Discourse structure, the generalized parallelism hypothesis, and FG 321hierarchically organized structure fits perfectly into a modular FG. This so-
lution can be said to have the following major advantages. First of all, it
permits us to unify the FG approach to the different discourse categories.
Second, it makes it possible to avoid two theoretically undesirable options:
(a) an unnecessarily recourse to other theories and (b) a costly and not al-
ways very convincingly justified multiplication of the modules within
MNLU on the other hand. Third, it allows us to redefine and extend the
central notion of embedding and to further refine the typology of embed-
ded clauses. Fourth, the perhaps most important gain is that this solution
provides the universal part of FG with a tool permitting us to describe and
explain in a more principled and unified way the similarities and the differ-
ences between natural language types as well as those between types of
natural discourse. Moreover, Dik’s (1997b: 415) insightful idea that a text
“can in many ways be likened to a piece of music” leads to the assumption
- hopefully to be verified by further research – that, if the transposability of
ADS turns out to be feasible, the solution proposed here can be viewed as
paving the way for a General Functional Theory (GFT) whose main task
would be to describe the general structure of the communicative process
and to account for its actualization in verbal and non-verbal modes of in-
teraction as well as providing particular theories for the various
communicative systems. Within such a general theory, FG would stand as
a member of a subset of linguistic theories to be compared to and evaluated
in relation with the theories of the non-linguistic subset. In this view, ADS
would be taken as the actualization in natural language of a more abstract
archetypal communication structure constituting one of the primitives of
this all-encompassing functional theory.
In this study, we have mainly been concerned with the underlying
(pragmatic and semantic) side of the structural parallelism between the dis-
course categories. I think that it would be of great interest to verify, in
future research, the extent to which a similar parallelism can be said to
also hold (within the grammatical module) at the surface (morphosyntactic)
level as well as the extent to which the latter parallelism can be taken as
resulting from a projection of the former.
Notes
- In arguing for the relevance of the discourse-type layer, Hengeveld (1997)
reports that in some languages (Turkish, Tauya and Krongo) certain verb
forms only occur in narrative texts. As for the discourse-style layer, its rele-