Language teaching and learning aim at practical outcomes in the real world, not simply at classroom outcomes. If we want to show that a feature of language has been taught or learned it is not enough to show that a learner can produce it under artificial con- ditions, in a grammar test for example, or as a controlled response to a teacher’s question. We need to see that it has become a consistent part of the learner’s language repertoire, that the learner can use it consistently as part of an act of communication. If we take this as the yardstick, then the findings of research into second language acquisition suggest that there is no direct link between teaching and learning (see, for example, Skehan 1998: 94–95). Learners may, for example, have a conceptual under- standing of the use of past tense forms when their attention is focused on producing the required form but, at the same time, they may fail to produce these forms when they are using language spontaneously. The conclusion from the research is that we should recognize that the relationship between teaching and learning is indirect. Even if a form is understood and produced under controlled circumstances this does not guarantee that learners will be able to use it.
It is, however, comfortable for teachers and teacher educators, and for the writers and publishers of teaching materials, to maintain that there is a direct relationship between teaching and learning. For teachers and teacher educators this belief offers security. It suggests that we know exactly what we are doing and where we are going. We can plan lessons and recommend methodologies with confidence. For writers and publishers it means that they can make clear, unqualified claims in terms of teaching and learning for the materials they produce. There is, then, built in resistance to the notion that learning is, at best, indirect. The inconvenient arguments from second language acquisition (SLA) are rejected on the grounds that they are unscientific, or that the work on which they rest is carried out under experimental rather than classroom conditions, or that one study often contradicts another with respect to the details, or that they are too diffuse to offer a firm basis for a teaching programme. But the fact that almost all the research points in the same direction casts doubt on these criticisms.
For my part I believe that the overall findings of SLA research are convincing. They reinforce my experience in the classroom. I am horribly familiar with the situation in which learners can produce a language form under controlled conditions, but cannot produce the same form spontaneously. In this chapter I will argue, taking language description as a starting point, that there are two good reasons why we should recognize that the link between teaching and learning is bound to be indirect. I will then go on to cite Widdowson (1979) to suggest a third reason for this phenomenon.
 
Three