OF NOMINATIVES AND DATIVES 493Chinookan Antipassive Formation:
Figure 12same hierarchy-bound fashion when the possessed noun is a human. The
result is an ergative pronominal for possessor in first position (compare the
ergative for underlying "D" in the inverse transitive) followed by an absolu-
tive pronominal in second position for the possessed (compare the absolu-
tive for underlying "O" in the inverse transitive). Yet a third nominative/
absolutive : dative schema occurs in the antipassive transform of the plain
inflection, where, for a regular transitive "A" — "O" schema usually
expressed with the split ergative : absolutive surface case-marking, the
antipassive codes the Agent in the surface dative case-marking pronominal
position, as shown in Figure 12. Note that if there is a lexical dative, an
indirect object, present, this is deleted obligatorily, replaced by the gram
matical dative of Agency. Two antipassive forms are found, a verbal or
predicating form in which there is no possible expression of the Patient, and
a nominal form in which the Patient is expressed in the nominative/absolu-
tive pronominal basically, this resulting two-position schema being subject
to the global alternation of form (see Figure 11) just like any possessed
noun stem.
It turns out that the "inverse transitive" inflection of some finite predi
cates is identical in properties to the antipassive form, and in turn to the
nominalized antipassive and other possessed noun forms. In fact, generaliz
ing the conditioning environment of the global ergative alternation over all
these constructions, we have done most of the work of case-marking in
Chinookan. To yield full plain inflections, we must simply prepose any "A"
or "S" coding pronominals to the left of the whole complex (in a sense, "to-
picalizing" them), so that the antipassive — "inverse transitive" — posses
sed noun inflectional schema with nominative/absolutive : dative can be
seen as a more remote underlying coding of all possible Chinookan inflec
tional morphology.