Discourse structure, the generalized parallelism hypothesis, and FG 313hospitable. In fact, one may say that ADS is potentially present in any
normal communicative event, whatever the discourse category through
which it is mediated. The difference between the discourse categories in
question resides, therefore, rather in how they house ADS. In general, the
full, explicit actualization of ADS takes place in the text; in the other dis-
course categories, only a part of ADS is expressed, the other part being
implicitly taken care of by the context/situation.
As may be deduced from the characterization of verbal interaction given
in Dik (1997b: 409), the discourse category through which a complete
communicative event takes place in the most explicit way is the text. This
is why NLUs speak in texts rather than in isolated sentences. We may thus
expect the full actualization of structure (17) to obtain optimally in a whole
text. This indeed emerges clearly from the contrastive examination of three
typologically quite different texts: Brunhoff's fairy-tale The story of Babar
the little elephant (already analysed from the point of view of Topic-Focus
assignment by Mackenzie and Keizer 1991), Najib Mahfouz's novel Han
Al-Halili ('Al-Halili Quarter') and Diderot's Jacques le fataliste.
It should be noticed here that the other discourse categories (clause,
term-phrase and word), as pointed out in many works (Dik 1997a, Mou-
taouakil 1993, and, in particular, Mackenzie 1998), cannot be said to be
less communicative. They also can be used to carry a complete piece of in-
formation, with the difference, however, that part of this information
remains implicit, i.e. is transmitted by contextual and/or situational means.
In other words, the difference between the discourse categories at hand
does not reside in their communicative capacity proper but rather in the de-
gree of explicitness of their communicativity.
Concerning the actualization of structure (17) in the discourse catego-
ries other than text (clause, term-phrase and word), data suggest that it
becomes more restricted as we run from the top to the bottom of DCH. As
regards this restriction, it will become clear through the following
examination of the internal structure of the clause, the term and the word
(where they stand alone as performing autonomous and complete speech
events) that (a) this is, in fact, a consequence of the decreasing hosting
capacity of these discourse categories, (b) it may affect the layering, the
operator values and/or the relations and (c) it operates according to a
certain directionality, i.e. from higher to lower levels and layers.
When used as a complete discourse unit, a clause can generally display
all the layers of the interpersonal and representational levels as well as,
though to a lesser extent, as (a) and (c) predict, those of the rhetorical level
(which typically occur in larger stretches of discourse). As regards func-