ARGUMENT LINKING IN DERIVED NOMINALS 423In fact, in LFG and configurational analyses (see Chapters 2 and 3 of
Nunes 1990 for discussion), some type of semantic justification is regularly
called upon to defend analytical stipulation — stipulation which is required
to prevent the theories' GFs or structurally defined GRs from ungrammati
cally associating certain vN arguments with particular syntactic positions in
the vNP. This clearly indicates that syntactic relations in the nominal can
not be fundamentally defined in terms of some type of grammatical form,
but must be defined in terms of semantic relations. Even in LFG, where
only semantically-restricted GFs are applied to the nominal, the GFs must
in some cases be manipulated via semantic-relations stipulations to prevent
ungrammatical NP-vN syntactic configurations (cf. earlier discussion of
Rappaport's analysis of experiential state vNs).
In RRG, no type of grammatical form is overlaid on semantic relations
in the English vNP. Semantic relations are used to define syntactic relations
directly, rather than being used to defend manipulations of structurally-
defined or thematically-regulated grammatical forms intended to define
syntactic relations. In combination with the fact that RRG does not make
the double genitive assumption, this means that the theory neither explicitly
nor implicitly presupposes any grammatical-form juxtapositioning in clausal
and vNP structure. Freed of both a grammatical form overlay and the dou
ble genitive assumption, an RRG analysis is able to investigate syntactic
relations in the English vNP in terms of the theory's notion of structural
layering (cf. Figures 1 and 2) and its semantic-relations constructs: nonar-
bitrarily-assigned macroroles (cf. the Α-U Hierarchy) and nonarbitrarily-
defined thematic roles. Within this framework, the foregoing analysis has
explained the linking of arguments to the postnominal of-marked direct
argument position in terms of the U > A hierarchy given in (8), and it has
explained the linking of arguments to the prenominal LDP in terms of the
hierarchy given in (55) and (74). This latter hierarchy has, in turn, been
explained in terms of the topical function of the nominal's LDP, a function
which in English appears to relate to argument affectedness (as defined by
principles of RRG theory) and animacy (broadly associated with NP argu
ments linked to the A macrorole and by definition associated with NPs
bearing the EXP thematic role).
Only further research and an extended application of this analysis to a
larger body of data will permit certain of its aspects to be more succinctly
and accurately stated. However, one thing seems clear: no grammatical-
form construct, such as the structurally-defined GRs of the configurational