Remarks on layering 295
murderer and John thinks he is the murderer or At that moment I thought he
was the murderer. A similar distinction exists in the use of speech act verbs.
But it does not exist in expressions of qualifications from time downwards.
It is actually important to stress that the layered system in (4), and the dis-
cussion relating to it, only concerns the performative ‘version’ of deontic
and epistemic modality and evidentiality. The status of descriptive uses is a
completely separate story. I will briefly return to this matter in Section 5 be-
low.
- If they do, they lose their epistemic status and become speech act modifiers:
Dutch misschien ‘maybe’, for instance, can figure in a question, but then it
does not express epistemic uncertainty, but acts to turn the question into a
tendentious one. See Nuyts (2001a).
- This relation between qualifications and clause linking immediately ex-
plains why there are so many diachronic and synchronic connections
between qualificational and discourse markers, as observed, for example, by
Traugott and König (1991), and why diachronic developments even of dis-
course markers do fit into the pattern of continuing subjectification, as
Traugott calls it, i.e. of climbing up the ladder of the qualificational hierar-
chy: they do so by virtue of the relation of discourse markers to
qualifications in the layered system.
- The difference between Hengeveld’s handling of discourse and the account
in the present framework could be summarized as follows. Hengeveld
seems to adopt an ‘accumulation’ concept of discourse production: the steps
taken at the interpersonal and representational levels accumulate at the ex-
pression level. Functional Procedural Grammar adopts a ‘tear apart and
code into language’ concept: large clusters of information going into what
will ultimately emerge in the hearer’s mind as a coherent discourse are pre-
sent in the speaker’s conceptualization and early pre-linguistic stages of
communicative processing only. The whole process of language production
is a matter of gradually singling out small chunks of information which get
coded in single utterances (albeit in discursively adequate ways). Grammar
actually only deals with the last part of that process, the single utterance.
- It may be relevant to refer to the fact that Systemic Functional Grammar
(e.g. Halliday 1994), too, deals with discourse organization in the semantic
stratum, and not in the lexico-grammar, for good reasons, which have been
elaborated at length in the Systemic literature.
- This does not necessarily mean that, as part of one’s conceptual representa-
tion of the world, descriptive qualifications are not distinguished from the
facts about the world which they qualify. Maybe they do maintain a special
status, for knowing about their qualificational status is obviously necessarily
part of one’s knowledge about them. So they may form a sort of intermedi-
ate level between the representation of the SoA and the hierarchy of
