The interpretive value of object splits

The goal of this paper is to account for the observation that in a broad range of genetically unrelated languages we find two classes of direct objects based on their syntactic and semantic properties. Specifically, we find splits in object case marking, object position, or the ability of the object to trigger verb agreement. This split always correlates with specificity or definiteness of the object, and in a subset of languages it also correlates with delimitation or boundedness of the event. We propose that this split in object properties is determined by the presence or absence of a feature [Quantization] on the object DP. This feature, which formalizes Krifka's characterization of the countability of nominals and events, may also be present on either the verb or object agreement (Agr-O). The observed cross-linguistic variation is attributed to the language specific choice between these two heads as follows: When [Quant] is a feature of the verb, it is interpretable and independently encodes delimitation/boundedness of the event. However, when [Quant] is a feature of the functional head Agr, it is an uninterpretable feature which only enters into a checking relation with a definite/specific direct object.