Search Results

This chapter introduces the two major puzzles posed by possession sentences: the too-many-meanings puzzle and the too-many-(surface)-structures puzzle. The too-many-meanings puzzle is that languages often use the same construction to speak of having a car, having a sister, and having brown eyes, amongst other things, despite these relations being notionally distinct. The too-many-surface-structures puzzle is that languages differ radically in the argument structure used to convey the same possessive meanings. The chapter goes on to lay out the theory of the architecture of the grammar assumed in the main body of the book, and sketches how that architecture gives rise to a solution to both puzzles. A particularly important facet of this solution is the idea that a head which introduces a thematic role in the semantics might fail to take a specifier in the syntax, causing the relevant role to be saturated higher in the structure (Wood 2015)—a circumstance this book refers to as delayed gratification, and which turns out to be commonly attested in the typology of possession sentences. The core predictions of the present approach are presented. A concluding section summarizes the structure of the rest of the book.

This chapter aims to summarize previous approaches to the too-many-meanings puzzle and the too-many-surface-structures puzzle, providing a uniquely detailed and up-to-date appreciation of recent literature on possession. It begins by setting out the scale of the puzzles by examining functional-typological work. The second section introduces the standard generative approach to the too-many-surface structures puzzle, which I refer to as the Freeze/Kayne tradition. This approach, associated with Freeze (1992) and Kayne (1993) (but with antecedents going back much earlier), proposes that the vast surface diversity in possession constructions is to be derived via movement from one or two underlyingly identical structures. The third section looks at extensions of and reactions to the Freeze/Kayne tradition. In the fourth section, various approaches to the too-many-meanings puzzle are discussed, much of it from the formal semantics literature. The main conclusions of this overview are (i) that the Freeze/Kayne tradition is correct to treat HAVE and BE as two realizations of the same element; but (ii) some of the surface differences among possession constructions involve real underlying differences in argument structure; and (iii) the meaning of possession sentences does not come from HAVE and BE themselves, but from other elements in the structure.

The approach of this book makes two important predictions different from those of the Freeze/Kayne tradition: (a.) possession constructions can vary in the place in the structure where the possessor is introduced, (b.) the different ways of building possession sentences permitted by (a.) could have somewhat different meanings, depending on the semantic contributions of the pieces that make them up. This chapter provides existence proofs that these predictions are correct, drawn from new fieldwork data on two understudied Quechua dialects. Prediction (a.) is supported by an analysis of two possession constructions in Cochabamba Quechua, dubbed the BE construction and the BE-APPL construction, which differ precisely in where the possessor is introduced into the structure. Prediction (b.) is supported via a comparison of the BE-APPL construction in Cochabamba Quechua with a similar construction in Santiago del Estero Quechua. Both case studies suggest that the applicative morpheme does not introduce a thematic role of its own, a fact that has important implications for applicative theory. The chapter closes with some preliminary remarks on why Quechua languages vary with respect to whether or not they have HAVE.

Stassen (2009) points out a set of possession constructions in which the possessee appears as the predicate in a copular construction. In these constructions, the possessee appears to be marked by a derivational morpheme, often an adjectivalizer or nominalizer. An example is the English –ed in contexts like John is blue-eyed. Stassen dubs this phenomenon Predicativization. In this chapter, I show that Predicativization cannot be related to more familiar types of HAVE or BE construction by movement, further undermining the Freeze/Kayne tradition. This is shown via a detailed analysis of the –yoq suffix in Cochabamba Quechua, in which Delayed Gratification plays a key role. Drawing on and amending ideas from Nevins and Myler (2014, submitted), the chapter then lays out a detailed typology of Predicativization structures cross-linguistically. The main parameters of variation in this typology are (i) the category of the Predicativizing morpheme itself, (ii) the size of the nominal substructure it selects, (iii) whether it requires modifier, and (iv) whether or not the modifier can be phrasal.

This chapter sketches an analysis of WITH-Possessives. It is argued that the analysis of Levinson (2011) is correct in its essentials, and that Levinson’s approach can be broadened to account for other subtypes of possession structure beyond temporary possession. A comparison of the Icelandic WITH-Possessive with similar constructions in Bantu languages reveals an interesting difference. While the Icelandic WITH-Possessive is semantically restricted in a way that suggests that Icelandic með ‘with’ retains its comitative semantics in possessive contexts, Bantu WITH-Possessives are not restricted in this way. I analyse the Bantu situation as a case of grammaticalization, understood as a subtype of reanalysis along the lines of Roberts and Rousseau (2003).

This chapter begins by giving a brief overview of the arguments made in the main body of the book. It then examines prospects for extending the approach to other constructions that involve HAVE and BE beyond possession, including aspectual auxiliary constructions and existential constructions. After discussing some other open questions, the chapter concludes by highlighting that the findings of this book force us to adopt a certain view of the place of thematic roles in the architecture of the grammar: thematic roles are not features assigned to particular positions by the syntax, but instead constitute (part of) the meaning of certain syntactic terminal nodes, and are relevant only in the semantic component.