The case of the direct object in Turkish: Semantics, syntax and morphology

In this paper, we investigate the interaction between semantic parameters and morphological constraints in determining the distribution of the accusative case marker-(y)I in Turkish. This marker is often discussed as an instance of differentiated object marking (DOM). The account of accusative marking based on a functional interpretation of DOM assumes that the case suffix marks a direct object if it is too similar to an archetypical subject. Other approaches to accusative marking in Turkish have been based on the observation that the accusative marker is closely related to the direct object’s specificity as such, rather than to the similarity of the direct object to a typical subject—and there is general agreement that typical subjects are specific. These approaches predict that specific subjects are also overtly case-marked; this is confirmed by the data. Enc (1991) explains specificity in terms of partitivity and argues that the accusative case marker indicates a partitive construction (or at least an implied partitive relation), and thus marks a specific direct object. In this paper we show that the conditions for the distribution of this case marker are quite complex and cannot be explained within the functional view of DOM. In particular, we argue that the suffix indicates specificity under certain morpho-syntactic conditions, rather than indicating just a contrast to the subject. This view is vindicated by the assignment of (genitive) case to the embedded subject that is determined by very similar morpho-syntactic and semantic conditions: the embedded subject receives genitive case if it is specific and no genitive case if it is non-specific. Furthermore we show that Enc’s definition of specificity in terms of partitivity must be modified for semantic as well as morphological reasons. We develop a more flexible notion of specificity in terms of referentially anchored indefinite NPs. We give additional evidence, based on the detailed analysis of the morphological conditions for partitives, which shows that partitives are not necessarily specific. In conclusion, we show that the accusative case marker can indicate the referential property of the direct object (such as specificity) in clearly defined morphological environments in a reliable fashion; in other contexts, it is not a reliable indicator of properties like specificity.