20th WCP: Davidson and Indeterminacy of Meaning

According to Quine's famous thesis of the indeterminacy of translation there are no facts of matter that could determine the choice between two or more incompatible translation schemes that are in accordance with all behavioural evidence. As Quine himself puts it:

Manuals for translating one language into another can be set up in divergent ways, all compatible with the totality of speech disposition, yet incompatible with one another. In countless places they will diverge in giving, as their respective translations of a sentence of one language, sentences of the other language which stand to each other in no plausible sort of equivalence however loose. (1) (p.27).

The indeterminacy is a consequence of the thesis of the underdetermination of theory by experience, as well as Quine' s behaviouristic view of language and his rejection of the analytic/synthetic distinction.

Donald Davidson has pursued, and modified, Quine's approach to language and interpretation. According to Davidson, to know what a speaker means by an utterance is to know what belief that speaker intends it to express. Identifying meanings is thus inseparable from the task of attributing beliefs. However, we could get access to a speaker's beliefs only when we could identify instances where a speaker holds a sentence true. In Davidson's method of radical interpretation, the radical interpreter has to notice the conditions under which the alien speaker assents to or dissents from a variety of his sentences. The relevant conditions will be what we take to be the truth conditions of his sentences. However, there will always be different ways of assigning truth conditions to sentences, hence a degree of indeterminacy will always be present. Thus, Davidson thinks "Quine is right ... in holding that an important degree of indeterminacy will remain after all the evidence is in; a number of significantly different theories of truth will fit the evidence equally well.". (2) This gives rise to what he calls "the thesis of indeterminacy of meaning" (IM). Davidson claims that IM is neither mysterious nor threatening. "It is no more mysterious than the fact that temperature can be measured in Centigrade or Fahrenheit (or any linear transformation of those numbers)". (3) And it is not threatening because the existence of empirically equivalent languages (that is languages equally consistent with all possible empirical evidence) does not question the reality or objectivity of the correct interpretation of utterances and their accompanying mental states (any more than the existence of various scales for recording temperatures or lengths would question the reality or objectivity of temperature or length). In this paper I aim to show that Davidson's theory of indeterminacy of meaning is not as innocuous as he claims it to be and has consequences which do not sit easily with some core elements of the Davidsonian overall approach to language and meaning.

The similarities and the differences between Davidson and Quine's view on indeterminacy of translation are: (4)

1-Davidson agrees with Quine that truth may be indeterminate. That is a certain sentence may be true according to one manual of translation and the same sentence false according to a second manual. But, as we shall see, he argues that the application of the principle of charity on an across the board basis would lessen the scope of this type of indeterminacy.

2-Quine has argued that logical form may be indeterminate. That is, two theories or manuals of translation for the same language may differ in their assignment of the underlying logic to the language being interpreted. Davidson limits the extent of this type of indeterminacy by attributing a uniform quantificational structure to all languages through the application of Tarskian truth schema by the radical interpreter.

3-Davidson also agrees with Quine that there may be differences in the references assigned to the same words and phrases, (the inscrutability of reference thesis). However, he argues that Quinean ontological relativity does not follow from the thesis of inscrutability of reference, for we cannot make sense of the suggestion that the ontologies of different languages may be totally different for the very idea that different languages may present differing conceptual schemes is unintelligible.

Despite these limiting factors, Davidson claims, there is still room for indeterminacy because "when all the evidence is in, alternative ways of stating the facts remain open". (5) The scope and import of the resulting indeterminacy, however, are less than that in Quine and it does not represent a failure to capture significant distinctions in terms of either meaning or truth.

The indeterminacy of translation implies that a sentence can be rendered true according to one acceptable system of interpretation and false according to a second, equally acceptable system of interpretation. Davidson claims that this view would appear paradoxical only if we think that there is a single "unique language to which a given utterance belongs. But, we can without paradox take that utterance to belong to one or another language, provided we make allowances for a shift in other parts of our total theory of a person." (6) Thus, Davidson's theory of indeterminacy of translation and meaning, in effect, becomes a thesis of indeterminacy of language ascription, because, "since we admit that it is not entirely an empirical question what language a person speaks; the evidence allows us some choice in languages, even to the point of allowing us to assign conflicting truth conditions to the same sentence". (7)

It seems highly counterintuitive, if not paradoxical, to claim that one can, in uttering a single sentence, be speaking in many languages. (8) Davidson removes the air of paradox by claiming that the same sentence is not spoken simultaneously in two different languages because there is no such a thing as a language as traditionally constructed. (9) Thus IM, which was presented as no more mysterious than having different standards of measurement, leads, firstly, to indeterminacy of language ascription and then to the even more controversial position that there is no such thing as a language, rather a host of maybe roughly overlapping idiolects. However, even if we accept this position, the indeterminacy of language ascription remains intact, but now it operates at the level of the ascription of idiolects. It seems equally counter intuitive to suggest that at any given moment I could be speaking simultaneously in indefinitely many idiolects.

