398 MARY L. NUNES(34) a. The love Jess has for that puppy is unshakable.
b. The desire that soldier has for a homecooked meal is great.
Some speakers are also able to get the possessive [of x] with certain mental
stative and envy subclass members, none of which can mark the y-ÑP with
for. For these speakers, the following vNPs are fine:
(35) a. the knowledge of Sue of the Bible
b. the envy of Jess of Ralph's car.^19
The acceptability of such constructions creates no problem for the theory,
since the two of s are semantically distinct and mark different types of con
structions — possessive and direct argument (cf. discussion of of markers in
2.3.1). In light of the fact that the possessive marker signals a strictly nomi
nal phenomenon, while the direct-argument marker marks a clausal entity
(the clausal undergoer), it is not surprising that the possessive construction
occurs closer to the vN head than does the direct argument construction.
One final point should be made regarding vNPs which include both a
post-vN possessor and a direct argument — i.e. constructions of the type
vNP[vN [of poss'r] [of U]]. In such nominals, the only argument which is
reflected is the U. The possessor NP is a strictly nominal entity and has no
LS-encoded thematic relation with the vN head.
Having recognized that the of χ phrases are in fact possessive, rather
than argument constructions, two data-untangling questions remain. First,
why must the for (of)^20 y phrase follow the of NPposs,r in possessive construc
tions, while vNPvN [of CL-U] constructions solicit the inclusion of a topical
LDP NP (cf. 37a and 38a)? Second, with the exception of idiomatic expres
sions like the hope of better days to come, why are members of the hope and
reverence subclasses unable to mark (at least easily) their y arguments with
of, as is exemplified in (36)?
(36) a. Joe's hope for/?of a new car
b. Bob's reverence for/*of the priest
Simply put, the first question is answered in terms of disambiguation; the
second question is answered, in part, by looking at the clause.
Turning, first, to the notion of disambiguation, notice, on the one
hand, how ambiguity in (37a) and (38a) between an argument vs. a posses
sive reading arises in constructions which include neither a topical LDP NP
nor a postnominal for (of) y phrase. On the other hand, notice how the
ambiguity is resolved when either the LDP NP or the for (of) y phrase is
included: