Abstract
One might take the significance of Davidson’s indeterminacy thesis to be that the question as to which language we can take another to be speaking can only be settled relative to our choice of an acceptable theory for interpreting the speaker. This, in turn, could be taken to show that none of us is ever speaking a determinate language. I argue that this result is self-defeating and cannot avoid collapse into a troubling skepticism about meaning. I then offer a way of trying to make sense of the idea that some utterances do belong to determinate languages even though there is no determinate language one can take another to be speaking. This, however, results in an uninviting picture of communication in which no speaker is really in a position to say what another’s words mean.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
And thus, someone whose approach to the problems surrounding meaning presupposes that there are languages in use.
Though I will not argue for this point here, I think that Davidson is right to draw this conclusion: we cannot take any two equally acceptable theories of truth for a given speaker to be agreeing on the language they take him to be speaking if we accept the kinds of differences Davidson thinks could obtain between such theories.
(Kripke 1982).
Successfully, in my mind, though it would take me too far afield to discuss this issue here.
In fact, as an anonymous reviewer pointed out, Davidson is not even denying that prior knowledge of a shared language helps explain actual cases of successful communication. He is only arguing that the possibility of successful communication does not rest on it, since successful communication can be had in the absence of such knowledge (and fail in the presence of it).
In addition, he has also suggested that the languages spoken by individuals on particular occasions are much less stable than we might have supposed.
Since the languages we are interested in contain context-sensitive elements, the theorems of a theory of truth will not really be material biconditionals, but generalizations over contexts of utterance. However, to simplify exposition, I am ignoring context-sensitivity, and thus treating the theorems of a theory of truth as if they were biconditionals.
A sentence is true if and only if it is satisfied by all sequences of objects.
For suppose you are given such a theory for a language you do not already understand. How could this yield an understanding of the language or the capacity to use it? You would, in addition, need to connect these intermediary entities to your own words, or to words you already understand. Your understanding of your own words, however, does not tell you how to do this.
I am here presupposing that the notion of truth, as it figures in a Davidsonian theory of truth, is such that a sentence’s being true depends on what this sentence means, or on how words in it are being used. For a different account of how we should understand ‘true’ in a Davidsonian theory of meaning, see Max Kolbel (2001), and perhaps also Michael Williams (1999) (though I am not sure that Michael Williams would deny that ‘true’ in a Davidsonian theory of meaning should be understood as I am understanding it.) In some work in progress, I take up the question of the compatibility of Davidson’s approach to meaning with an “immanent” approach to truth (which takes the meaning of ‘true’ in application to one’s own sentences to be exhausted by the disquotation schema ‘‘s’ is true iff s’, and defines ‘true’ in application to the sentences of others in terms of this immanent notion of truth and some notion of translation). There, I argue that we cannot both adopt immanent approach to truth and take a theory of truth to serve as, or play a significant role in, a compositional meaning theory.
It is important not to misunderstand what a theory’s being interpretive requires. It does not require that any T-sentence that is a consequence of the theory be interpretive. For if ‘‘Snow is white’ is true in English if and only if snow is white’ is a consequence of a given theory, so is the sentence ‘‘Snow is white’ is true if and only if snow is white and either I have a dog or I do not have a dog’, since it is a logical consequence of the first T-sentence. What is required for a truth-theory’s being interpretive needs to be understood in terms of the idea of a “canonically derived” T-sentence. To be canonical, the derivation of a T-sentence from the axioms of our theory “must demonstrate, step by step, how the truth value of the sentences depends on a recursively given structure” (“Semantics for Natural Languages”, Davidson 1984, p. 61). For a truth theory to be interpretive, only the T-sentences canonically derived from its axioms need to be interpretive. Even apart from concerns with a theory’s being “interpretive”, the notion of a canonical derivation from the axioms needs to be invoked in explaining how a recursive theory of truth gives a constructive account of the truth-values of sentences: for only the canonical derivations will display the role of each expression in determining the truth values of sentences.
For an in depth account along similar lines, see Part I: “Historical Introduction to Truth-Theoretic Semantics”, in Lepore and Ludwig (2005). See also Evans and McDowell (1976), McDowell, “Truth-Conditions, Bivalence and Verificationism” and “Meaning, Communication and Knowledge”, both in McDowell (1998). Though I disagree on a number of points with both McDowell’s interpretation and Lepore and Ludwig’s, I believe these do not affect the point of this paper.
