ON DEVIANT CASE-MARKING IN LATIN 347Each of the four Aktionsart classes is represented among those verbs
taking dative objects. Statives include the inverse verbs, e.g., libet; these
differ from other statives taking dative objects, e.g., fido, in that the
nominative argument represents not a cognizer (i.e., locative) but a cog
nized item (theme). The RRG analysis of inverse verbs is straightforward.
"Irregular" statives like libet, having one macrorole, must assign that mac-
rorole U status, as there is no activity predicate in the LS. Of the two
thematic roles connected with these Stative predicates, locative and theme,^4
theme outranks locative for U; the locative, as a non-macrorole core argu
ment, will accordingly appear in the dative. The theme, as the single mac-
rorole-bearing core argument, is given PrP status, and hence appears in the
nominative.
We now examine the application of the case-marking principles to
verbs like fido, non-inverse statives taking a nominative and dative argu
ment.^5 Such verbs require a marked linking of U to the thematic role loca
tive. The cognizer (locative) is PrP, as evidenced by its nominative case,
while the cognized item (theme) is a non-macrorole core argument, as evi
denced by the fact that it is dative. According to the A/U hierarchy, how
ever, theme outranks locative for U, and it is therefore the theme which
would be predicted to occupy the role of PrP. Hence, this predicate can be
said to license a marked linking, and this fact should be noted in its lexical
representation.
Unhappily, however, the fact that fido allows this marked linking
appears to undermine the validity of coding principle (26), viz.: ablative or
genitive case-marking of "objects" signals the marked linkage of locative to
U. Such predicates as fido require the marked linkage, and their objects
appear in the dative. This fact is actually unproblematic if we recall that
dative case-marking is, as codified in the case-asignment principles of (24),
the standard means by which a non-macrorole core argument is coded. As
mentioned earlier, the dative case-marked argument is by far the most com
mon type of "deviant object", licensed by predicates representing each
Aktionsart class. In contrast, ablatively and genitively marked "deviant
objects" — with the as yet undiscussed exception of activity verbs denoting
use — occur only with those statives and causatives of lack, abundance, and
recollection which require the aforementioned marked linkage.
Because the dative represents the default coding of non-macrorole
direct core arguments, it should be no surprise that dative deviant objects
can occur with any verb having "quirky transitivity", even when the marked