362 LAURA A. MICHAELIS
ing" and "depriving" are distinct from one another in the following impor
tant respect: in the former instance the ablative argument — or, more fre
quently, the ab-headed PP requiring an ablative NP complement — repre
sents a location (2b); in the latter case it represents a theme (15a, e.g.).
Further, no semantic difference appears to justify the separation of "verbs
of giving" from "verbs of supplying". Case-pattern differences alone seem
to provide the entire motivation for this semantic division: in the former
case, the theme is coded by an accusative argument; in the latter case, the
theme is coded by ablative argument. It is certainly not clear that these dis
tinct coding possibilities reflect distinct semantic verb classes. Indeed, Pink
ster acknowledges (p. 171) both that the classes of verbs of "giving" and
"supplying" are "semantically rather close" and that such variable-valence
transfer verbs as circumdo ("I surround") and dono ("I give"), which can
appear with both ablative and dative third arguments (2a, 14c), defy the
apparent dichotomy of case-patterns based on verb class. Variable valence
and the existence of the two case-patterns among semantically similar verbs
Pinkster eventually attributes to an animacy division, viz., the dative marks
animate third arguments, while the ablative marks inanimate third argu
ments.
This animacy division, Pinkster argues, is well supported by three-
place predicates. He concedes, however, that "...the clear-cut division
found with third arguments of three-place verbs...is less prominent with
two-place verbs (p. 174). The ablative, again, almost always marks inani
mate things, but the dative is used both for animate and inanimate entities,
with only a slight preponderance of animate beings." He nevertheless
decides to uphold the animacy division on the following grounds: as men
tioned in section 2, there are verbs which can govern either a dative or
accusative, with a concomitant change of meaning (the example of consulo,
"I consult", is given in fn. 2). The dative, Pinkster argues, is used only
when the object-referent is animate. This, however, is a line of argumenta
tion whose validity, as noted in fn. 2, is in later work eclipsed by Pinkster
himself (to appear: 8):
...there is [apparently] no opposition between the accusative and dative
case, but [it appears that] they mark different types of constituents: either
[the verb in question] is a two-place verb governing a normal accusative as
its second argument and the dative constituent is a benefactive satellite, or
[it] is a three-place verb with both an accusative argument and a dative
argument. In specific contexts, either one or the other, or both, may be