Truth in Virtue of Meaning Reconsidered

Abstract

The positivists defined analyticity as truth in virtue of meaning alone and advocated the view that the notion of analyticity so defined is co-extensive with both the notion of an a priori truth and that of a necessary truth. For a number of reasons, this notion of analyticity is nowadays held to be untenable, and the related doctrines about a priori truths and necessary truths are almost unanimously rejected. Against this consensus, I will argue that, if correctly understood, the positivists’ version of the analytic/synthetic distinction is defensible. Moreover, I will propose partial and somewhat qualified defences of their linguistic doctrines about a priori truths and necessary truths.

Notes

1 Following a common practice, I restrict the notion of analyticity to true sentences. A more inclusive notion, which allows for analytic falsehoods, can be obtained by altering the definition in the following way: a sentence is analytic if, and only if, its truth value depends solely on its meaning and not on how things are.

2 In the wake of Boghossian’s 1996 article, it has become usual to refer to the positivists’ conception of analytic sentences as ‘metaphysical analyticity’ and contrast it with an epistemological conception of analyticity. Because I will only discuss the former conception, I will continue to speak simply of analyticity.

3 Defences of Quine’s arguments from 1951 can be found in Harman (1996) and Margolis and Laurence (2001).

4 I am indebted to Glock (2003, pp. 158–159) for the reference to Lewis. Current defenders of the argument include, among others, Sider (2011, pp. 119–120), Williamson (2007, pp. 58–9), Hale (2002, pp. 305–7), Horwich (1997, p. 436), and Yablo (1992, p. 878). Topey (2018, footnote 7) provides an even more comprehensive list.

5 Different forms of conventionalism, not always in agreement with the one discussed in this article, are still defended by Sidelle (2009), Glock (2003), Garcia-Carpintero and Otero (2009), and Topey (2019).

6 A discussion of this objection can already be found in Malcom (1940, p. 200).

7 Actually, Russell claims that if ‘Mohammed Ali’ is introduced as a substitute for ‘Cassius Clay’ the reference determiner of the former name is ‘contained’ in, but does not coincide with, the reference determiner of the latter (2008, p. 83). This complication will be ignored, as it is immaterial to the present discussion.

8 I would like to thank David Dolby and an anonymous reviewer for their comments on previous drafts. Further to this, I received helpful suggestions from Severin Schroeder and the members of his Wittgenstein seminar with whom I discussed some of the topics treated in this paper.