164 Michael Fortescue
Nootka understood as ‘pattern’ (or grammatical ‘template’) presents
ample support for Dik’s treatment of both nominals and verbals as underly-
ing predicates.^16 On the other hand, the question of which element to treat
as main clausal predicate and which as argument (and of how many clauses
to split up one proposition into) is hardly relevant to the pattern perspective
at all if, as I have shown, this is determined by higher discourse-pragmatic
factors, which must ‘reach down’ from the topmost level. It would only be
relevant if there were a more or less direct link between the State of Affairs
to be expressed and a particular predicate frame corresponding to it in the
lexicon (built up usually around a verbal core), but this link is precisely
what is extremely tenuous in Nootka, where the main predicate could just
as well be an adverbial or a quantifying expression, for example. It is diffi-
cult to see how the pattern perspective could accommodate ‘percolation’
down from the level of illocution (or beyond) in the way that Hannay
(1991) has suggested, since this would have to extend right down into the
lexicon itself – besides which, ‘percolation’ is surely a process and not part
of a pattern description at all.
The nature of the link between clause/predication structure and dis-
course articulation is also much less direct in Nootka than in English,
where the clause can accommodate several arguments and satellites of sin-
gle events. By contrast, a unitary predication may in Nootka be broken into
several clauses, e.g. with one salient referent in each. The Nootka clause is
structurally very simple and less flexible than in English as regards meet-
ing more elaborate discourse needs than just indicating ‘newsworthiness’.
It makes sense to describe the pattern grammar of Nootka as being without
syntactic – and indeed with very limited pragmatic – function assignment.
(This is something the FG model already allows for.) Thus Focus in the
general sense of ‘newsworthiness’ could in theory be left right out of the
grammar as an etic, not an emic category, although in the serialization con-
struction, where it triggers an actual ordering choice, it had better be
retained.^17 The morphological ‘inverse’ can in turn be treated as a matter of
predicate formation in the Fund rather than of syntactic function assign-
ment, just like causative (and other morphological) valency-shifting
derivation processes. In other words, all that has to go from the traditional
pattern model is the direct link between lexical items and specific SoAs.
From the ‘process’ perspective, on the other hand, one can indeed en-
visage higher-level pragmatic choices (at the level of the illocution or
above) percolating all the way down to the basic predicate level and affect-
ing the choice of (main) predicate. Compare the following two
representations of sentence (1) above (I would work early, with early cho-