1 The pursuit of this program has turned into a minor industry. For references, see LePore.

2 Oxford: Basil Blackwell1989

3 This way of understanding Davidson’s work is strongly encouraged by Davidson’s ‘Truth and Meaning’ (in Inquiries into Truth and Interpretation [Oxford: Oxford University Press 1984)).

4 In Inquiries into Truth and Interpretation, 56 n.3, Davidson points out that this does not amount to equating truth conditions with meanings. For other works connecting truth-conditions with meaning, see Davidson, Donald and Harman, Gilbert eds., The Logic of Grammar (Encio, CA: Dickenson 1975)Google Scholar, Appiah, Anthony For Truth in Semantics (Oxford: Basil Blackwell 1986)Google Scholar, and Lycan, William G. Logical Form in Natural Language (Cambridge, MA: MIT Press 1984)Google Scholar.

5 On the indeterminacy of interpretation, see especially Inquiries ... , 228 and 239f.

6 Interestingly, Ramberg himself stops short of taking this next obvious step. Instead, he immediately proceeds by summarizing (and then developing) a rather more tangled argument which we will examine below.

7 For some discussion of this requirement, see Bar-On, ’semantic Verificationism, Linguistic Behaviorism and Translation,’ forthcoming in Philosophical Studies 1992.

8 This is because the requirement says that there can be nothing in principle ‘hidden’ about the structure and meaning of a speaker’s speech.

9 To ascribe to a speaker semantic knowledge in Ramberg’s sense is not necessarily to ascribe to her knowledge of semantics, in the sense of, e.g., having an articulate truth definition for her language. Michael Resnik has objected that all the speaker needs to know to understand the sentence in question is what it is for the sun to shine, and that this knowledge is not about words, and thus not semantic.

10 Strictly speaking, the appropriate form would be: ‘is true in speaker S’s mouth at time t and place p.’ This allows for the specification of truth-conditions for sentences containing indexicals, demonstratives, etc. It also gets rid of any explicit reference to languages, which reference should be illegitimate if, as Davidson eventually concludes, there are no languages.

11 Interestingly, Michael Dummett seems to use considerations similar to those just mentioned in arguing against Davidson’s use of the notion of truth as the central notion in a theory of meaning. He apparently takes the inevitable dependence of interpretation on what is held true to give the lie to any attempt to construe meanings in terms of evidence-transcendent truth-conditions. For a reading of Dummett which supports this, see, e.g., Lycan (Ch. 10, sec. 2).

12 Bold claims for the principle of charity (as well as humanity) have been made by a number of writers. See, for instance, Hollis, Martin ‘The Limits of Irrationality,’ Archives europeenes de sociologie 7 (1967) 265-71CrossRefGoogle Scholar, and ‘Reason and Ritual,’ Philosophy 43 (1967) 231-47; Lukes, Steven ’some Problems about Rationality,’ Archives europeenes de sociologie 7 (1967) 247-64CrossRefGoogle Scholar; Macdonald, Graham and Pettit, Phillip eds., Semantics and Social Science (London: Routledge & Kegan Paul 1981)Google Scholar; Root, Michael ‘Davidson and Social Science,’ in LePore, 272-304Google Scholar; and the essays in Hollis, Martin and Lukes, Steven eds., Rationality and Relativism (Cambridge, MA: MIT Press 1982)Google Scholar. Ramberg’s analysis correctly shows that insofar as these authors have relied on the precondition role of PC, their arguments are unsound.

13 See, for instance, Devitt, Michael Realism and Truth (Princeton: Princeton University Press 1984)Google Scholar; Grandy, Richard E. ‘Reference, Meaning, and Belief; Journal of Philosophy 70 (1973) 439-52CrossRefGoogle Scholar; Hollis Martin, ‘The Social Destruction of Reality; in Hollis and Lukes, 67-86; Steven Lukes, ‘Relativism in its Place; in Hollis and Lukes, 261-305; and MacDonald and Pettit.

14 We may, e.g., get"’ ... " is true iff Rama is a witch: Here, ‘ .. :will be held true, yet interpreted by something the interpreter takes to be false. Notice that the T -sentence is still true (both sides of the biconditional are false). But we have broken the link between the native speaker’s holding a sentence (i.e., ‘…’) to be true and its being true, thereby ascribing a false belief to the speaker.

