CLAUSE COMBINING IN NOOTKA^237
styles, as contrasted with something like 34% for English conversation and
46% in English letters. Mithun defines poly synthesis as subsuming pronom
inal incorporation, and she notes that not all such languages are so sparse in
subordination, obtaining a figure as high as 28% for Tlingit (Na-Dene). She
suggests that the low proportions of subordination might be partly explaina
ble from the free word order made possible by the obligatory pronominal
affixes on the verb. There remains, of course, an empirical question of the
extent to which the proportion of this sort of stylistic device can be
explained internally within a language's structure. There are very likely
fashions in this respect that may spread over an area — one thinks of the
Latinate style that spread over western Europe with the development of lit
eracy, so in this respect Mithun's discussion of the development of a literary
tradition as encouraging heightened subordination is entirely to the point.
The other influential contribution comes from works of the inves
tigators who have presented a framework for typological generalizations
under the label Role and Reference Grammar. Part of this is a typology of
clause linkage, with two parameters (see Van Valin 1984, also Foley & Van
Valin 1984:238-320, Foley & Olson 1985). Most significant for our purposes
is that of nexus, with three types of relations. In addition to well-recognized
coordination and subordination, they propose one of cosubordination, in
which one clause is not embedded within another but is nevertheless depen
dent upon it for the expression of certain of its grammatical categories.
They furthermore recognize three layers in the clause: innermost nucleus,
intermediate core, and outermost the whole clause itself (originally called
the periphery', I am herein following the revised terminology of Van Valin
1987 and this volume). This corresponds to three types oí juncture, accord
ing to which layers are structurally joined. The intersection of nexus and
juncture yields potentially nine types. A syntactic bondedness hierarchy,
later called inter clausal relations hierarchy, is developed (Foley & Van
Valin 1984:267, Van Valin 1984:555, this volume), embodying the assump
tions that juncture takes precedence over nexus and that cosubordination is
a tighter linkage than subordination. It was originally thought that subordi
nation is absent from nuclear junctures, but this possibility is now recog
nized by Van Valin (this volume). We have then nine types of linkage, var
ying from weakest clausal coordination through intermediate core subordi
nation to tightest nuclear cosubordination. The application of these
categories to Nootka seems to clarify certain relationships, especially as
regards cosubordination at all three levels of juncture.