Abstract: Expressions such as pronouns, definite descriptions, and demonstratives can all refer to a familiar entity in a discourse, though they differ in relative distributions and possible interpretations. While the interaction among these expressions are evident from natural language data, they are often assumed to be separate semantic elements, making it difficult to derive such an interaction from their underlying denotations. I propose a unified semantic account of these definite expressions, where they share the same underlying semantic structure and only differ in the amount of restrictions that they carry. I discuss the main theoretical and empirical motivations for such an account, focusing on a phenomenon in bare argument languages where the presence of pronouns block the use of bare nouns in anaphoric contexts. I also discuss the implications the account has on the semantic analysis of pointing that is used to refer to entities in both spoken and signed languages.