15 This may happen when the interpretees’ beliefs concern matters about which the interpreter has no opinion. Perhaps the interpretees’ knowledge of galaxies is much more advanced and they pronounce on their configuration, etc., where we would remain silent. Or perhaps there are phenomena or kinds of objects in their surroundings about which we know nothing. In such cases we may be unable to identify or state the appropriate truth-conditions until the gap between us closes, and their beliefs may seem odd to us. Alternatively, they may seem to have elaborate explanations of phenomena we have never gotten around to — or had interest in — explaining. But the difference may also be due to the fact that the interpretees utilize different ‘basic categories’: for instance, the interpretees may have a primitive vocabulary item referring to rabbits-in-sunny-weather (the example is Ramberg’s, 84f.). And this may just be the tip of an iceberg. Our subjects may have very different ways of carving things up, as a consequence of their idiosyncratic needs, interests, etc. Ramberg is explicit that PC in its precondition role ‘does not counsel us to assume that speakers ... see things the way we do; it is not a tool of cognitive imperialism’ (Ramberg, 85). But the question is whether PC in its adjudicating role would also allow for differences of type (3). If it did not, proponents of PH would quarrel with it.

16 It is wrong, then, to see the proponents of PH as attempting to bring anthropological considerations to bear on the precondition role of PC (contra Ramberg, 77). Ramberg eventually gets around to this understanding of PH, but rejects it as ‘barren,’ on the grounds that ‘the anomalies are expressions of discrepancies between patterns, between structures, not beliefs held individually. We are not told which T -sentence is the source of trouble because no one T -sentence is the source of trouble’ (Ramberg, 88f.). It would be, however, implausible to limit PH to the evaluation or explanation of individual beliefs. When the beliefs of an individual or group are explained, it is typically a body of interrelated beliefs that is subject to explanation. We want to know why, for instance, the Azande maintained their witchcraft beliefs, or how such beliefs function in their society and cohere with their rituals and practices. Humanity is brought to bear on patterns of belief no less than Charity (in its adjudication role).

17 Ramberg acknowledges all this (91), but does not seem to recognize how it plays into the hands of the proponent of PH. He even cites approvingly David Lewis’s ‘credo’ that the possible existence of two conflicting but acceptable theories of interpretation is always proof that we have not yet found all the relevant constraints on interpretation (Ramberg, 94). He indicates no awareness that this actually undermines his argument for the indeterminacy of language and his insistence that Charity is the only relevant principle constraining interpretation. It may be worth emphasizing that the acknowledgement of the relevance of psychological and other considerations can be reconciled with an insistence on the ’supremacy’ of Charity by properly separating the precondition and adjudication roles. Charity is paramount in breaking into a language and gathering an evidential base. Quite possibly, competing sets of T -sentences can be projected from that base, all equally confirmable through empirical testing. In adjudicating among them, it is Humanity, rather than Charity, that must be followed (if we are to avoid epistemological occultism and cognitive parochialism).

18 Ramberg gives the impression that the rejection of convention and the rejection of language are separate consequences (he refers to them as ‘these conclusions’ in the paragraph following the one just quoted). But it emerges that the ‘more recenf rejection of language by Davidson is actually supposed to follow from the rejection of convention. This is at any rate what we take to be the best reading of the argument, based on Ramberg’s discussion on pages 100ff. Not much hangs on this for our purposes, since our main criticism will focus on the first part of the argument, the alleged inference from (1’) to (2’).

19 This follows Appiah; also cf. Davidson, ‘A Nice Derangement ... ,’ 436ff.

20 Ramberg identifies this as a premise shared by Davidson and one of his chief opponents, Michael Dummett. Whereas Dummett is said to accept its antecedent, Davidson rejects its consequent. See Ramberg, 106f.

21 Cf. Tarski, Alfred ‘The Semantic Conception of Truth,’ Philosophy and Phenomenological Research 4 (1944) 341-76CrossRefGoogle Scholar; and for discussion, Harman, Gilbert ‘Logical Form,’ Foundations of Language 9 (1972) 38-65 and LycanGoogle Scholar.

