Advances in Role and Reference Grammar

88 ROBERT D. VAN VALIN, JR.

In both of these languages, there are constructions in which actor and

potential actor NPs do not receive their canonical morphosyntactic treat­

ment, and in some of them they retain their pivot properties and in others

they lose them. The difference turns crucially on whether the constructions

involve preventing an argument that qualifies to be an actor from being

assigned actorhood, a lexical phenomenon in terms of Figure 21, or

whether it entails assigning an actor to a non-canonical (normally

peripheral) syntactic status, a syntactic phenomenon in terms of Figure 21.

Thus the lexical-syntactic opposition, which falls out naturally from the

RRG linking schema in Figure 16, is strongly supported empirically.

5.4 Focus structure and linking

In the linking diagrams in Figures 17-20 the linking procedure appears to

relate the semantic LS representation to a simple LSC syntactic representa­

tion, but this is in fact misleading, for it omits a major factor in the determi­

nation of the form of the sentence: focus structure. A more accurate rep­

resentation of the linking relation is given in Figure 22, which includes all