Discourse structure, the generalized parallelism hypothesis, and FG 315to another; it rather resides in the fact that they all borrow their internal or-
ganization from one and the same archetypal structure. It seems to me that
the parallelism phenomena at hand can be accounted for in a psychologi-
cally more adequate way in the light of the assumption that NLUs organize
their discourse according to a common archetypal communication structure
than if we consider that they do so by projecting the structure of a given
discourse category, especially since there is no agreement about the dis-
course category which is the source of this projection (clause or text).
Interestingly enough, this permits us to avoid the undesirable ‘clause-
centricity’ of FG.
4.1.2. ADS actualization in integrated use
As mentioned above, the discourse categories lower than the text can be
used in two ways: they can constitute complete communicative units or oc-
cur as (embedded) parts of larger categories. In the latter case, words can
constitute parts of terms which stand as parts of a clause; a clause can be
embedded in another clause yielding a complex clause; (simple and com-
plex) clauses, when coherently sequenced, form a text. In the previous sub-
section, we were concerned with the free use of the discourse categories;
our aim in what follows is to examine the way in which ADS is actualized
in these discourse categories when they are integrated into each other, fo-
cusing particularly on the embedding of clauses into clauses and clauses
into texts.
In the FG framework, several works have been devoted to embedding
phenomena (cf. Bolkestein 1990, Moutaouakil 1987, Hengeveld 1996 and
Dik 1997b, among others). Once reinterpreted in terms of the actualization
of ADS, the basic ideas advocated in these works can be reduced to the fol-
lowing general assumptions:
(a) The actualization of ADS is more restricted in integrated (embedded)
parts than it is in integrating (embedding) parts;
(b) Integrated parts ‘inherit’ features (layers or operator values) from in-
tegrating parts;
(c) The inheritance process is a matter of degree.
As regards assumption (a), it is now well established in FG circles that
the subordinate part of a complex clause need not be a complete clause: as
argued in Hengeveld (1996), it can be a clause or a proposition or even a