A final chapter has necessarily a partially retrospective character: I shall begin this one by outlining the extent to which I hope in the remaining pages to look back, and to look forward. The chapter falls into two parts. In the first I want to focus upon what has, in the preceding chapters, been stated or implied, that discourse is systematic and its structure is predictive, and that these related concepts, that might at a superficial view be dismissed as altogether too abstract and general in character to be interesting to anybody except committed professional linguists, are, on the contrary, productive of insights that teachers can put directly to use. In discourse expectations are set up, are tested, and are confirmed (or not) and then, with the next unit, the next exchange, the cycle starts again. Much of what has gone before has to do with this process, and with what can be observed of the way newcomers to a particular sort of situated discourse learn to participate in it, to know what to expect and what is expected. There is some evidence that the expectations that are transmitted are remarkably durable. The predictive structure of discourse is recognisable in the text produced by teachers and learners of different ages, in varying situations, and despite all the differences of topic and presentation, the acquisition of specialised vocabulary, and the use of very much more complicated grammatical construction, that attend the passage from the reception classroom to the sixth form and beyond. Further, there is some evidence, and I shall want to examine it, that the structure of classroom discourse resists change between the generations, and shares the conservatism that has often been noticed as characteristic of educational processes generally. Some evidence may be available in answer to the question: how are expectations transmitted? how does it come about that they are so durable in the face of serious and well-publicised and deliberate efforts to modify them?