Process and pattern interpretations 173
constructions (like Chinese and Japanese) seem to fall somewhere between
English and Nootka as regards constraints on how much information can be
packaged in a single clause. The issue is complicated by the further ques-
tions of how (and how much) new information is distributed in successive
clauses (for example, one such unit per clause, as Chafe 1994: 109ff. sug-
gests) and whether semantically obligatory serial verb constructions in
Papuan languages, for instance, really represent more than one predicate per
clause (as opposed to a single complex one).
- Typically of Nootka, neither 3rd person subject nor object are overtly
marked when context can supply them, as here. Nakayama glosses the ‘in-
verse’ marker as ‘SHIFT’, since, as he points out, its function is broader
than that suggested by the former term.
- As in kusanartu-mik qaja-liur-puq (handsome-INSTR kayak-make-
3SG.INDIC) ‘he made a handsome kayak’. There is no way the affix here
could be construed as ‘enclitic’, by the way – it must follow the ‘incorpo-
rated’ head. In West Greenlandic there are even a few affixes, for example
- nirar- ‘say that’, which can take as base an entire proposition with its own
tense or modality marker (stripped of only person/mood inflection), as in
(i); though again its position is fixed
i. imiq nillir-sima-nirar-paa
water be.cold-PERF-say.that-3SG/3SG.INDIC
‘He said that the water had been cold’
- Engberg-Pedersen points out to me that sentences such as these are com-
pletely parallel to similar utterances in Danish Sign Language (pers.
comm.).
- It is not that traditional FG has any particular problem with languages that
have only two parts of speech (as can be argued is the case for Nootka),
namely nominals and verbals, which are distinguished almost entirely by
morphological criteria, i.e. by the type of inflectional and derivational proc-
esses to which they are subject respectively (Dik 1989: 163 specifically
mentions neighbouring Salishan languages in this context). What is more
problematical is the association of such a wide array of lexical stems with
both verbal and nominal (and other) uses. Term predicate formation in
Dik’s sense (also corresponding derivations from semantically adjectival
and adverbial elements) would on the one hand have to apply across a high
percentage of all sentences and, on the other, a wide array of potential SoAs
would have to be associated with individual lexical entries.
- Deciding what is to be the main predicate given a complex communica-
tional intent can also be problematical with languages which do not display
the same ambiguity of major word classes as Nootka. Thus consider the ini-
