Discourse structure, the generalized parallelism hypothesis, and FG 307(15) a. Hey there, frankly, I don't like your hat!
b. *Frankly, hey there, I don't like your hat!
It becomes clear from (15a–b) that interactional expressions preferably oc-
cupy the absolute initial position preceding all the other constituents
including illocutionary satellites (such as frankly), which corroborates the
assumption that they pertain to a layer higher than the illocutionary one.
In the discussion of the extension of FG, the problem of the number and
the nature of the relevant units (or layers) has received more attention than
the relations holding between these units. Yet the notion ‘structure’ im-
plies, as is well known, constituency as well as relationships. In order to
give a complete and precise picture of discourse structure, we must also
take into account the various relations it involves.
In FG, relational structure subsumes two main types of relations: func-
tional and referential relations. Since the FG definition and characterization
of these two types of relations are largely well known, I will not go into
this matter in any more detail. Let us only keep in mind the following three
salient facts. First, the two types of relations differ from each other in the
sense that functional relations hold within the discourse structure itself, i.e.
between its constituents, whereas referential relations obtain between the
structure as a whole and what it refers to. We can speak, therefore, of
strictly ‘internal’ relations as opposed to ‘external’ relations. Second, as re-
peatedly pointed out and emphasized in Dik (1997a and 1997b), the world
within which the discourse refers is a mental ‘Discourse Model’ which
represents either S's conceptualization of the outside world or a purely
fictitious world. Third, as we will see below, not all functional relations are
involved in all the discourse categories distinguished so far.
Still on constituency, Dik (1997b: 415–422) draws attention to the im-
portance of “the global discourse decisions” that S takes in building up a
discourse, such as choosing a discourse type, a discourse style, a discourse
world, etc. The importance of such decisions, as Dik convincingly shows,
lies in the fact that they co-determine the internal structure of (the sub-part
of) the discourse they take in their scope. This obviously means that these
decisions must be represented in the underlying discourse structure. As re-
gards the way in which this can be done, I think that the following general
reflections may be of some relevance.
The exhaustive list of the discourse decisions in question, as established
by Dik, includes those which relate to the choice of discourse event (in par-
ticular the deictic centre), discourse genre (conversational, narrative,
argumentative, etc.), discourse style (formal/informal, polite/familiar, etc.),