ON DEVIANT CASE-MARKING IN LATIN 365for example, that noceo ("I harm") was a one-place predicate at any stage
of its development. Further, although this analysis may be capable of exp
laining why a non-accusative "object" is dative as opposed to ablative
(perhaps in terms of whether that argument was originally a satellite coding
a source — ablative — or beneficary — dative), it cannot apparently
account for either the presence of genitive objects or the synchronic alter
nation between ablative and genitive object-coding among, e.g., stative
verbs of abundance. Further, although Pinkster's more recent analysis does
not rule out "semantic justification[s] for the use of specific cases," he does
not propose again the animacy dichotomy suggested in his 1985 analysis.
This dichotomy, despite the flaws enumerated above, did account for both
the use of the dative to code indirect and non-accusative direct objects and
the existence of case-pattern parallelisms between certain causative and
quirky stative predicates. This more recent analysis, abandoning that
dichotomy, does not appear to explain these facts. Finally, it seems that this
analysis, like that previously discussed, contains no account of the strong
correlation between quirky case and impersonal passivization, a correlation
which Pinkster attempts to explain away in the manner discussed in the pre
vious section. It seems that for Pinkster, although such verbs as faveo may
have been intransitive at some preliterary stage, they are transitive from the
point of view of Classical Latin, despite their lack of an accusative object.
Hence it seems that Pinkster's analysis would make the prediction that
verbs characterized by deviant case-patterns, like "normal" verbs, form
personal passives — a prediction which we know to be false. An example
adduced by Pinkster (p.4) in which the "quirky verb" noceo ("I harm")
apparently sanctions a personal passive seems to indicate a regularization of
the active rather than any correlation between the quirky case-pattern and
the personal passive.
At best, it seems, the correlation between non-accusative-object verbs
and the impersonal passive would have to be treated here as a sort of coin
cidence, and handled by fiat. As an explanation for this correlation falls out
naturally from the RRG account, it seems that one should favor it. It
remains to be shown, however, in what way the present account represents
an improvement over a third account, which appears to handle the facts of
Latin impersonal passivization with equal ease: that of Jensen (1983).