Relators and linkers: The syntax of predication, predicate inversion, and copulas (review)

Marcel den Dikken's book constitutes an outstanding contribution to the syntactic study of predication, a topic that has been on the research agenda of generativist syntacticians since at least Higgins's 1979 groundbreaking dissertation and Williams's 1975 and 1980 foundational articles and that has given rise to an impressive body of work: Heycock 1994 and Moro 1997, 2000 (to which D steadily and extensively refers), but also Rothstein 1983, Stowell 1983, Heggie 1988, Napoli 1989, and Déchaine 1993 (which are also discussed), among many others. What sets D's book apart is that he manages to bring together a broad range of intriguing and challenging syntactic data relating to predication and copular constructions, some of them new, others already known but never before considered in this perspective, under an original and impeccably organized analysis. This analysis is built around two simple ideas. The first is that 'all predication relationships are syntactically represented in terms of a structure in which the constituents denoting the predicate and the subject are dependents of a connective or RELATOR that establishes the connection—both the syntactic link and the semantic one—between the two constituents' (11). This claim, which at first sight only recapitulates a well-established traditional conception, is less innocuous than it seems. It must be interpreted as meaning that the proposal that 'the RELATOR accommodates the predicate and the subject in its MINIMAL DOMAIN' (12) is ALL that the grammar has to say about predication relations. They are both configurational and hierarchically asymmetrical, that is, defined in X'-theoretic terms, AND nondirectional, contrary to what Rothstein 1983 and Déchaine 1993 propose. There are thus two ways of linking a predicate to its subject: either the predicate stands in the complement position of the RELATOR, as is standardly assumed in the small-clause approach to predication, or it occupies its specifier position. Reverse-predication constructions, which instantiate the predicate-specifier structure, should thus be distinguished from straight-predication ones, which exemplify the predicate-complement structure.

The second core idea of the book is no less innovative than the first. D adopts Moro's 1997 proposal that predicate-inversion constructions involve A-movement of a null-headed small-clause predicate around the subject. In a wide variety of constructions, well-formedness appears to be dependent on the projection in the tree of a LINKER element, surfacing between the inverted predicate and the subject. The gist of D's proposal is that the projection of the LINKER should be viewed as the result of the syntactic constraints imposed on the inversion process. Its function is to extend the domain of predication beyond the RP projection and to make a specifier position available to host the raised predicate.

D insists that the RELATOR should be distinguished from the other elements that have been taken to provide the connection between a subject and a predicate, in particular from Bowers's 1993 Pr head, a 'new functional category' according to its proponent. D's RELATOR is an 'abstract functional head', a 'placeholder' for any functional head in the structure that mediates a predication relation, not a designated category. That head can be a copula (The earth must be round), a prepositional element (They regard him as a strong president; an idiot of a doctor), T (Imogen fell), or a head in the periphery of the sentence, such as Topic (Brian, Imogen really adores: a RELATOR of some sort is needed to connect the initial DP topic and the TP comment). The question arises of whether v should be included in this set. D's answer is that there is no need to introduce a light verb v to mediate the relation between the verbal predicate and its subject in unaccusative constructions, because the T-head suffices to achieve this result, but that wherever it occurs, in particular in transitive constructions, v relates a predicate to a subject, even if its presence...