Main Article Content
This paper addresses a long-standing debate concerning the derivation of de se construals. One camp holds that there is a dedicated mechanism of ‘de se binding’, which results in a de se pronoun being interpreted as a variable ranging over the doxastic alternatives of the attitude holder (e.g. Chierchia 1990). Another treats de se as a special case of de re under the acquaintance relation of identity (e.g. Lewis 1979, Reinhart 1990). This debate is premised on the assumption that the two different routes to de se result in identical truth conditions. I argue that this assumption is incorrect for a class of cases that can be delineated in a principled fashion — counterfactual attitude reports involving counter-identity, such as Ivanka imagined that she was Melania and she was giving an interview. The argument builds on Ninan 2008, who noticed that de re construal works differently with counterfactual attitudes, and that this has consequences for de se interpretation in this type of sentence. I spell out these consequences more precisely, drawing on a novel, crosslinguistically robust generalization about unambiguously de se expressions such as PRO (the ‘De Se Generalization’). I argue that a treatment of such expressions that appeals to de se-as-de re cannot account for the De Se Generalization in a principled way, and hence that a dedicated mechanism of de se binding must be included among the expressive resources of the grammar.