No.1

Under the popular interpretation of indefinite and definite articles, an indefinite DP introduces a new referent while a definite DP refers to the old/given referent, and thereby establishes an anaphoric relation to an antecedent. However, the following example provides evidence against this traditional understanding: (1a) Along came a Samaritan woman i , and Jesus k talked to her i. (1b) She i was surprised that he k would talk to a Samaritan i. (Grønn and Saebø: 2011) There is a co-referential relation between "a Samaritan woman" in (1a) and "a Samaritan" in (1b). But we cannot replace the latter "a Samaritan" by a definite description such as "the Samaritan", keeping the context of the sentences. This is a puzzle both for syntax and semantics/pragmatics. In this paper, we discuss usage restrictions between definite and indefinite descriptions. Here are two points. First, co-indexing in (1b) violates Binding Condition C ("R-expressions must be free"). But the violation seems not to be at work for the relevant interpretation. Note that the indefinite DP in (1b) is specific, hence does not have a non-specific quantificational interpretation. Secondly, if we introduce the semantic concept of "guise" (function from worlds to individuals: Heim 1988), the violation of Binding Condition C can be avoided because the pronoun "she" and "a Samaritan" in (1b) have different guises. From the semantic and pragmatic point of view, there must be another binding configuration with respect to the interpretation of (1b). Therefore we need two layers of co-indexing relationships: guises and actual extensional referents. The former has to do with intensional, contextual, and cultural interpretations of the referential expressions. Our main goal is to reconsider the status of Binding Condition of C, possibility of its violation so that indefinite DPs can establish an anaphoric interpretation.