One way to bring out the charge that IM is not a non-threatening thesis is to look at the disanalogies between the localised examples, such as different conventions for measurement, that Davidson cites and the all embracing question of indeterminacy of language ascription.

One important disanalogy, as noted by Davidson himself, is that in the cases of measurement of temperature or weight "our linguistic interactions with others allows us to agree on the properties of the numbers and the sort of structures in nature that allow us to represent those structures in the numbers. We cannot in the same way agree on the structure of sentences or thoughts we use to chart the thoughts and meanings of others, for the attempt to reach such an agreement simply sends us back to the very process of interpretation on which all agreement depends." (10) To put it slightly differently, in case of measurements of temperature there are certain pre-existing conventions which allows us to correlate the two structures. Such conventions are not present for the totality of a language.

In addition, in cases of "indeterminacy of temperature" sentences expressing measurement according to one scale are easily reducible into sentences giving measurements in the second. Furthermore, the two measurements scales can be easily reduced into a third (base) scale or language. In the instances of indeterminacy of translation cited by Quine, such reduction does not seem possible. (11) Even more significantly, in cases of localised indeterminacies, such as that of measurement, there does not seem to be any indeterminacy of truth while the indeterminacy of language ascription arises out of indeterminacy about truth. This is because in the instances of indeterminacy of measurement there is a shared invariant and accepted 'fact of the matter', or agreement about how things are in the world, which makes reduction possible and mitigates against the possibility of indeterminacy about truth. It is difficult to imagine, however, what the invariant facts of matter for any language, in its totality, are or can be.

Davidson reduces the scope of indeterminacy about truth and attempts to overcome the above disanalogies through the use of the principle of charity. In its original formulation the principle of charity emphasised consistency and agreement on truth. Davidson had argued that "To the extent that we fail to discover a coherent and plausible pattern in the attitudes and actions of others we simply forgo the chance of treating them as persons." (12) In our need to make sense of other speakers "we will try for a theory that finds him consistent, a believer of truth, and a lover of the good (all by our own lights, it goes without saying)". (13) In recent years he has argued that in attributing beliefs and hence meaning to users of language, at least in some basic cases, the object of a belief should be identified with the cause of that belief . He now clearly distinguishes between two elements in the principle of charity: the "Principle of Coherence" and the "Principle of Correspondence". The Principle of Coherence "prompts the interpreter to discover a degree of logical consistency in the thought of the speaker; the Principle of Correspondence prompts the interpreter to take the speaker to be responding to the same feature of the world that he (the interpreter) would be responding to under similar circumstance". (14) The Principle of Coherence endows the speaker with a modicum of logical truth, the principle of correspondence endows him with a degree of true belief about the world. The principle of correspondence, thus, ensures that there is a "fact of the matter" in common to different but empirically equivalent languages. In other words, Davidson thinks that we are able to interpret a speaker and hence assign a specific language to her, because we share a world with her. The principle of coherence provides us with the means for correlating the two languages. Thus, the two interpretative principles reduce the differences between the Quinean and the more common or garden, Davidsonian, interpretations of the thesis of indeterminacy of translation.

The difficulty with this approach lies with Davidson's use of the notion of causality. The concept of causality, as Quine and Putnam have pointed out, is applied to the world according to human interests. Davidson agrees that specific causal relations are often singled out of a myriad of such possible relations on basis of the interests of the speakers involved. Consequently, there always can be too many candidates for the status of the "common cause of any utterance" - for example any large slice of the history of the universe up to a time before the speaker or speakers were born. That particular slice may be a common cause of two speakers being disposed to assent to "that's red" but it would be a cause of every other disposition of both speakers. (15) The converse is also true. That is, one and the same slice of the world, or event, may be the cause of very different utterances. Thus, everything can be the cause and hence the content of a given belief and the project of interpretation and belief ascription will not get off the ground.

Davidson argues that it is possible to narrow down the choice of the relevant cause by taking into account what is salient for both the speakers and their interpreters. "Salience" is defined by him in terms of similarity of responses. According to him "What makes communication possible is the sharing, inherited and acquired of similarity responses. The interpreter's verbal responses class together or identify the same objects and events that the speaker's verbal responses class together. Furthermore, "the innate similarity responses of child and teacher is necessary for language acquisition", and "a condition for being a speaker is that there must be others enough like oneself". He believes that there is no need to be "worried by the dependence of the concept of the cause on our interest; it is our shared interests, our shard similarity responses, which decide what count as a relevant cause." (16)

However, given the overall Davidsonian framework there is ample reason to worry. It appears that a new element has been added to the principle of charity. A radical interpreter should not only assume the fundamental reasonableness and truthfulness of the alien speakers but also that in their interaction with their environment and the verbal categories used for this interaction the speaker and the interpreter share fundamentally the same interests. Does the Davidsonian approach entitle us to make such an assumption? Quine and Putnam, in very different ways, have argued for the thesis of the interest dependence of ascriptions of causal relations and Davidson has tended to agree with them. But their arguments are not available to Davidson.

