494 MICHAEL SILVERSTEIN
Chinookan Ascribed Habitual Relative Clauses (Coreferent NPs):
Deverbative nominalizations
Figure 13
Now, from the point of view of cross-clause linkage relations, it turns
out that the antipassive, as we might expect, plays the major role above a
certain tightness of linkage. Most of the clause linkage relations in
Chinookan are handled by simple sequencing of clauses plus deletion of
noun phrases for coreference (recall the inflectional pronominals always
remain on the constituents that govern the nouns). But for relative clauses
expressing ascribed habitual action (one who is supposed to/by nature can/
whose duty or role it is to do something or be some way), the language
shows nonfinite, nominalized normal forms, where transitive predicates
occur in antipassivized nominative/absolutive : dative construction, with the
dative pronominal position (i.e., the adnominal dative or genitive, which
globally alternates with ergative) being coreferent with the head of the
noun phrase and coding the "A" and "S" underlying case-relations, or
other case-relation derivatively regimented into this form, as in Figure 13.
Other case-relations are not handled with this construction, but rather with
a regular finite clause. So there is not only severe constraint on the linkage
tightness in Chinookan antipassivization, but there is the characteristic
restriction of what case-relations are available for coreference marking,
here "A" and "S".
4. The "focality" of nom/abs : dat
We have been pointing out the details of both clause-level and discourse-
level morphosyntactic facts of these two languages from the perspective of
a unified typological account, summarized in Figure 14. We can compare
the placement of the nominative/absolutive : dative case-marking schema
within the systems of inflection of these two languages now with a better
sense of the similarities and differences. In Djirbal, the normal form of