32 ROBERT D. VAN VALIN, JR.
icate focus) and (12b) (sentence focus) are given.^16 Unlike English, the
PFD is not coextensive with the clause, and all prenuclear elements in these
sentences are topical. All three projections of the English sentence What
did John give Mary yesterday? are given in Figure 14a.
It was noted in section 1.2 that there is no analog in the LSC to the VP
grouping which is basic to X-bar analyses of clause structure. However, a
glance at the predicate focus constructions in Figures 12 and 14 reveals that
the focus domain in these constructions corresponds to what would be a VP
in an X-bar analysis. This is no accident. VPs, to the extent that they exist
in languages, are nothing more than the grammaticalization of the comment
in the topic-comment predicate focus construction; they are not primitive
categories in clause structure. All languages have predicate focus construc
tions, the universally unmarked focus structure, but the same cannot be
said for VPs. (Cf. e.g. Van Valin 1987a, Mohanan 1982.) Lambrecht (1988)
also shows that there is no evidence for an NP-VP bipartite structure in sen
tence focus constructions cross-linguistically. The clearest syntactic evi
dence for a VP-like category in English comes from imperatives, so-called
"VP fronting" and "VP anaphora", as in (18).
