Codeswitching worldwide Ed. by Rudolfo Jacobson (review)

BOOK NOTICES 205 retical topics addressed but are loosely united in their attempt to ask how well current versions of generative theory (either the principles and parameters framework or the minimalist program) fare in their treatment of linguistic phenomena in different languages . The emphasis of the volume rests far more on adjusting the apparatus of generative theory than attempting to discover the full range ofrelevant typological variation. As a whole, however, the book asks intriguing questions about how differences between languages might be unified with the proper syntactic theory. Hubert Haider (17-33) addresses the issue of head-complement ordering. He provides a number of empirical problems for the position promoted in recent years by Richard Kayne and Noam Chomsky. He offers an alternative view which generates verbs on the periphery of clause structure and then moves the verb, rather than its nominal arguments, to achieve the surface order of basic constituents. Anoop Mahajan (35-57) demonstrates that ergative case marking correlates with two other linguistic properties. First, an ergative system is restricted to verb peripheral languages. Second, ergative languages typically lack a distinct verb have. Mahajan proposes that the correlations are derivable from a deeper organizing principle of grammar. Specifically , he argues that the auxiliary verb have is actually an oblique form of the verb be and is created by adposition incorporation. In certain languages, the incorporation is blocked, so have cannot be derived. In these cases the adposition arises either as an ergative particle or affix. Chris Wilder (59-107) supplies an overview of ellipsis phenomena in English coordination. He identifies two primary types of ellipsis—backward and forward—and demonstrates constraints on their formation . He provides several arguments that ellipsis structures are not created by movement but rather are produced by phonologically empty elements generated at deep structure. David Adger (109-34) examines how the structural position of weak quantifiers is involved in their interpretation. Using data from Germanic languages and Scottish Gaelic, he suggests that the interpretation of certain weak quantifiers is accomplished by lowering at LF and that the lowering is governed by a restriction on derivations called effability. M. Rita Manzini (135-53) elaborates on her previous work on the notion of locality. She argues that all movement is subject to the same restrictions, essentially operating between adjacent minimal domains . This conception differs from work on relativized minimality which treats adjunct WH-movement distinctly from argument WH-movement. Manzini employs evidence from English WH-movement as well as polarity items in Italian and Albanian to exemplify how her definition of locality can unify various sorts of dependencies. Paul Law (155-79) focuses on subject/object asymmetries in Malagasy, which are seemingly opposite of what one finds in English. Law accounts for the Malagasy facts in terms ofageneral constraint on argument binding and justifies the constraint by demonstrating that it also accounts for a correlation between voice morphology and word order. Since Malagasy manifests a basic VOS constituent structure and has a three-way voice distinction (active vs. passive vs. circumstantial), both of which are typologically unusual, Law's article is particularly effective in pointing out types of variation that are not easy to account for undercurrent assumptions in generative theory. Elly van Gelderen (181-96) argues that there are strong features in AgrO which must be checked in the course of a derivation, either by an NP or by an expletive. She offers evidence from French, other Romance languages, Germanic, and Indie to support her claim. She also briefly raises the utility of her account for agreement in Uto-Aztecan but concedes that it only works well for languages like Hopi which have a relatively fixed word order. Jamal Ouhalla (197-218) discusses genitive subjects in Berber. He proposes that such subjects arise in clauses which contain the functional projection AGRgen whereas nominative subjects arise in clauses which lack this feature. In order to avoid complete circularity in his argument, he identifies certain word order facts which also receive some account under this analysis. Finally, Gaberell Drachman (219-48) argues that the syntactic properties of clitics provide evidence for different projections in clause structure: operator projections, agreement projections, and lexical projections. His primary evidence...