ARGUMENT LINKING IN DERIVED NOMINALS 405Finally, in (46d), the interpretation of the direct argument of the plural
result vN, investigations, is ambiguous between an A and an U reading. It
is possible to interpret the subcommitee either as the A performing the
investigations or as the U undergoing the investigations. In the case of the
A interpretation, the construction behaves as is expected for ACT-based
vNPs (or result nominals derived from iteratively-used ACMs) ; in the case
of the U interpretation, the construction behaves as is expected for ACM-
based vNPs.
These constructions strongly suggest that the same susceptibility of
activities to accomplishment interpretations which is found in the clause
occurs also in the vNP. This would explain why the (46a,b) constructions
instantiate the ACM pattern of direct argument selection and permit a
bounded interpretation wherein completed may be used to describe the
event, while the (46c) construction instantiates the pattern predicted by
RRG for ACTs and permits only an unbounded interpretation wherein
completed cannot be used to describe the ongoing state of affairs. Such sus
ceptibility would further explain why the direct argument of the plural
result vN in (46d) is ambiguous between an ACT-Α and an ACM-U read
ing.
Notably, where the NP Sgt. Brown is the direct argument (cf. 46b, c),
it is interpretable as the CL-A only where it is followed by the indirect y
argument in (46c). In the absence of the y argument (cf. 46b), Sgt. Brown
takes an U interpretation. This is not surprising in light of the numerous
interacting factors discussed up to this point: the susceptibility of ACTs to
ACM interpretations, the inherent syntactic "intransitivity" of vNs, and the
conflicting macrorole choices RRG predicts for the direct-argument linking
of ACTs (select CL-As) vs. ACMs (select CL-Us). Given these interacting
factors, it is not surprising that in the absence of the disambiguating into y,
the default interpretation of investigation falls to the class of verbal bases
which refer to bounded events, rather than unbounded states of affairs. In
naming nondynamic and bounded events, nominalizations interpreted as
ACMs are more easily reified as "things,"^23 placing them closer on Ross's
verb-noun cline to prototypical Ns like spatula (cf. discussion in 1.1) than
the more V-like unbounded and dynamic states of affairs denoted by Ad
verb nominalizations.^24
Because the y argument of investigation can be marked with a preposi
tion other than the direct-argument of, an obvious hypothesis is suggested
by the behavior of this vN: in general, only where the y argument of a two-
argument activity can be prepositionally marked, permitting it to follow the