This will require finding the speaker’s beliefs to be largely in agreement with our own, but it will also require being able to explain our disagreements.
Similarly, consider the following assignments of satisfaction conditions to a predicate ‘x is F’ in L:
-
For any o, o satisfies ‘x is F’ in L if and only if o is G
-
For any o, o satisfies ‘x is F’ in L if and only if o is H
Call these equivalent if for any sentence of the metalanguage containing ‘G’, the result of replacing all occurrences of ‘G’ in it with ‘H’ will yield a materially equivalent sentence of the metalanguage. (To simplifify exposition, I am here speaking of an object itself as satisfying or failing to satisfy a one-place predicate. Strictly speaking, however, satisfaction as defined in a T-theory is a relation between sequences of objects and formulas of a language.)
-
An anonymous reviewer made the following suggestion. We can get underdetermination of interpretation theories without any prima facie incompatibility between their theories of truth, provided these theories yield prima facie incompatible assignments of meaning to the same sentences. The idea is that two theories of truth can be compatible (in the sense that no contradiction or falsehood ensues from taking their claims to be jointly true) without generating compatible or equivalent interpretations of a speaker’s sentences. For instance, the following T-sentences are equivalent:
-
‘Snow is white’ is true in L iff snow is white
-
‘Snow is white’ is true in L iff 2 + 2 = 4 and snow is white
However, taking each of these to be interpretive would yield the following prima facie incompatible claims:
-
‘Snow is white’ means in L that snow is white
-
‘Snow is white’ means in L that 2 + 2 = 4 and snow is white.
Suppose we have two truth theories T1 and T2 whose T-theorems are equivalent and such that (1) is a canonically derived T-theorem of T1, while (2) is a canonically derived T-theorem of T2. Suppose, further, that the constraints governing radical interpretation could confirm two total theories one of which contains T1 and the other T2. Then, because T1 and T2 yield prima facie incompatible claims like (3) and (4), we would have to describe the two total theories as underdetermined by these constraints, even though their component theories of truth are themselves perfectly compatible.
While I am sympathetic to the idea that two T-theories can generate incompatible interpretations of a speaker’s sentences without being themselves incompatible, I doubt that this idea can be usefully exploited here, for two reasons. First, I do not think that the constraints governing interpretation could confirm two T-theories like T1 and T2 as both interpretive, though this is a matter that deserves further discussion. (I do not think that two T-theories could assign equivalent truth-conditions to a speaker’s sentences, be both confirmed as interpretive, and yet generate intuitively incompatible assignments of meaning to the speaker’s sentences.) Secondly, though (3) and (4) do intuitively seem incompatible, it is hard to describe what their incompatibility comes to, or how a contradiction or falsity might result in taking them to be jointly true.
-
Davidson himself seems to be making the same point in the following passage: The question whether a theory of truth is true of a given language (that is, of a speaker or group of speakers) makes sense only if the sentences of that language have a meaning that is independent of the theory (otherwise the theory is not a theory in the ordinary sense, but a description of a possible language)…. If the question can be raised…the language must have a life independent of the definition (otherwise the definition is merely stipulative: it specifies, but is not true of, a language). (Davidson 2005a, pp. 36–37)
What Searle actually says is that if “there cannot be empirically well-motivated translations of the words of one language into those of another”, then “there cannot even be “correct” translations from a language into itself” (Searle 1987, p. 131). Davidson (and perhaps Quine) would retort that the indeterminacy thesis does not mean that there are no correct or empirically motivated translations from one language into another, but that there are more than one. My way of putting Searle’s point avoids this objection.
Searle mounts an even more radical objection to the indeterminacy thesis by appealing to the first person case. He argues that if the consequences of the thesis for self-interpretation were true, the thesis itself would be unintelligible to us. In other words, our understanding of the thesis implies its falsity” (Searle 1987, p. 131).
Some passages suggest that Davidson himself might have rejected this assumption . Consider:
It should not be concluded from the fact that a person is restricted to a unique way of interpreting himself (if this can be called interpretation: it would be better to say that aside from pathological cases, our way of interpreting others has no application to ourselves) that therefore his words have unique reference (“Indeterminism and Antirealism”, Davidson 2001, p. 80).
