240 WILLIAM H. JACOBSEN, JR.
essence these are potentially predicative forms with no outer-layer (core or
clausal) overt markings. They constitute a primary reason why it is difficult
to be certain of clause boundaries and hence to specify the number of
clauses in a passage. The following gives a good indication of the problem:
(37:2)
big-iNCEP-3 water stirring in waves-DUR-3.
This was translated as "the waves grew bigger and bigger", treating the sec
ond word as the subject of the first, but it seems possible to regard this as
two clauses, which would mean literally "it grew bigger, the waves were
stirring". In the preceding line a predication based on this stem in fact
occurs:
(37:1)
water stirring in waves-iTER-MOM-FiN--
"now, it is said, water began to stir upward in waves".
In the suggested alternative interpretation we would have an instance of
clausal cosubordination. It is indeed the absolutive of a predicate that is the
form occurring in cosubordinations of all three juncture types. We will see
below that another uncertainty about clause boundaries hinges on the occa
sional difficulty in distinguishing between clausal and nuclear cosubordina
tion.^10 Swadesh (1939:78-79; quoted in extenso in Jacobsen 1979a:85-86)
reacted to such facts by claiming that there is only one major part of speech
and only one syntactic relation, that of supplementation, and moreover that
word order is not an unequivocal guide to the scope of this relation. I have
previously (Jacobsen 1979a) tried to suggest (using Makah data) that this is
an exaggeration: different parts of speech can be distinguished, albeit with
overlapping properties, and various different syntactic relationships can
also be seen to obtain. Nevertheless, in surveying again this material, I feel
quite acutely the uncertainties that he was expressing.^11
5. Main clauses
Turning now to a more specific examination of these texts, let us first con
sider the main clauses in the two styles. It is convenient to speak of three
groups of markings of predicates: subordinate (including nominalized),
absolutive, and independent. Independent clauses are not subordinate or
cosubordinate and have complete peripheries marked for illocutionary
