126 Julia Horvath
immediately for this proposal is of course: If so, how come instances of other case-
assigning overt heads, such as Vs and Ns governing oblique case, do not have the same
option available under (14) (as shown e.g. in (7) and (8) above)? Specifically, why can
V and N not satisfy the Generalized Case Realization Requirement (14) by virtue of
being overt heads assigning their oblique case feature, as does P?
A most straightforward interpretation of these facts is that they indicate that V and
N, in contrast to P, do not possess oblique case features. From this it directly follows
that V and N could not possibly satisfy the oblique case realization requirement for
an undeclined noun phrase complement. It also entails that it must be some element
other than V or N that assigns the oblique case in examples which under previous
accounts were assumed to involve V-governed or N-governed instances of oblique
case. What could then be the source of the oblique case feature in such examples?
Having found that P but not V or N can possess oblique case features, the obvious
conclusion is that in these cases too it must be P, namely, a phonologically null P that
acts as the oblique case assigner. Such null Ps, just like their overt counterparts, would
be heads selected by particular verbs or nouns, and would assign “oblique” case to, i.e.
be copied onto, their noun phrase complement. Notice that the existence of phono-
logically null instances of various functional categories is a common state of affairs, as
attested by determiners, tense morphemes, among others; thus the occurrence of null
Ps can hardly be seen as unexpected or exceptional. Let us then consider how under
the proposed account, violations of the oblique case realization requirement (dem-
onstrated in (7) and (10b)) would arise precisely in cases stipulated by W&Z’s Case
Realization Condition (9).
On our account what appear to be oblique noun phrase complements of V or N
(and what earlier literature claimed to involve V-governed and N-governed oblique
cases) are analyzed as complements of phonologically null Ps, selected by the V or N
head. The reason why these complement noun phrases must have an element within
them morphologically realizing the oblique case assigned is now straightforward. It
is precisely because although the P possesses – and under Pesetsky’s (2013) particu-
lar theory, crucially copies onto the complement noun phrase the relevant categorial
feature (the “case” feature P) – but being non-overt (null), such a P-morpheme cannot
satisfy the Generalized Case Realization Requirement (14). This is in contrast to an
overt, phonologically realized P, which as suggested above, satisfies the requirement
(14). Thus, under the crucial assumption that “oblique case” is a phenomenon associ-
ated only with the category P and is not attested with any other categories, the prima
facie puzzling asymmetry between (alleged) V-governed and N-governed versus
P-governed oblique cases resolves. The apparent realization asymmetry reduces to the
simple distinction between overt versus non-overt (null) manifestations of the “oblique
case” feature (alias the category feature P in the assignment domain), which derives
directly from the case realization requirement attested independently in the language.