84 Advances in spoken discourse analysis
fully fitting response will confirm the accuracy of the statement; that any
relevant response will concern the redness or otherwise of what it refers to.
However, if the context suggests that the initiation is to be interpreted as a
warning to stop a vehicle, a fully fitting response will couple an acknowledgement
with some prompt verbal action.
In my judgement, the creation and maintenance of prospections should
be the defining criterion of an exchange. The initiating move creates prospections
which then determine the minimum extent of the exchange. Further moves
in the exchange may make further prospections, but there is no need to
classify them as initiations.
The control of prospection thus takes priority over the information model
of the exchange, reducing its significance but not necessarily its relevance.
It would not be appropriate here to assess the strengths and weaknesses of
an information model; it is enough, I hope, to point out that the state of
information transfer does not determine the structure of the exchange in the
way that the state of the prospection does.
MULTIPLE CODING
The second point raises the vexed question of multiple coding. Levinson’s
critique of discourse analysis (1983:289ff), and Tsui’s reply to it (1986)
give strong accounts of the two positions. Since human behaviour is infinitely
specific and infinitely subtle, it is ridiculous to assume that one designation,
especially from a small set of choices, is sufficient to describe its total
effect. But it is impossible to make an exhaustive description, and invidious
to make a selective description. So the alternatives to a single coding are
not really any better.
There are two strands of argument in favour of providing a single label
for each discourse act and move. One is that people actually talk about
discourse in terms of invitations, agreements, promises, etc., as if there was
psychological reality to the labels, and the language contained a rich variety
of labels. The other argument is that the language provides a full set of
closed classes for secondary acts, such as responses, cues, reiterations and
markers. The abundant use of these classes is only explainable by assuming
that to each there corresponds a discourse function, and that speakers understand
the discourse, at a primitive level at least, in terms of a small number of
mutually exclusive alternatives. A well-known example of a closed class is
the marker, which in classroom discourse is realized by one of well, OK,
now, good, alright, right.
There will always be counter examples. For example fictional spies conduct
exchanges which have two simultaneous meanings. All sorts of codes are
used, for example by adults with children present. These are specialized and
marginal types of discourse, perhaps of a similar status as ironies—in fact
the literary figure of dramatic irony is one such example. But just as the