ARGUMENT LINKING IN DERIVED NOMINALS 411(56) is the only restriction placed on preposing in vNPs headed by such vNs,
the direct arguments of these vNs are always preposable, although where a
vNP is headed by an -ing form, the resultant construction may be a gerun
dive nominal, rather than a vNP (cf. 57f,g).
(57) One-argument ACH and ACT nominals^29
a. The old manU died. (ACH)
b. the death of the old manυ dir arg
the old man's death
d. The planeU landed. (ACH)
e. the landing of the planeU dir
f. vNp[The (plane's) landing] was smooth.
8- ger. nom.[The (plane's)^30 landing on time] surprised Joe.
h. The old typewriterA performed as if it were new. (ACT)
i. the performance of the old typewriterA dir arg
j. the old typewriter's performance
(58) Two-argument ACH and ACT nominals ("0" = no macrorole)
a. The planeU arrived at gate nineø. (ACH)
b. the arrival of the planeU dir arg at gate nineø
. gate nine'sø arrival (of/by the plane )
d. the plane'SU arrival (at gate nineø)
e. The dogA attacked the childø. (ACT)^31
f. the attack of the dogA dir arg on the childø
g. *the child's ø attack (of/ by the dogA)
h. the dog'sA attack (on the childø)
2.3.4.2 Prenominally-permitted arguments with vNs capable of taking both
macroroles
In an analysis which is not based on the double genitive assumption, but
which is first concerned with the linking of arguments to the direct-argu
ment position, a question which immediately presents itself in regard to
argument preposing with vNs capable of taking both macroroles is the fol
lowing: are any restrictions placed on the prenominal occurrence of an indi
rect argument linked to a macrorole in vNPs which include a direct argu
ment? Since U's vN [of Adir,arg ] is not possible in English (cf. fn. 32), the
question posed above refers to constructions of the type A's vN [of Udir arg ].
Table 1 reflects that restrictions do apply in the case of Type A vNPs.