Here is how I think we could undermine option 1 by appeal to self-interpretation, though I have come to realize that the argument that follows is likely to be met with resistance. As already noted, the only difficulty in offering a theory of truth for one’s own sentences is in the construction, rather than the testing, of a particular theory. For where s is one of my own sentences, a canonically derived T-sentence for s is directly tested against my understanding of s. And the best theories of truth I can construct for my own sentences, using these very sentences, are ones tied to the homophonic translation manual: that is, theories such that the sentence used on the right hand-side of each canonically derived T-theorem is the sentence mentioned on its left-hand side (I am here simplifying by ignoring context-sensitive expressions). Any true and meaning-giving theories of truth for my own sentences—as I presently understand them—will thus deliver the same canonically derived T-sentences. (It is these assumptions that I think are most likely to be found problematic, though I do believe they are ultimately defensible, even in the context of a Davidsonian approach to interpretation.) Now, according to option 1, in constructing a recursive theory of truth for a body of sentences, I am specifying a particular language. It follows, in particular, that in constructing a theory of truth for my own sentences, I am specifying a particular language. But this, together with the result that the best theories I can construct for my own sentences will deliver the same T-sentences, give, I think, enough substance to the idea that there is only one language I can specify and correctly attribute to myself. For even if the different acceptable theories I could construct differed in some of their axioms, the fact that they yield the very same T-theorems enables us to view their differences as merely amounting to different ways of assigning to each expression the same role in a language. So we can plausibly maintain that on any given occasion, there is only one language I can correctly take myself to be speaking on that occasion. But if there is a unique language I can correctly take myself to be speaking, it is, I think, safe to conclude that this is the language that I am speaking, rather than explain my inability to attribute to myself more than one language as a shortcoming due to my limited perspective.
In light of this, let us reexamine the idea that in constructing a theory of truth for the sentences of another, I am assigning to each term a determinate referent and to each of sentence something that fixes a truth value for it. I can only be doing this if my own terms refer to determinate objects and my own sentences have determinate truth values. But it is supposed to be a consequence of referential indeterminacy that no term in anyone’s repertoire has a determinate referent. Therefore, since my own terms do not refer to determinate objects, I cannot use them to assign to someone else’s words determinate referents. As a result, Davidson’s claim that “words refer relative to a language”1 cannot be accepted as true if we accept referential indeterminacy, think of a language is the sort of thing I can specify in constructing a theory of truth, and think of a word as “referring” only if it has a determinate referent.
For instance, Lepore and Ludwig argue as follows: “Suppose our two theories are relativized to different languages. Then if each is as good a theory of the speaker’s language, but they are different languages, we are committed to saying that his language is identical to each of two distinct ones not identical with each other, which is a contradiction”. (Lepore and Ludwig 2005, p. 240f) Lepore and Ludwig’s point is that the same language cannot be correctly identified with two different languages. For a different kind of argument, see Ramberg (1989, pp. 99–100), and Bar-On and Risjord (1992, pp. 165–166). The arguments they sketch is from the indeterminacy of language attribution (rather than to their being more than one language correctly attributable to a given speaker) to the conclusion that there is no such thing as the language spoken by an individual.
What this account seems to imply is that the meanings of object-language sentences containing the word ‘mass’ are strictly speaking inexpressible in the interpreter’s language. This example is borrowed from Field (1973), though he there offers a different, and in my view, much more convincing account of it than the one I have sketched here.
See Lepore and Ludwig (2005, p. 239).
See Lepore and Ludwig (2005, pp. 228–236).
Against this interpretation, consider the following passage in “Indeterminism and Antirealism”:
There is a question, to be sure, whether underdetermination in Quine’s sense can occur, for it requires that there be empirically equivalent, but incompatible, theories. Indeterminacy is not like this: the empirically equivalent theories it accepts as equally good for understanding an agent are not incompatible, any more than the measurement of weight in pounds or kilos involves incompatible theories of weight (Davidson 2001, p. 76).
Lepore and Ludwig do consider, but only to reject, the idea that “assignments of meanings to sentences should be relativized to an interpretation theory.” (pp. 240–241) Against this idea, they argue that claims about the meanings of sentences do not require any further relativization, beyond the familiar relativization to a language. But one can agree with this, without denying, as they do, that claims about the meanings of sentences need not be relativized to interpretation theories. For if you think that claims about what language someone is speaking, or a given utterance belongs to, themselves require relativization to an interpretation theory, then this gets you the result that meaning is relative to an interpretation theory, without its being relative to anything more than a language.