Quine has argued that our sense of similarity is fundamental to our thought and language. (17) Without it learning in general and language learning in particular would not be possible — nor would induction and prediction. According to him:

People have to be in substantial agreement, however unconscious, as to what counts as similar if they are to succeed in learning, one person from another, when next to assent to a given observation sentence. Here, then is an irreducible kernel of relativism: all sensory evidence as reflected in observation sentences is relative to the neural organization that determines what different triggerings of nerve endings will favor the same response. Subjects radically at odds in this neural way could never learn observation sentences or anything else from one another. Our training even of a dog, horse, bear, seal, or elephant hinges on a conformity of his inarticulate similarity standards to our own. Such, then, is the residual relativism of empirical evidence. (18)

Given Davidson's long-standing rejection of all hues of relativism the above conclusion should be unacceptable.

Furthermore, the assumption that human beings share an innate similarity-standard is in line with Quine's project of naturalised epistemology since for him the boundaries between philosophy and science are artificial and unnecessary. Davidson's arguments, however, have always had a transcendental rather than a naturalistic aspect to them. Davidson compromises this dimension of his work once he allows biological assumptions to become a part of the principle of charity.

Putnam, on the other hand, has argued that the notion of causation is interest relative in the sense that whether we say that A caused B or not depends on what we take to be the relevant alternative. "The truth of a judgement of the form A caused B depends upon the context and the interests of the people making the judgement (for example what the speakers want to know in a particular context)." (19) A causal statement is true or false "only when a certain framework of pre-understandings is in place, including which conditions should be considered "background conditions" and which conditions should be considered "bringer-about" of effects. [But] there isn't a distinction in the physical facts themselves between background conditions and bringers-about of effects independent of the existence of human beings with human interests and human capacities". (20) Our interests in particular situations and our assumptions about what constitutes the relevant alternatives in any given context are informed by the type of projects we are engaged in. These projects, in turn, are shaped by the socio-historical conditions in which we find ourselves, including the particular historical stage of our cultural development within which we are located. Without the assumption of a great deal of similarity between the projects of diverse speakers of language, and hence without a great deal of uniformity in socio-historical conditions and stages of knowledge-acquisition, the assumption of similarity of interests, at least in its Putnamian version, would not go through. If our interests and acquired similarity standards are given to us within localised frameworks of historically informed knowledge and understanding, then the assumption that there are universal similarity standards is untenable and Davidson cannot add it to his principle of charity.

In Putnam's account of the interest-dependence of the causal nexus, the salient features of a causal connection are picked out relative to a "certain framework of pre-understanding" and knowledge of what counts as relevant. Even sameness of meaning, for Putnam, is interest relative. Putnam embraces this conclusion since he believes that conceptual relativism is an unavoidable but non-threatening consequence of our correct understanding of the relationship between mind, language and the world. Such a conclusion should be unwelcome to Davidson, as one of the main aims of the Davidsonian project is the rejection of scheme/content dualism and its attendant conceptual relativism.

The introduction of the idea of the uniformity of similarity-response into the principle of charity has some further unwelcome consequences for the Davidsonian project. Davidson has argued that intertranslatability is the necessary condition of languagehood. (21) Without being able to translate we cannot attribute beliefs, desires and intentions to a creature. Hence, we cannot make sense of the idea that there can be beings with propositional attitudes, i.e. thinking beings, whose language is not translatable into ours. More recently Davidson has been telling us that interaction among similar creatures is a necessary condition for speaking a language; and that communication is possible between these similar creatures insofar as they share a common stock of similarity responses. Putting these theses together we have the rather surprising conclusion that all language users in the universe must be largely alike in their similarity responses. We have acquired an interesting empirical result by engaging in apriori philosophy of language! The assumption that all human and non human speakers, despite their differing evolutionary histories and biological make-up, must share roughly uniform interests and similarity responses, may help to undermine the threat of relativism, but is, on the face of it, quite implausible. Yet this seems to be one consequence of Davidson's views on language. If we are prepared to bite the bullet and accept that sharing trans-socio-historical interests and cross-species similarity-responses is the precondition for the existence of language, then we are in danger of not recognising divergence and imposing too much homogeneity on the speakers of all languages. A homogeneity that does not stop at the level of ascription of beliefs, desires and truth but also intrudes into the biological substratum.

I have argued that, despite Davidson's protestations to the contrary, his views lead to a non-trivial thesis of indeterminacy of translation, i.e. an indeterminacy in language ascription, which is not captured by the mundane examples of indeterminacy of measurement that he frequently cites. Davidson makes a liberal use of the principle of charity in order to lessen the effect of this indeterminacy. However, the more recent formulation of the principle either introduces an element of relativism or imposes too much homogeneity on all language-users. Consequences that could not be welcome to Davidson.