Wednesday, March 23, 2005

composition of regular possessives

an analysis of the semantic composition of regular possessives might look like the following (based on work by barker 1995, jensen and vikner 1994 and partee & borschev 1998):

possessive: john's car

structure: [[john's][car]]

in this analysis, possessa (car, in this case) are uniformly of type >. to get this type from common nouns whose basic type is , we assume that they undergo a type shift that is conditioned by lexical information (see vikner and jensen 2002 for a theory of constraints on this operation). the lexicon piece will be important, but for the time being let's take it for granted.

the basic meaning of car is Lx[car(x)], type . (L = lambda)
the type-shifted meaning for car might be LyLx[car(x) & owns(y)(x)]

the extra relation which is in this case owns can vary, but is constrained by the lexical information of car (vikner & jensen 2002) and by contextual factors such as the definiteness of the possessive (storto 2003).

let the meaning of john's be simply the entity j.

the composition results in the following formula for john's car: Lx[car(x) & owns(j)(x)]

or: the set of things that are cars such that john owns them.



composition of generic possessives

based on the analysis for regular possessives.

example: butcher's knife

structure: [[butcher's][knife]

the structural difference between regular and generic possessives is in the categories. as argued in munn 1995, strauss 2005, and elsewhere, regular possessives are DPs and their possessors are in Spec-DP, while generic possessives are subphrases of DP, perhaps NPs.