1 I take it that the purpose of any language course is to develop in learners the ability to engage in communicative behaviour and this, I have argued, must mean that there has to be a concern for capacity, for the procedural activation of competence. To coin a slogan: no course without discourse. But language courses have generally concentrated on competence and left capacity out of account. The structurally ordered course concentrates attention on linguistic competence as such but does not effectively indicate how this competence can be drawn upon as a communicative resource. It is true that words and sentence patterns will often be associated with situations, but these situations are designed simply to reveal the symbolic signification of linguistic forms. The direction of fit, as it were, is situation to language.
2 In courses which have a notional/functional orientation, the focus of attention is on the schematic level and the direction of fit is reversed. That is to say, the starting point is a particular notional frame of reference or, more usually, a particular functional routine: asking the way, asking and granting permission, apologizing and so on. The language is then brought in to service the presentation of these schemata. In both cases the whole business of language behaviour is presented as a straightforward matter of projecting knowledge. One gets the image of the language user as somebody going around with bits of language in his head aiming for the appropriate occasion to insert them into the right situational slots.
3 But actual language use is not like this at all. It is rather a series of problems that have to be solved on the spot by reference to a knowledge of linguistic systems and communicative schemata. This knowledge does not provide ready-made solutions which are simply selected from storage and fitted in. But language courses have generally been based on the assumption that it does. Whether they are structurally or functionally oriented, what they have tended to do is to present and practise solutions. What they need to do, I suggest, is to create problems which require interpretative procedures to discover solutions by drawing on the knowledge available as a resource. In other words, they need to encourage the exercise of the capacity for negotiating meaning and working out the indexical value of language elements in context.