Context-Dependence, Perspective and Relativity

On words and thoughts about oneself (p. 253-254)


James Higginbotham


Abstract: In (Higginbotham 2003) the peculiar nature of first-personal or de se contents (including thoughts, memories, and utterances) was attributed to their reflexive character, where a content for an agent x is reflexive to the extent that it presents x as the bearer s(e) of the state or action e of thinking or saying it, and it registers in logical form the roles r(e') that x would play in events e' classified by its predicates. Here I extend the discussion so as to take into account critical remarks in Recanati (2007) and Morgan (2009), and to explain how I see the account going within a modal conception of the nature of (structured) propositions. On the linguistic side, I am specifically critical of the view that human languages with first-personal or other indexicals that are sensitive to embedding involve “context shift” in the sense of Schlenker (2003) or Anand (2006), rather than being special cases of anaphora. Departing also from Lewis (1979), whose internalist views did not permit a robust sense of de re thought, I argue for an account that rules out context shift; pins the de se on reflexivity of content; and restores de re content along standard proof-theoretic lines, as in Hintikka (1962). Finally, using examples from English and Norwegian, I observe (with Castañeda [1987], I believe) that the distinction between truly de se and merely de re thoughts about oneself extends to certain extensional as well as intensional simple sentences, where the question of context shift evidently cannot arise, but where the distinction between reflexive and non-reflexive thoughts can be made out in logical form.

1. Introduction


Indexical singular terms are understood as they are used by their speakers to refer to things; moreover, their behavior, their contribution to what is said, is not in general dependent upon their sentential position, or their depth of embedding. On the other hand, there are a number of cases in human languages, such as those surveyed in Anand (2006 and references cited there, where the same forms when embedded do not behave as they would in isolation. One way of looking at these phenomena, advanced in particular in Schlenker (2003), is to hold that they involve “context shift.”

It has been very useful to consider these phenomena, and their sophisticated discussion by the above authors has advanced both linguistics andphilosophy. In this note, however, I shall expand upon some considerations sketched in Higginbotham (2003, 2008), suggesting that context shift does not exist, and that having recourse to it is a misdiagnosis, in part an artifact of two aspects of the formalism of Intensional Logic, as adapted to the Logic of Demonstratives in Kaplan (1977), or its “extensionalization,” as in Schlenker (2003). The first aspect is the view that the fundamental task for semantics is the characterization of the notion of truth in a context (at a possible world, or relative to another such parameter), so that contexts are taken up as manipulable objects of modification or quantification. The second is the confinement of anaphoric relations to explicit arguments, as explained more fully below. If, contrary to these assumptions, the object of inquiry is the target truth conditions of potential utterances by competent speakers, and implicit arguments (as in the case of Tense) are brought into the anaphoric picture, then context shift is out of the question (the context of an utterance is whatever it is). But the phenomena given in the literature fall out straightforwardly through the anaphoric use of indexicals; or so I shall argue, looking here through the narrow, but I believe representative, prism of the first person.