Priorities in discourse analysis 81recourse, but the work of scholars like Grice, Labov, and more recently
Sperber and Wilson opens up conventions of description which are too
sociologically dependent for the linguistic realities to be thoroughly observed
and described.
It was conceded from the start that as the linguistic units increased in
size, some of the description would have to be couched in non-linguistic
terms, as in Ventola (1987) a move may be simultaneously a directive and
a Sale Initiation (as the Italian ‘dica’, for example). This is not a double
coding because the provenance of the two analyses is quite different. A
directive is an Initiating move in the general structure of discourse, and a
Sale Initiation is a category of sociolinguistic description which is not directly
related to any linguistic unit or criteria.
Far from being a weakness, the lack of specificity of the higher units of
the original model was seen as an element of flexibility, adaptable to the
genre analysis of the future. Even the modest suggestion of transaction
boundary markers turned out to be less than reliable in teacher talk, one
major variable being the class size. Warren (forthcoming) points out that
real life does not always measure up to the structural sequences that are
expected of it. He also suggests that the study of spoken discourse may
have been over-affected by the use of telephone calls and quiz programmes
as data. They are much more predictably patterned than less specialized
discourse; at the beginning and end of telephone calls there are set routines,
no doubt stabilized because of a lack of shared environment.
One enduring problem is the rigorous description of the topic of conversations.
To the observer it is an obvious feature of talk that it is about something;
topics are proposed, supported, developed and concluded by the co-operative
linguistic behaviour of participants. In current work, Hazadiah (1991) shows
how Topic Frameworks form a rank between exchange and the transaction.
The relation between utterances and their discourse value was originally
seen as being partly determined by aspects of the situation in which the
language occurred. So it was said that if a teacher said ‘Can you swim a
length?’ there is a potential ambiguity. If the teacher and pupil are poolside
and the pupil suitably dressed, if swimming is not a proscribed activity at
the time, and if other conditions favour it, the utterance may be taken as a
command; otherwise a mere question.
This seems to rely far too much on the situation, and not enough on the
context of the discourse. Commands to swim do not just appear in discourse
out of the blue; they will be prepared for quite elaborately.
Subsequent work (Sinclair and Brazil 1982) built up a rule for interpreting
initiations based on their grammatical structure. However, the relative importance
of cotext and context has not been seriously explored. It is readily assumed
by most commentators that since a listener has access to large quantities of
knowledge of the world and its affairs, this knowledge will be deployed
according to hints given in the text, using inferential processes. Thus is a
text interpreted.