Mixed categories: The morphosyntax of noun modification by Irina Nikolaeva and Andrew Spencer

Abstract

Reviewed by: Mixed categories: The morphosyntax of noun modification by Irina Nikolaeva and Andrew Spencer Antonio Fábregas Mixed categories: The morphosyntax of noun modification. By Irina Nikolaeva and Andrew Spencer. Cambridge: Cambridge University Press, 2019. Pp. xi, 396. ISBN 9781108415514. $125 (Hb). A scientific theory is judged by its ability to provide a principled explanation of less straightforward facts that lie at the margins of the data normally discussed and captured by other theories. To the extent that those data naturally follow from the new proposal, a theory is successful in defining a new paradigm. Mixed categories, by Irina Nikolaeva and Andrew Spencer, is an excellent example of how to deliver a maximally explicit morphological theory in a way that complies with these scientific desiderata. As always, one can agree or disagree with some of the assumptions [End Page 951] made, but there is no doubt that this book makes a significant contribution to our understanding of grammatical categories and the relation between lexical and syntactic representations. In this respect, the title of the book might be a bit misleading in that it suggests that the book is devoted to a particular empirical phenomenon. That is not the case, even if the empirical phenomenon that names the book is discussed in depth and analyzed in a broad range of manifestations. This monograph goes far beyond the specific empirical domain that names it, as it defines in a maximally explicit way a full-fledged theory about the lexicon that addresses some of the foundational questions in morphology, from the divide between inflection and derivation to the integration of lexical information with syntactic structures. N&S present in this monograph an expansion of Spencer's generalized paradigm function morphology, an inferential word-and-paradigm theory that integrates Stump's (2016) paradigm function morphology with recursive function application in order to capture in a principled way both concatenative and nonconcatenative morphology. The theory itself challenges syntacticocentric approaches to word formation, such as Halle & Marantz 1993. But before I discuss N&S's contribution to these foundational issues, let us do justice to the phenomenon that the authors chose as title of the book by presenting their specific analysis. N&S follow Spencer 2013 in defining 'mixed categories' as 'lexical types that have not only some properties of the morphology, syntax, or semantics of one category, but also some properties of the morphology, syntax, or semantics of another category' (22). This categorial mixing is widely illustrated in the book, with examples that frequently have been obtained by the authors themselves, from a broad variety of languages. The Upper Sorbian possessive adjective bratowe in 1 (Corbett 1987) is a good example of a mixed category in the nominal domain (97). (1) The categorial mixing is visible in that the possessive adjective acts both as an adjective and as a noun. Externally, bratowe is an adjective in that it agrees in case and number with the head noun that it modifies (nominative plural). Internally, bratowe behaves as a noun that licenses the presence of a possessive determiner mojeho, which agrees with the nominal base bratr- in gender and number (masculine singular) and inflects for genitive case. For some syntactically inclined morphologists (such as myself), it has always been tempting to use these data to support a syntactic analysis of inflection and derivation. Roughly, the analysis would be to propose that the possessive adjective suffix -ow combines not with a morpheme 'brother', but with a syntactic phrase corresponding to 'my brother'. This dual projection analysis (discussed in detail by N&S in Ch. 4) would correspond roughly to 2. (2) N&S, however, show that this analysis cannot be right, specifically that the base of the possessive adjective cannot be a phrase (137). A coordinated base is, for instance, ungrammatical. (3) A dual projection analysis could try to argue that coordination is different from other phrases, but in any instance 3 shows that a phrasal analysis cannot be adopted without other (perhaps idiosyncratic) provisos. N&S then argue in detail that the solution for mixed categories is to be found in the lexicon, but in a very specific model of the lexicon...