334 LAURA Α. MICHAELIS
higher-ranking actor of praeficio, ei ("they"), is realized in the clause (as
the complement of an afe-phrase). Hence, neither "equi" nor zero anaphora
appears to select its pivot on purely semantic grounds.
Zero anaphora is in fact regarded by Foley & Van Valin (1985:306) as
"[a] good grammatical frame of reference for distinguishing pragmatic from
semantic pivots." They state (ibid):
Zero anaphora is only permitted of highly topical referents, and the con
trol of this process by an NP type is a good diagnostic indicator that it is a
pragmatic pivot; provided, of course, that its selection is not predictable in
purely semantic terms. That is, if we find a language like Enga in which
zero anaphora is always restricted to actors, then the pivot is a semantic
one. However, [if]...both actors and undergoers [are] subject to zero
anaphora, [this proves that] the...pivot is a pragmatic one, rather than a
semantic one.
The presence of a pragmatic pivot in Latin does not, however, rule out the
possibility that semantically selected pivots are involved in clause-internal
grammatical processes. In fact, it will be argued below that in order to
account for the genitive "subjects" shown in (19), the specification of the
"raising to object" construction in Latin must be such as to allow SmPs
lacking PrP status to fulfill a core argument position in matrix and embed
ded clauses linked via core coordination. It will also be shown, however,
that where semantic and pragmatic pivots are represented by different argu
ments, PrP preferentially "raise".
Having motivated the assignment of PrP status to the nominative argu
ment in (24a), we might now turn to the case-marking principle governing
oblique arguments in (24). Principle (24c) requires that dative case be
assigned to any non-macrorole core argument which is not a prepositional
object. Thus, in (2b), a non-macrorole core argument coding a "source"
receives ablative case; the fact that it does not appear in the dative is pre
dicted by the fact that this argument is indirect — it is the object of the
preposition ab.
One apparent difficulty with this last coding principle must be dealt
with here: as stated earlier, "sources" need not always be coded by an ab-
phrase; they may also appear as direct ablative arguments. The latter type
of coding is, however, relatively uncommon, as one can discover through a
cursory examination of the entries in the Oxford Latin Dictionary for such
three-place removal verbs as separo ("I separate"), distinguo ("I distin
guish"), and removeo ("I remove"). In each of these entries, the preposi
tional-phrase coding of the source argument is far better represented. For