90 John Connolly
In other words, what phenomena belong under the heading of discourse
studies and thus merit consideration in respect of the design of frameworks
for discourse representation?
We may begin by noting that discourse can be regarded either as a pro-
cess of communicative interaction or as the product of that process. Both
these perspectives need to be accommodated, and we shall seek to do so
here, using the term ‘discourse’ in whichever sense is appropriate in any
particular context.
The approach to discourse studies presented in Dik (1997b) may be
summarized as follows. Discourse as a product is viewed as a coherent
stretch of spoken or written language, which normally contains more than
one clause, and which may also contain one or more extraclausal constitu-
ents (ECCs). Discourse as a process can take the form of a monologue or
dialogue, and the interlocutors (or participants) are labelled S (for
speaker/writer) and A (for addressee) as appropriate.
Any discourse process will be embedded within a Discourse Event,
which is a type of social phenomenon. There are several important aspects
of discourse which apply to the event rather than the content of the lan-
guage employed. These include (i) the identity of the interlocutors and their
relationship to each other, (ii) the time, place and setting of the discourse,
and (iii) the prevailing social conventions which govern the acceptable use
of language. When embarking upon a discourse, several decisions need to
be made which tend to affect the construction of the whole piece of lan-
guage concerned. These include the choice of genre (such as a lecture or an
interview) and the selection of style (which may be formal or informal, po-
lite or impolite, and so forth).
As the discourse proceeds, the interlocutors develop a mental picture
which enables each new utterance to be interpreted in the light of what has
gone before, and also in the context of the interlocutors’ pragmatic infor-
mation concerning the discourse situation and, indeed, the wider universe.
This evolving mental representation, which has a degree of commonality
among all the interlocutors, is termed a Discourse Model (DM). Because it
is constantly changing, a DM is dynamic, and because it can never include
all the pragmatic information that is conceivably relevant, it constitutes
only a partial model of the discourse concerned.
As stated in Dik (1997a: 300–304), each individual clause has a basic il-
locutionary value which is to be viewed as a coded instruction whereby S
seeks to modify A’s pragmatic information. The possible illocutionary val-
ues are declarative, interrogative, imperative and exclamative. These types
are recognized on the basis that they tend to be coded differentially in the