358 LAURA Α. MICHAELIS
Such a conception of verbs like pudet would, however, be problematic,
in that it would entail a violation of the principle of macrorole assignment
proposed in the "Synopsis", sect. 3.3.2, namely, that the single macrorole
with an intransitive verb is an actor if the verb has an activity predicate in
its LS; otherwise it is undergoer. Any accomplishment predicate, consisting
of an activity predicate linked to an achievement predicate by a causal
operator, would then assign actor status to any single macrorole. It is clear,
however, that if those verbs requiring genitive "subjects" license only a
single macrorole, as their case pattern suggests, this macrorole is an under
goer, as reflected by the accusative case of the single macrorole-bearing
argument. Granted, undergoer is not equivalent to object — undergoers, as
has been shown, can be subjects — but arguments bearing the actor mac
rorole apparently receive accusative case-marking in Latin only when
"raised to object", as in (20a). And although Van Valin (ibid) cautions
against literal construals of the macrorole labels, the argument receiving
accusative case — the metaphorical recipient of some emotional "force" —
is a fairly good example of the undergoer prototype suggested by Foley &
Van Valin (1984). How then can we explain the predilection of the "flip"
verb for an accusative (rather than nominative) macrorole-bearing argu
ment? The Latin "flip" verbs do constitute a special class — if the inverse
class of (3-5) is to be taken as typical, then a non-PrP cognizer is ordinarily
given dative, rather than accusative coding. If we should chose to view the
flip verbs as an exceptional class, we might then attribute to this exceptional
character their apparent defiance of the aforementioned principle of mac
rorole selection.
Of course, it may also be the case that the accusative argument
included in the case frame associated with this class, like that coding the
theme-argument of distransitive verbs (28a-b), does not in fact bear the
undergoer macrorole. Evidence bearing on this question comes from the
passives of verbs of this class. If the accusatively coded argument of, e.g.,
pudet ("shames") were a bona fide undergoer, one might expect that it
could be promoted to PrP status via that component of the passive linking-
rule licensing the "foregrounding" of undergoers. Such promotion does not
appear to occur. Where passive forms of these verbs are found (gerundives
like pudendus from pudet are common), their subjects code not the sufferer
but the cause of the emotion. An example of such an idiomatic gerundive
can be seen in (30):