ARGUMENT LINKING IN DERIVED NOMINALS 399(37) #The love of the child is unshakable.^21
b. The parents' love of the child is unshakable.
c. The love of the child for the puppy is unshakable.
(38) . #Th admiration of Sam is sincere.
b. The students' admiration of Sam is sincere.
c. The admiration of Sam for the dean is sincere.With these nominals, the of immediately following the vN may mark either
a possessor or a vN direct argument, creating ambiguity where the addi
tional NP is not included. This would not be the case, of course, if these
state vNs were treated as alienable possessions like book in the book of
Sam's, where Sam is marked with both the possessive of marker and the
genitive 's. As inalienable possessions, however, these state vNs do not
take such doubly marked possessors — *the love/ admiration of Sam's.
Hence, only the inclusion of the second x/y-NP permits the vN of NP to be
disambiguated between possessive and argument readings.
The second question raised by the state data — why do vN members of
the hope and reverence subclasses prefer to mark their arguments with
for? — can be answered in the case of the hope subclass by looking at the
clause. The verb sources of these vNs belong to a group of two-place sta-
tives which have a propositional argument, realized either as a clause (Joe
hoped [that he would get a Mercedes]y) or as an NP marked by non-predica
tive, purposive for (Joe hoped for a Mercedesy) (Jolly 1987: 110-111). In the
vNP, NPy arguments of derivative vNs simply inherit the for marking.^22
Why the vN derivatives of revere verb types are restricted to the same
for marking in the vNP is not clear. In fact, exactly how this for functions in
the vNP and why some two-place Stative vNs may mark their arguments
either as a direct argument or with for (the love subclass), while those
belonging to the reverence subclass are limited — where their verb sources
are not — to marking their arguments only with for are issues which must
be left for future investigation.
2.3.2.4 Activity-verb nominalizations
Activity-based vNs comprise the final class of vNs to be considered in terms
of the RRG direct-argument prediction. Since ACT verbs only take actors,
never undergoers, RRG predicts that derivative vNs will select arguments
linked to the CL-A as vN direct arguments. However, the prediction is trans
parently borne out only with vNs derived from single argument activities.