The Philosophy of Philosophy. Timothy Williamson

uttered, if indexicals are present. By contrast, the appeal to meaning in the case of analytically true sentences is not to the proposition expressed on some particular occasion but rather to the linguistic meaning of the sentence, which is invariant across contexts, even if indexicals are present.

      11 ¹¹ To handle ambiguity, treat it as homonymy: distinct sentences with the same superficial form. The reification of meanings in the definition can be eliminated at the cost of circumlocution. Note also that the utterance of a modal-analytic truth may be false if the context shifts during the utterance: consider “If it is now exactly noon then it is now exactly noon.” Similarly, an utterance of “If John is a bachelor then John is unmarried” may express a falsehood if the wedding ceremony is completed between the utterance of the antecedent and the utterance of the consequent. Taking such complications into account would not help friends of analyticity.

      12 ¹² The notion of modal-analyticity is similar to the notion of deep necessity in Evans (1979), where the truth of the sentence does not depend on any contingent feature of reality.

      13 ¹³ See n. 6 for this terminology.

      14 ¹⁴ The term “Frege-analytic” is from Boghossian (1997), with reference to §3 of Frege (1950) (as Boghossian suggests, the interpretation of the passage is not entirely clear). He classifies the notion of Frege-analyticity as neither epistemological nor metaphysical but semantic (1997: 363); for convenience, it is treated here under the heading of metaphysical notions of analyticity.

      15 ¹⁵ Quine (1966: 111) notes that so-called truth by definitions (“Every vixen is a female fox”) depends on prior logical truths (“Every female fox is a female fox”).

      16 ¹⁶ Note that the epistemological issue is not how we can know that s is a logical truth; it is how, given that s is a logical truth, we can know the simple truth of s.

      17 ¹⁷ For more discussion and further references to the controversy over the nature of logical consequence see Williamson (2000b).

      18 ¹⁸ See Salmon (1986), especially 133–5.

      19 ¹⁹ See Kripke (1979). This contradicts Dummett’s claim that “It is an undeniable feature of the notion of meaning – obscure as that notion is – that meaning is transparent in the sense that, if someone attaches a meaning to each of two words, he must know whether these meanings are the same (1978: 131). For more general theoretical considerations against such claims see Williamson (2000a: 94–107). See also Horwich (1998: 100–1).

      20 ²⁰ The contexts of utterance and circumstances of evaluation here are not restricted to the actual world. If the content of an expression has a structure which reflects the grammatical structure of the expression, then sameness of intension does not imply sameness of content, and sameness of intension in every context does not entail sameness of character, that is, sameness of content in every context. See Kaplan (1989) for relevant background.

      21 ²¹ Boghossian argues that many a priori truths are not Frege-analytic (1997: 338–9).

      22 ²² This point is related to the paradox of analysis: how can a conceptual analysis be both correct and informative? The paradox goes back to Langford (1942).

      23 ²³ See Fodor (1998: 69–87) and Williamson (2000a: 31–3) for further discussion.

      24 ²⁴ For helpful discussion see the essays in Part IV of Stalnaker (2003). He sometimes use the terminology of “descriptive semantics” and “foundational semantics” rather than “semantics” and “metasemantics” respectively.