A semantic reanalysis of the partitive constraint

In this article a semantic reanalysis of the Partitive Constraint is given that is based on a distinction between determiners that quantify over entities and determiners that quantify over sets of entities. This leads to a straightforward analysis of problems that arose within earlier analyses of the Partitive Constraint. The reformulated Partitive Constraint simply states that NPs that are allowed in partitives must be entity-denoting if the upstairs determiner quantifies over entities, and set-denoting if the upstairs determiner quantifies over sets. NPs that denote restricted sets of entities satisfy the Partitive Constraint. That explains the presence of a modifying phrase in certain cases where the embedded NP is headed by a weak determiner. All in all, I will argue that the Partitive Constraint can and must be maintained as a semantic condition, and cannot be reduced to some kind of pragmatic principle.