422 MARY L. NUNES
encoded PATs (cf. fn. 45) and EXPs, on the other hand, are themselves the
conditioning factors.
In the analysis presented here, the ergative orientation of the English
nominal is evidenced not only in the U > A linking hierarchy for direct
arguments, but in the mandatory presence of the U (or the argument
treated as an U with ACTs receiving a delimited ACM interpretation) in
any process nominal not capable of receiving a strict ACT interpretation. In
general, this requires that the U be present in the vNP as the direct argu
ment. However, with result-state ACM nominals (or performance object
ACTs, which may take an ACM result-state reading), the U may occur as
the topical LDP NP, and the nominal retains the process reading of its RS-
verb source (or RS-verb interpretation, in the case of the ACT perfor
mance objects).
In result nominals, too, the significance of the U is demonstrated by
the typical incorporation of its meaning into the result vN (where "result" is
given the narrow "thing" definition used throughout the preceding discus
sion). In fact, only two types of verbally- derived nominals seem to regu
larly have U-less result vNs: result-state nominals having a clear PAT (i.e.
an effected, rather than an effected PAT: e.g. destruction) and "concrete"
ACM experiential states (e.g. amusements — cf. fn. 42 and discussion in
1.2). Thus, only with these two types of result vNs, and with vNs having an
ACT component in LS and capable of taking an accumulated action
interpretation are Us regularly omitted from the vNPs included in the data
base.3. Concluding comments
As the analysis presented here has underscored, Anderson was clearly on
track in associating affectedness with preposability. In not having access to
a theoretical framework in which affectedness could be nonarbitrarily
defined, however, she was unable to give the notion a principled, non-intui-
tively applicable form. Significantly, in using the notion as a defense of her
structural manipulation, not only did she expose the inadequacy of the
structural framework in its application to syntactic relations in the English
nominal, but she inadvertently indicated the essential role of semantic rela
tions in explaining what the configurational framework's structurally
defined GRs could not — syntactic relations in the vNP.