Grammatical categories: Variation in Romance languages by M. Rita Manzini and Leonardo M. Savoia (review)

This new book by M. Rita Manzini and Leonardo M. Savoia presents a coherent picture of the authors' views on foundational questions concerning the morphosyntax of natural language. M&S's theoretical approach is best described by quoting one of the last few sentences of the book: 'Indeed, a key property in terms of which our approach differs from the others reviewed is its uncompromisingly lexicalist stance, whereby structures are projected from actual lexical terminals' (311). By this, they mean to contrast their theory with various approaches in which structures are built up from abstract functional heads with universally fixed content and possibly occurring in an invariant order across languages, as in the cartographic tradition emanating from Cinque 1999. Their book offers a large-scale reinterpretation of the conventional functional categories like complementizers, negation, auxiliaries, and case, based on a comparative study of Romance (largely Italo-Romance) and Albanian varieties. The general drift of the argumentation is that each of these functional heads is better understood as belonging to one or another of the lexical categories N, V, Q, and D. Variation is determined by the intrinsic lexical content of the items that lexicalize these categories. That is, crosslinguistic variation is reduced to differences in the lexicon. Another interesting aspect of their analysis is that the syntactic categories they postulate also occur inside units that are traditionally regarded as morphological words. The empirical material is mostly drawn from M&S's own fieldwork as reported in a number of their other publications and most notably in the monumental Manzini & Savoia 2005.

The analysis proposed in Ch. 1, 'The structure and interpretation of (Romance) complementizers', starts out from the observation that the k-based complementizers, such as Standard Italian che, are generally also used as interrogative words; for example, Standard Italian che also means 'what'. Since this identity in form is widely found across the range of Romance varieties, even though the actual morpheme found in one variety is not necessarily a cognate of the morpheme used in another, M&S conclude that it is not accidental. (This is reminiscent of the argumentation in Kayne 2010.) They conclude that a k-complementizer is a lexical item of category N that introduces a variable that can be restricted either by a proposition (che as a complementizer) or by a nominal (che as an interrogative in che cosa/libro 'what/what book'). Thus, the (Romance) complementizer is not a functional head on the verbal spine, but rather a regular lexical item, a noun taking a clausal complement.

This characterization of the complementizer leads to the expectation that the complementizer, like any other N, should come with its own 'left periphery' of phrases with discourse-related interpretations. Thus, the complementizer adds a second left periphery to the left periphery associated with the verb in its complement clause. M&S argue that this provides a better understanding of the distribution of clause-initial foci and topics with respect to the k-complementizer. More generally, they argue against accounts based on various cartographies of the left periphery emanating from Rizzi 1997. In section 1.3.1, they raise important issues regarding the relative merits of theories in which semantic relations like focus and topic are encoded in the syntax, as in the cartographic approach, or not, as in M&S's own theory. No implementation of the idea that interpretation is directly dependent on the relative order of phrases in the left periphery (rather than encoded by features) is offered, however.

The conclusions from Ch. 1 are corroborated by the data examined in Ch. 2, 'Variation in Romance k-complementizer systems', which deals with the complementizer systems of Italo-Romance varieties. One of the most interesting parts of the discussion in this chapter is devoted to the varieties that have a split complementizer system. Typically, the distribution of the two distinct complementizers depends on...