clear that the behaviorist notion of transfer is quite different from the notion of native language influence (cf. Section 2.2). For one thing, the behaviorist notion of transfer often implies the extinction of earlier habits, whereas the acquisition of a second language need not (and normally does not) lead to any replacement of the learner's primary language.