3 Exchange structure
Malcolm Coulthard and David Brazil
DESCRIPTIVE PROBLEMS
Introduction
Following any piece of research one is faced with the problem of demonstrating
the validity and generality of one’s findings and of showing that an explanation,
based of necessity on a fairly small sample of data, is applicable to similar
data collected by other investigators. During the past thirty years this problem
has been elegantly solved within traditional linguistics by the development
of generative grammars. A linguist can now present and exemplify his findings
quite briefly and then encapsulate them in a few abstract rules which will
generate all and only acceptable instances of the phenomena. The reader is
then able to insert his own lexical items and check the outcomes against his
own data or, more usually, his own intuitions and thereby evaluate the
description for himself.
By contrast most of the descriptive problems in the analysis of spoken
discourse remain to be solved. There has, so far, been no detailed theoretical
discussion of the peculiar nature of verbal interaction nor of the components
and categories appropriate to describing it—there is no Discourse Structures
or Aspects of the Theory of Discourse. Indeed, it is by no means certain that
the kind of generative description that grammarians have used so successfully
is an appropriate tool for handling interaction. As a result there are virtually
no commonly agreed descriptive categories; it is still not even clear what is
the largest structural unit in discourse and descriptions tend to concentrate
on fragments.
One notable obstacle to the development of a description of interaction
is that speakers seem to have weaker intuitions about permissible sequences
of interactive units than they do about permissible sequences of grammatical
units. Of course it may be that this is only the case because relatively little
work has so far been undertaken on the structure of interaction, but, nevertheless,
we have found the safest working assumption to be that, in the co-operatively
produced object we call discourse, there is no direct equivalent to the concept
of grammaticality. Indeed, the concept of competence, as it has been understood