Verb first: On the syntax of verb-initial languages (review)

The idea that verb-initial languages might constitute a uniform set has been discredited at least since McCloskey 1996, and this volume amply demonstrates that, indeed, they do not. A brief introduction by the editors precedes sixteen papers covering a variety of Austronesian, Celtic, and other languages that have been claimed to be verb-initial (in some way). The book is divided into two parts.

Part 1 looks at ‘VP-movement vs. head-movement’, and a good start is made by S andraC hung’s ‘What fronts? On the VP-raising account of verb-initial order’. Many analyses inspired by the research program initiated by Kayne (1994) have suggested that (some) V-initial languages are derived by VP fronting. Chung systematically works through the kind of evidence that would have to be adduced in order to test this claim. She suggests that if whole (or remnant) VPs do front, then, like other moved constituents, they should (i) raise across clause boundaries, but (ii) not permit movement from within themselves—in other words, VPs should be islands. The first test is definitely failed by VPs: they do not raise across clause boundaries. The results of the second test prove more equivocal: in only some of the languages investigated are VPs islands. Chung then asks which independent principles would force VP raising. In some languages, the properties associated with the Spec, TP position are apparently not confined to subjects, but may apply to VPs, too. The extended projection principle (EPP) will therefore play a role in forcing VP raising. The paper is carefully argued and suggests directions for future research.

Along with the V-fronting and predicate-raising classes of verb-first languages, there are at least two other ways to derive the observed word order: subject-lowering, as established by Chung for Chamorro, and, as proposed by henry davisin ‘Coordination and constituency in St’át’imcets (Lillooet Salish)’, via ‘right-conjunct extraposition’. Davis argues that St’át’imcets, which displays a VSO/VOS alternation, is conventionally configurational, with arguments in argument positions rather than adjunct positions à la Jelinek. There is clear evidence for a VP (i.e. excluding the subject). Davis shows that superficially, one might think that subject lowering was the right analysis of St’át’imcets, too, but coordination evidence proves this to be incorrect. The theoretical implications are interesting, because no current model captures the generalizations (57). Again, this is a carefully argued paper.

Y ukoO tsuka, in ‘Two derivations of VSO: A comparative study of Niuean and Tongan’, again makes it clear that V-initial order arises differently in different languages, even in these two closely related Polynesian languages. Otsuka agrees with Diane Massam that in Niuean, the derivation involves VP-remnant movement, but argues that in Tongan, it is V-to-T-to-C movement. The evidence involves the distribution of clitic pronouns and scrambling. Note that nonverbal predicates are also initial in Tongan, an order accounted for by proposing that such predicates are also heads, rather than XPs, so like verbs they undergo movement to C.

F eliciaL ee, in ‘Force first: Clause-fronting and clause typing in San Lucas Quiaviní Zapotec’, argues for both the overt and covert raising of whole clauses. Various question and other particles appear in an expected clause-final position, despite the strong head-initial character of the language. Lee argues that these particles represent Force heads, and are generated in a head-initial position, but that the surface order is derived by fronting the remainder of the clause to the specifier position of ForceP. This analysis is supported by the fact that very similar question particles in the language do have a sentence-initial position. Lee concludes that ‘clause typing’ (Cheng 1997) in this variety of Zapotec is realized either via question particles, which are base-generated...