Definiteness spreading in the Hebrew construct state

Abstract

The Construct State (CS) in Modern Hebrew displays a phenomenon known as Definiteness Spreading (DS), often characterized as having the definiteness value of the CS determined by that of its embedded genitive phrase. This is shown to be an oversimplification: semantically, DS gives rise to no less than four different interpretation patterns in definite-marked CSs. We examine the implications of these semantic facts for a Minimalist analysis of DS in terms of the operation Agree. It is argued that the formulation of Agree given in Chomsky (2000, 2001) does not provide the tools needed to account for these facts. A further problem for a syntactic analysis based on Agree is posed by the structural configuration found with adjectival CS modifiers, where agreement takes place despite the lack of the c-command relation required by Agree. This paper argues that both problems can be solved by viewing the Agree operation as a feature sharing operation, as proposed independently by several authors. Using this approach, all four semantic patterns can be derived using an independently motivated hypothesis regarding the interpretation of features at the syntax–semantics interface.