Language& Communicafion, Vol. 3, No. 3, pp. 271-304, 1983 Printed in Great Britain. 0271-5309183 53.00+.00 Pergamon Press Ltd. zyxwvutsrq LISTENERS’ COMMENTS ON CONVERSATION GRAHAM MCGREGOR A concentrated amount of work on discourse analysis over the last ten years has demonstrated that discourse is highly organized and amenable to analysis using traditional linguistic concepts such as, ‘sequential and hierarchic organization’, ‘system and structure’, and so on. Such work has gone some way to meeting Labov’s (1972, p. 120) criticism that, ‘Linguists have not made much progress in the study of discourse. By and large they are still confined by the boundaries of the sentence’, by examining larger units of structure (see, for example, the notion of ‘exchange’ in Coulthard and Brazil (1979)). Though linguists have considered the possibilities of applying their theoretical principles to the systematic analysis of ‘actually occurring conversation’, research continues to exclude any kind of reference to what kind of activity ‘conversation’ is. Labov and Fanshel(l977, p. 26) note, Many profound questions have been raised about the sequencing of individual speech acts or larger units of surface organization, but comparatively little attention has been given to the actions and reactions that are dependent on the position of the speakers in the social networks of family, group, or therapeutic situation. Approaches which seek to reduce social actions to linguistic categories are, however, likely to exclude any reference to the concerns of the participants and as such are liable to contribute to the ‘language myth’ which Harris (1981) warns us against. . . . the language myth assumes that a language is a finite set of rules generating an infinite set of pairs, of which one member is a sound-sequence or a sequence of written characters, and the other is its meaning; and that it is the knowledge of such rules which unites individuals into linguistic communities able to exchange thoughts with one another in accordance with a pre-arranged plan determined by those rules (Harris, 1981, p. 11). But semantic processes are relative to individual design and comprehensibility as Harris recognizes. The notion of an ideal speaker-hearer who operates a sophisticated cypher in encoding and decoding messages is a gross misconstrual of the complicated matrix of linguistic, social and psychological processes which mesh interactionally and determine the nature of verbal interactions. Language is a process of making communicational sense of verbal behaviour. Our training in language is a training to use words in such a way that, in the context of a particular situation, our total behaviour will make the kind of sense to others that we intend it should, and effectively implement our inter- actional objectives. It follows that language cannot be studied in isolation from the investigation of ‘rationality’. It cannot afford to neglect our everyday assumptions concerning the total behaviour of a reasonable person. These include assumptions about his probable utilisation of the linguistic resources available to him, but stop well short of assuming that everyone in a language community uses and interprets words in exactly the same way. On the contrary, it is manifest that if individuals actually behaved in accordance with the principle of mechanical uniformity conjured up by the language myth, most of their attempts to communicate would be bound to end in failure (Harris, 1981, p. 165). In other words, participants in various kinds of verbal exchange may presumably be supposed to know ‘what is going on’ and this knowledge presumably influences and is displayed in how they participate in ‘what is going on’. But we do not say anything useful about that knowledge of ‘what is going on’ simply by listing the superficial features 271