Reviewed by: Agreement restrictions Milan Rezac Agreement restrictions. Ed. by Roberta D’Alessandro, Susann Fischer, and Gunnar Hrafn Hrafnbjargarson. (Interface explorations 15.) Berlin: Mouton de Gruyter, 2008. Pp. 325. ISBN 9783110200652. $137. This collection of papers aims to present the current state of research on agreement restrictions. Unity or consensus is not sought; the independent contributions bring out both the achievements and the disagreements of the field. Ten of the eleven papers deal centrally with the restrictions known as the person case constraint (PCC, Bonet 1991) and the person restriction (PR, Taraldsen 1995), or with their superficially close kin. I comment on the papers in accordance with these themes. The remaining paper is Hamid Ouali’s ‘On C-to-T ϕ-feature transfer: The nature of agreement and anti-agreement in Berber’ (159–80). The PCC and PR belong to the family of argument interactions involving phi-features. In both restrictions, a dative prevents a first- and second- but not third-person argument from otherwise obligatory cliticization (the PCC in French (1)) or agreement (the PR in Icelandic (2)). 1. 2. The following issues emerge in current work (e.g. Anagnostopoulou 2003): • PCC/PR in syntax vs. morphology: Some agreement restrictions rely on surface morphology, but it does not differentiate the well- and ill-formed examples in 1 and 2, which appeal to more abstract structures (see papers by Arregi and Nevins, Rivero, Wiltschko). • Configuration where PCC/PR obtain: 1 and 2 suggest that the dative must c-command a first or second person that bears structural Case and needs to agree or cliticize (Anagnostopoulou, López, Richards, Sigurðsson and Holmberg). • Role of the dative: The dative does not control the same agreement or clitic series as the first and second person it affects, unlike in ‘primary’ object applicatives of the English or Mohawk type (Boeckx, Bonet, Richards, Sigurðsson and Holmberg). • (Non)repairs: Suspending cliticization or agreement does not automatically lead to the goodness of pronouns that otherwise must cliticize or agree, including in 1 and 2, but it sometimes does (Bonet, Sigurðsson and Holmberg). • PCC vs. PR: The PCC typically occurs in transitive sentences and affects clitics or agreement, while the PR typically appears in intransitive sentences and affects agreement. Other factors may or may not derive this, such as the absence of dative subjects in French and object clitics in Icelandic. In ‘Icelandic dative intervention: Person and number are separate probes’ (251–80), Halldór Ármann Sigurđsson and Anders Holmberg greatly advance our knowledge and understanding of the Icelandic PR. The core of the chapter shows that the PR involves syntactic intervention of the dative for person agreement with the nominative, and in some varieties also for number agreement. Remarkable support comes from elimination of the PR by moving the nominative past the dative. It is good evidence here for the key role of syntactic structure in the restriction, shored up by other movement effects on the PR and the PCC (Kučerová 2007, Rezac 2008, Karimi 2010, Baker 2011). The PR itself is attributed to a conflict of features coming from multiple agreement with the person of the nominative and the third person that all datives are assumed to bear (cf. Ormazabal & Romero 2002 on the PCC). Default agreement does not help, but syncretism of the exponents for the conflicting person values does (cf. McCreight 1988 on Case). The PR is therefore related to the morphological realization of agreement, whose interaction with syntax is left somewhat vague (271f.). Two issues seem essential to further exploration of this proposal. One is the status of the syncretism effect. It is reported that for most speakers, it eliminates the PR; yet Sigurðsson’s (1996:33ff.) study suggests instead that it ameliorates the PR to a degree that is commensurate with the amount of syncretism in a verbal paradigm. The other is the PR in nonagreeing infinitives. It is expected to be absent, but the evidence seems ambivalent for now (271, 276, n. 29; Boeckx 2003:ex. 24, Nomura 2005:§3.2.3, Bobaljik 2008:319, n. 27). Sigurðsson and Holmberg make use of one of the...