154 Michael Fortescue
the overarching nature of pragmatics seriously must, it seems to me, result
in a process interpretation of language that does not set discourse and
grammar side by side on the same ontological level, but rather treats gram-
mar as abstract pattern or template at the service of communicative
purposes. Let us avoid formalizing the patterns of pragmatics as if they
were merely an extension of the patterning of grammar. This amounts to
having one’s cake and eating it as regards the choice between incorporating
a discourse ‘level’ into the existing FG model (in the manner of Hengeveld
1997) or interfacing it with a separate ‘discourse grammar’ (in the manner
of Hannay 1991). It is in fact compatible with Hengeveld’s more recent
proposal (this volume) for combining the two approaches to discourse
within the FG model.^5
In this chapter I shall be extending the notion of letting the highest (dis-
course-pragmatic) level of a processual interpretation of the FG model ‘dip
down’ to the lowest, predicate level, as proposed for predicate formation
rules in Koyukon and West Greenlandic (Fortescue 1992: 131). Languages
vary on this dimension, some allowing whole propositions as input to cer-
tain derivational rules, others strictly limiting input to stem predicates (as
English, by and large, does).^6 In the following section I shall specifically
look at relevant phenomena in Nootka. My representations will focus on
the level of the (extended) predication (= Whiteheadian proposition, as in-
gredient in the ‘judgment’, which corresponds more or less to Dik’s
‘proposition’), since that is where I claim the link to psychological reality
must be made. The ‘missing’ relationship between this level and the lexi-
con (with its reflection of SoAs) will be returned to in the final section.
- Choice of predicate in Nootka
In Nootka, as in the other Wakashan and neighbouring Salishan languages
of western Canada, the distinction between lexical nouns and verbs is noto-
riously slippery. Moreover, the choice of constituent that is treated as
(main) sentence predicate often goes contrary to European expectations.
What corresponds to an adverbial or a quantifier, for instance, might be
chosen as predicate, followed by a bare (structurally indeterminate) verbal
or nominal expression whose relation to the predicate – e.g. as subject – is
very loose (and must be inferred). Thus the following sentences appear in
Mithun (1999: 188–189, citing Nakayama 1997):