Function Domains in Variable-Free Semantics

Sauerland ( 1 999) used such examples as the basis of an argument in favor of the traditional representation of pronouns as indexed variables, and against Jacobson's Variable Free Semantics (VFS). In her response, Jacobson defends the VFS frame­ work and argues that it can account for a broader range of examples than Sauer­ land's proposal. In particular, Jacobson proposes that we can account for such examples by representing pronouns as restricted functions, and appealing to the contrastibility of their domains. Finally, Sauerland (2000) abandons the alphabetic variants proposal as unworkable and proposes an account which involves the trans­ lation of the stressed pronouns as covert definites (E-type pronouns). 1 In the present paper I consider some issues connected with the incorporation of function domains into the Variable-Free Semantics framework. I show that al­ though there is no technical obstacle to introducing function domains, for technical and empirical reasons they should not be used to account for the contrastive stress data. After a short introduction to Variable Free Semantics, I review Sauerland's ( 1 999) argument against VFS and Jacobson's counter-proposal. Section 4 discusses the use of domains in VFS, and their application to the contrastive stress problem. Section 5 compares the domains required for evaluation with the notion of domains assumed by the domains analysis. Finally, section 6 argues that contrast cannot be predicted by simply comparing semantic translations, and discusses some tentative suggestions for a suitable account.