252 Advances in spoken discourse analysis
Up to then Chris had not said anything. We both got out on to the flat
roof at the top. Then someone in a garden on the opposite side shone a
torch up towards us. Chris said: ‘It’s a copper, hide behind here.’ We hid
behind a shelter arrangement on the roof. We were there waiting for
about ten minutes. I heard some more policemen behind the door and the
policeman with me said ‘I don’t think he has many more bullets left’
Chris shouted ‘Oh yes I have’ and he fired again. I think I heard him fire
three times altogether, the policeman then pushed me down the stairs and
I did not see any more. I knew we were going to break into the place.
I did not know what we were going to get—just anything that was
going. I did not have a gun and I did not know Chris had one until he
shot. (The whole confession is reproduced in Appendix 2 below.) I now
know that the policeman in uniform is dead.Bentley is supposedly telling a narrative of how he came to be with Craig
on the rooftop; then, after a series of unremarkable positive assertions he
suddenly says ‘Up to then Chris had not said anything.’ There is no preceding
justification, no subsequent take up, no apparent reason for him to deny this
particular occurrence; there were a great many things Chris had not done up
to that point. Shortly afterwards the negatives come thick and fast. What
has not happened is rarely reportable in narratives; why, one wonders, should
Bentley want to report that Chris ‘had not said anything’ and why should
he subsequently, in what is a narrative text, produce five clauses in succession,
four of them denials, all concerned with facts, not events? The simplest
explanation is the one Bentley himself advanced, that the police asked him
questions about things that were important to them, he replied in the negative
and the exchanges were incorporated into the text as negative monologue
sentences.
Discourse structure
The difficulty with evidence based on discourse structure is twofold: firstly,
we have the acknowledged fact that nothing can be regarded as ‘undiscoursical’,
i.e. anything can occur and so the analyst is forced to argue at best by
appeal to probabilities; secondly, we currently lack a data base, that is a
substantial collection of texts with which one can compare a suspect text.
However, it looks as though concentration on the the rank of sequence,
that is the unit of discourse where topic is handled, is likely to yield the
best results. One of the marked features in interviews which are acknowledged
to be authentic is the occurrence of sequences of topic-linked exchanges,
what others might characterize as a series of follow-up questions
Pc: I would like you to tell me about that red Fiesta
B: I bought a car from a breakers yard...
Pc: The car was stolen...are these the tools used in the burglary