As its title suggests, this book presents an analysis of the semantic effects of incorporation, a phenomenon most often studied from a syntactic or morphological perspective. Farkas and de Swart adopt Discourse Representation Theory and show how this framework allows them to capture a range of semantic aspects of incorporation in several different languages. In particular, they claim that their analysis offers a superior account of the differences between languages in terms of the discourse transparency of the incorporated nominal. I should point out that their analysis is reasonably accessible to linguists who aren't familiar with Discourse Representation Theory. I have only a rough idea of how the theory works, but I was able to follow the analysis without any problem (though my lack of familiarity does make it difficult to critically assess certain aspects of their analysis, as I make clear later on).
Chapter 1 begins with an introduction to the phenomena under consideration. The authors begin by stating that they include under the umbrella of "incorporation" both Baker (1988)-style head movement and Massam's (2001) pseudo incorporation. In other words, Farkas and de Swart are not concerned with the analysis of the syntax of incorporation. They note, however, that unlike van Geenhoven (1998), they do not consider all narrow scope indefinites to be semantically incorporated. Instead, they insist that incorporation is always morphologically marked.
Farkas and de Swart then provide an overview of the semantic particularities of incorporation that need to be addressed. First, there is cross-linguistic variation with respect to whether or not an incorporated nominal can serve as the antecedent of a pronoun in subsequent discourse (discourse transparency). Second, incorporated nominals always take narrow scope with respect to operators such as negation, quantifiers, modals, etc. Third, argument structure plays an important role (e.g., subjects of individual-level predicates never incorporate, and the incorporated nominal acts more like a predicate modifier than a true direct object, allowing for "doubling" in some languages). Fourth, although incorporated singular nouns are always unmarked for the singular/plural contrast, incorporated plurals [End Page 69] (in languages that allow them) are interpreted as plural. Finally, incorporated plurals are more discourse transparent than singulars.
Chapter 2 sets out Farkas and de Swart's modified version of Discourse Representation Theory. Their core modifications concern how discourse referents are introduced, the representation of thematic roles, and the nature of plurals. In "standard" Discourse Representation Theory, a DP such as the student introduces both a discourse referent and a descriptive condition. Farkas and de Swart, however, claim that determiners introduce discourse referents. Moreover, they claim that predicates are n-place relations (this is the representation of thematic roles). In order to fill in arguments of a predicate, the authors define a process called instantiation. A(rgument)-instantiation replaces a thematic argument of a verbal predicate with a discourse referent. D(eterminer)-instantiation replaces the thematic argument of a nominal predicate with a discourse referent and indexes the thematic argument with the discourse referent. After D-instantiation, the Discourse Representation Structure (DRS) for A student left would be as in (1) (after the determiner a has introduced the discourse referent u and this referent has A-instantiated the argument of leave):
The analysis of plurals is a little more complicated. Farkas and de Swart argue that the plural feature on a noun introduces a presupposed discourse referent. It is possible to resolve this presupposition by binding (in case there is a determiner that introduces a discourse referent) or by accommodation (in the case of bare plurals). This use of presupposition for plurals will be crucial in the authors' account of the discourse transparency of incorporated plurals.
Chapter 3 contains the core analysis of incorporation. In order to account for the special semantic properties of incorporation, Farkas and de Swart propose a new process of combining a nominal argument and a predicate: unification. The result of this process is a complex predicate, where...