Analysing everyday conversation 141a rank, which is inconsistent with the fundamental principles of rank-scale
analysis.
Our data sample happens to be a complete interaction; it contains ‘opening’
and ‘closing’ transactions which define the interaction boundaries. We do
not wish, however, to elevate this observation into a generalization about
‘boundary transactions’, since we do not have enough data. The majority of
our students’ transcripts, on the basis of which this description was formulated,
did not consist of complete interactions, but were randomly selected stretches
of discourse within them. A much larger body of complete interactions
needs to be investigated before any structural statements as to their boundaries
can be made.
Secondly, there is the question of whether interactions exhibit any
further evidence of internal structure. Here the problem is that where
interactions do appear to be structured, this is usually the result of situational
factors. For example, the structure of a doctor—patient interview is affected
by such factors as institutional setting and the goals of the participants.
The fact that examination will precede diagnosis, and diagnosis will be
followed by prescription has little to do with linguistic structure. And
for everyday conversation, we have no evidence even of this sort of
structuring.
Following Sinclair and Coulthard, then, we describe the interaction as
‘an unordered series of transactions’, bearing in mind that this does not
mean that they do not display order but that this ‘order’ has not as yet been,
and perhaps cannot be, characterized in linguistic terms.^3
Having set out the framework of the analytical system, we will now go
on to look at some of the key areas and problematic issues, particularly
those concerning exchange structure.
RESTRICTIONS ON MOVES
On pp. 136–9 we set out the elements of structure of different exchanges
and the moves which may realize these elements. The following diagram
summarizes the restrictions upon where in an exchange a move may occur:
This shows, for example, that an eliciting move may realize I or R/I but
not R. This is, of course, consistent with the predictive nature of the
eliciting move (see pp. 145–9 below): it must realize a predictive element
of structure.