Process and pattern interpretations 153fer to all procedurally interpreted levels of the model, from the choice of
basic predicate (and its relationship to the lexicon) up to the highest level
of illocutionary intention (and its relationship with still broader discourse
context).
My own position as regards the upper limit of the model as such is thus
twofold: from the pattern point of view (the actual ‘emic’ coding choices of
grammar) it is the illocution, just as Dik proposed. From the process view,
illocutions are simply an intermediate stage of organization whereby com-
plex intentions (perlocutions, etc.) find their way to expression via the
available linguistic code. The upward limit from this viewpoint would cor-
respond to the Whiteheadian ‘subjective aim’ defining the individual
communicative act. Such an intentional stance is, along with propositional
content, an integral part of the Whiteheadian judgment (which, unlike those
of standard propositional logic, is not limited to the assignment of positive
and negative truth-values). The way in which the intention to utter such a
judgment is worked out with the help of the ‘template’ of grammar is one
of the constraints on pure expression. A grammar as such is simply a set of
abstract ‘forms of definiteness’ (Whiteheadian ‘eternal objects’), and the
FG model (as pattern) is best regarded as a higher-level set of generaliza-
tions across the grammars of many languages. The obvious direction to
move in describing a process interpretation of FG is thus top-down (com-
pare also Hengeveld this volume).
Now a cornerstone of FG is the notion that pragmatics contains seman-
tics, which in turn contains morphosyntax (as in Dik 1989: 7). So if
pragmatics really embraces all levels of grammar it should be seen to ex-
tend down to the very lowest level of the process model it embraces,
namely the organization of predications. This will be my starting point in
what follows. Also for Hannay (1991: 146ff.) the ‘place to start’ is with
pragmatic/message organizing (choices to do with new versus given infor-
mation, etc.).^4 Note the difference from my proposal, however: Hannay
sees his message management module or component as an interface be-
tween a product-oriented model and a process one (of verbal interaction).
In fact he indicates choice of ‘mode’ as taking place at the highest level of
the extended FG model, i.e. as part of the ‘pattern’ grammar, e.g. TOPIC
(mode) DECL E 1 : [X 1 : ...]. My own approach is not really modular at all:
both discourse and grammar have their pattern and their process aspects,
only the pattern aspect of discourse is far more heuristic, relying heavily on
inference and background in ways that are not necessarily coded ‘emically’
in any grammar, and its patterns are across situations of use and means of
achieving ends, not across coding categories. The consequence of taking