In an earlier essay (Noble, 1981) I pointed to certain problems with Gibson's ecological theory of perception (sometimes called a 'direct' theory-see Michaels and Carello, 1981, for a recent overview of the paradigm). It was argued that the place where this theory gets itself into trouble is at the interface between perception and language. For all the power of the theory to account for 'natural world' perception, it comes unstuck in the face of human perception mediated as it is by language. At a 'low' level (language as information at second-hand) the theory takes account of the languageperception connection, but it does not begin to address itself to the way language actually 'goes on' in the real world. Thus, the theory can make little headway towards accounting for the general run of human experience, even though it is trying to aim for such generality. (Reed, in this volume, argues a case for Gibson's treatment of language and perception as more thoroughgoing than what has just been implied. Gibson recognises that language functions in fixing and expanding awareness, in socializing through training and teaching. And Gibson brings language, as expression, into the material and out of the idealised realms. Missing still, however, is appreciation of the normative and rhetorical functions of language in the control of conduct.)