22 See Tarski in Corcoran, John ed., Logic, Semantics, Metamathematics: Papers from 1923-1938 by Alfred Tarski, 2nd ed (Indianapolis: Hackett 1983), 164Google Scholar, and in Linsky, Leonard ed., Semantics and the Philosophy of Language (Urbana, IL: University of Illinois Press 1952), 18fGoogle Scholar.

23 Note that this would be so only if we conveyed our understanding by actually introducing into their language a new predicate. But there may be ways of conveying our understanding to them without expanding their language. We could teach them English, or point to an analogy with the case in which they constructed a truth theory in their language for some other language.

24 This would go contrary to premise (3’), which ties the viability of the notion of convention to the applicability of the same truth theory to different speakers or at different times.

25 It might be thought that this way of eliminating indeterminacy provides a recipe for rejecting the crucial premise (2) in Argument I. It is attractive because it does not involve resorting to principles beyond the PC. However, it should be noted that the cost to be paid is the recognition that truth theories are incompletable. And this, Ramberg thinks, in tum leads to the dispensibility of language thesis, as shown by Argument II. Thus, we can only take seriously a Davidsonian assurance that indeterminacy is unworrying, thereby rejecting Argument I, if we accept a key premise in Argument II. (This may provide a further, charitable explanation of why Ramberg does not develop Argument I, but rather focuses on Argument II.)

26 Ramberg sometimes seems guilty of this conflation; see, e.g., Ramberg, 62, 79.

27 It is interesting to note that, on p. 90, Ramberg uses some of the same materials used in Arguments I and II to draw a much less radical (and more acceptable) conclusion, to wit: ‘Once the full import of Davidson’s holistic empirical strategy has soaked in, we will not make this mistake [of thinking there are "pre-existing" determinate meanings or languages], because we will see that the only possible criterion of identity for language is the successful application of one and the same theory of truth. Since we never apply the exact same theory to any two speakers, or even to any one speaker at different points in time, this makes the identity of languages a matter of degree’ (our emphasis).

28 We thank Simon Blackburn for pressing this point. Ramberg himself does discuss the role malapropisms play in Davidson’s attack on convention and language (Ramberg, 101-4). However, as we saw earlier, there are clear indications that he takes the incompleteness of language/truth theory to provide a crucial starting point for the attack. We have argued that incompleteness (in either sense we have identified) does nothing to undermine the concept of a language. Ramberg also fails to recognize that, if the lesson of the malapropism phenomenon is as Davidson — and he himself — see it, it is completely idle to invoke incompleteness. Ramberg is apparently moved to bring in the issue of incompleteness because of his concern to weave Davidson’s views about language into a coherent whole, and to establish that Davidson’s recent rejection of language is a natural consequence of his semantic program. We argue against this in the next section. Given the inadequacy of Ramberg’s two senses of incompleteness in supporting claim (2’), it may be most charitable to take him to be using, for the purpose of Argument II, incompleteness in yet a third sense: the indefinite flexibility of interpretation witnessed by the phenomenon of malapropism. If, however, as we argue below, claim (2’) nonetheless remains unsupported, Argument II would still fail.

29 For a very congenial discussion of the malapropism phenomenon and a critique of Davidson’s treatment of it, see George, Alexander ‘Whose Language Is It Anyway? Some Notes on Idiolects,’ Philosophical Quarterly 40 (1990) 275-98CrossRefGoogle Scholar. On the social character of linguistic norms see Dummett, Michael ‘The Social Character of Meaning,’ in Truth and Other Enigmas (Cambridge, MA: Harvard University Press 1978);Google Scholar also see Dorit Bar-On, ‘On the Possibility of a Solitary Language,’ forthcoming in Nous (1992).

30 George (283ff., esp. n.24) makes the related point that, even if the malapropism phenomenon helps show that the assumption of a shared, stable set of linguistic conventions does not suffice to explain what goes on in linguistic communication, it does not show that the assumption is not a necessary one. The Davidsonian rejection of language, however, requires both claims (cf. Ramberg, 104).

31 This is echoed in (Ramberg, 64).

32 For the relevant notions, see Quine, ch. II).

33 We thank Simon Blackburn, Bill Lycan, Mike Resnik, and especially Keith Simmons for helpful comments on an earlier draft of this paper.