Committee: Alec Marantz (chair), Stephanie Harves, Richard Kayne, Pieter Muysken (Radboud University, Nijmegen), Anna Szabolcsi
This dissertation is fundamentally about the place of thematic roles in the architecture of the grammar. One contemporary viewpoint takes thematic roles to be formal features present in the syntax, analogous to phi-features and abstract Case features, which must be checked via (internal or external) Merge in the course of the derivation (the movement theory of control of Hornstein 1999 et seq. relies on such a view). A different view, visible in work by Heim and Kratzer (1998),Schäfer (2008), Marantz (2009, 2013), Wood (2012), Bruening (2013) and many others, takes it that thematic roles are not syntactic entities at all, but are rather determined post-syntactically, in the semantic component. From the perspective of Distributed Morphology (Halle & Marantz 1993), this latter perspective gives rise to some intriguing expectations concerning parallels with phenomena at the PF interface. First, we expect that the denotations associated with certain terminal nodes could vary depending on the surrounding syntactic structure, just as the PF realizations of terminal nodes can vary in a manner determined by the syntactic context (i.e. we should find conditioned “allosemy” at the LF interface, analogous to conditioned allomorphy at PF). Secondly, just as there are zero morphemes at the PF interface (i.e., terminal nodes which are effectively ignored by the phonological component), we expect to find that terminal nodes can be “semantically zero”in certain circumstances, effectively being ignored by the interpretive component.
The main argument of this study is that a fresh look at the typology of how possession sentences are built and interpreted in natural languages provides novel evidence that both of these expectations are correct (reinforcing conclusions reached on different grounds by Schäfer 2008; Wood 2012; Bruening 2013; Marantz 2013), thus supporting a view of argument structure in which thematic roles are determined post-syntactically in a way mediated by the conditioned allosemy of certain argument-introducing functional heads. It is shown that such a perspective yields novel solutions to two major puzzles in the syntax and semantics of possession sentences- what I call the too many meanings puzzle (why do possession constructions across languages have the ability to convey a myriad of unrelated meanings,like kinship, body parts, permanent ownership, abstract attributes, etc.?), and the too many (surface) structures puzzle (why do possession constructions vary so much in their surface syntax across languages, from transitive HAVE constructions,to existential BE constructions containing an oblique possessor, to copular BE constructions containing a PP possessee?; Why is the transitive HAVE pattern relatively rare?). The post-syntactic conception of thematic roles has a consequence that proves to be a crucial advantage in solving both of these puzzles: whether a given argument-introducer takes a complement or specifier is determined in the syntax, but whether it assigns a thematic role or not is determined in the semantics.
This means that the notion of “syntactic argument of head X” is potentially independent of the notion of “semantic argument of head X”, and this independence is key to understanding the syntactic structures of possession sentences and how they give rise to “possessive” interpretations. The theory is illustrated via two main case studies- (i) a micro-comparison, based on original fieldwork, of the syntax of possession constructions in Cochabamba Quechua (a BE-language of Bolivia) and the closely-related Santiago del Estero Quechua (a HAVE-language of Argentina), and (ii) a unified syntactic and semantic analysis of the uses of have in English (and similar verbs in other European languages). Other chapters offer a review of generative and typological work on possession sentences since the early 1990s, and extend the analysis to the broader typology of possession constructions as uncovered in recent generative and typological literature.