Lepore and Ludwig are here assuming that whenever there are different possible ways of representing features of a certain system in terms of those of another system, this is to be explained by the latter system being essentially richer than the former. I am not convinced that that this is right, though I do not currently have any counterexamples to it.
I take the guiding idea of an interpretivist approach to be that there is nothing more to what speakers mean by their words than what they can take each other to mean in trying to make sense of each other’s sayings and doings. Davidson and Dennett, both prominent interpretivists, further require that any question about what an individual means by his words or what his attitudes are, can only be answered in the context of a particular scheme for making sense of the individual, where there are supposed to be many equally successful schemes. Davidson wants to describe any two such schemes as equivalent, while Dennett thinks they can involve “substantive disagreements” (Dennett 1991, p. 46; “Indeterminism and Antirealism”, Davidson 2001). I think that as long as this restriction to a particular way of making sense of the individual is imposed, the indeterminacy of language-attribution is unavoidable. But I also think that there are ways of dropping this restriction to a particular scheme without straying away from an interpretivist approach to meaning.
References
Bar-On, D., & Risjord, M. (1992). Is there such a thing as a language? Canadian Journal of Philosophy, 22(2), 163–190.
Davidson, D. (1984). Inquiries into truth and interpretation. Oxford: Oxford University Press.
Davidson, D. (2001). Subjective, intersubjective, objective. Oxford: Clarendon Press.
Davidson, D. (2005a). Truth and predication. Belknap Press: Harvard University Press.
Davidson, D. (2005b). Truth, language and history. Oxford: Clarendon Press.
Dennett, D. (1991). Real patterns. The Journal of Philosophy, 88, 27–51. doi:10.2307/2027085.
Dummett, M. (1993). What is a theory of meaning (I). In M. Dummett (Ed.), The seas of language (pp. 1–33). Oxford: Clarendon Press.
Evans, G., & McDowell, J. (1976). Introduction. In G. Evans & J. McDowell (Eds.), Truth and meaning. Oxford: Oxford University Press.
Field, H. (1973). Theory change and the indeterminacy of reference. The Journal of Philosophy, 70, 462–481. doi:10.2307/2025110.
Hacking, I. (1986). The parody of conversation. In E. Lepore (Ed.), Truth and interpretation: Perspectives on the philosophy of Donald Davidson. Cambridge, MA: Blackwell.
Kolbel, M. (2001). Two dogmas of Davidsonian semantics. The Journal of Philosophy, XCIII(12), 613–635. doi:10.2307/3649462.
Kripke, S. (1982). Wittgenstein on rules and private language. Cambridge, MA: Harvard University Press.
Lepore, E., & Ludwig, K. (2005). Donald Davidson: Meaning, truth, language and reality. Oxford: Clarendon Press.
McDowell, J. (1998). Meaning, knowledge, and reality. Cambridge, MA: Harvard University Press.
Pietroski, P. (1994). A defense of derangement. Canadian Journal of Philosophy, 24, 95–118.
Ramberg, B. (1989). Donald Davidson’s philosophy of language: An introduction. Cambridge: Basil Blackwell.
Rorty, R. (1998). Davidson between Wittgenstein and Tarski. Critica, 30(88), 49–71.
Searle, J. (1987). Indeterminacy, empiricism, and the first person. The Journal of Philosophy, LXXXIV(3), 123–146. doi: 10.2307/2026595
Stroud, B. (1998). The theory of meaning and the practice of communication. Critica, 30(88), 49–71.
Williams, M. (1999). Meaning and deflationary truth. The Journal of Philosophy, XCVI(11), 545–564. doi:10.2307/2564623.
Acknowledgment
I would like to thank John Campbell, Hannah Ginsborg, Niko Kolodny, John MacFarlane, Matt Parrott, Barry Smith and Barry Stroud for their useful feedback on earlier versions of this paper.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Khatchirian, A. What is wrong with the indeterminacy of language-attribution?. Philos Stud 146, 197–221 (2009). https://doi.org/10.1007/s11098-008-9251-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-008-9251-z