Kirk Ludwig, University of Florida
I want to thank Simon and Piers for their comments. It is real pleasure to have such perceptive commentators. Both Simon and Piers put the argument for the impossibility of radical interpretation at the center of their comments, and I will do the same with the aim of shedding more light on where the fundamental difficulties lie.
Radical interpretation is the centerpiece of Davidson’s philosophy. It incorporates his suggestion for using a Tarski-style truth theory, adapted to a natural language, to do the work of a compositional meaning theory. It puts at center stage Davidson’s conviction that meaning and related matters must be understood from the third person point of view because of the essentially public nature of language and communication. It exhibits his commitment to explicating the family of concepts involved in understanding language not by giving analyses one by one but by trying to show how evidence which does not presuppose their application can be marshaled in support of a theory deploying them. And in arguing that radical interpretation represents, in some suitable sense, the basic methodological and epistemic position of one speaker with respect to another, Davidson puts himself in a position to argue that whatever must be assumed for radical interpretation to succeed is constitutive of the interpreter’s subject matter.
There is no denying the essentially public character of language. The central question about Davidson’s philosophical work is whether what is undeniable is strong enough to support the view that interpretation can succeed from the extremely restricted evidential position of the radical interpreter. The basic idea is that speakers are by their nature interpretable, the function of language being communication between speakers. Davidson interprets this as requiring interpretability form the standpoint of the radical interpreter. Why? We can see why this follows if we interpret the publicity requirement as in (1).
(1) Any speaker is interpretable in principle by any other speaker in any environment.
This may seem very strong, but if this is true, then given the deployment of the principle of charity in interpretation, we get immediately that any speaker must be mostly right about his environment,
that there can be no radically different conceptual schemes, and that what is not recoverable from purely behavioral evidence is no part of meaning – all theses which Davidson has endorsed.
Lepore and I think that a weaker principle accommodates the essentially public character of language, namely,
(2) Any speaker is correctly interpretable by some speaker in some environment.
This is too weak, though, to support most of what Davidson has wanted to maintain about the essential character of language and the propositional attitudes. The stronger principle, which would underwrite the possibility of radical interpretation, we try to show runs into difficulties from the standpoint of radical interpretation itself.
The radical interpreter, if we are right about Davidson’s ultimate commitments, is restricted to purely behavioral evidence. Davidson often describes the radical interpreter as starting with knowledge of a speaker’s hold true attitudes. However, he also describes this as an intermediate stageofinterpretation. Hepresupposesthattheradicalinterpretercanidentifyholdtrueattitudes on the basis of more primitive evidence. It is important that in describing the evidence as purely behavioral, we are excluding any knowledge of the speaker’s psychological states, and any general knowledge about the type of psychology he has. All of that has to be justified in terms of the behavioral evidence. And it is not in the cards, on Davidson’s view, to appeal to another speaker being a conspecific and to infer from knowledge of one’s own mind what the other’s mind, adjusting for different histories, is likely to be like on that basis. This would be to make the first person point of view in an important sense primary in the epistemology of interpretation.
In approaching the question whether radical interpretation is possible, as Simon noted, we identify as a fundamental assumption of Davidson’s project that the concepts invoked in the theory of interpretation are purely theoretical. They are purely theoretical if their import is exhausted in their role in the theory in accounting for the behavioral evidence for it. We challenge this on the basis of the observation that we seem to apply the concepts of the theory both on the basis of behavior and directly to ourselves. This suggests prima facie that their content is not thought of by us as exhausted by their role in imposing an intelligible pattern on behavior in interpretation. If Davidson had a way of making sense of this compatibly with his overall position, then he would have a response to this objection. We argue that his accounts fail. If the concepts are not purely theoretical, then a real question arises about which of two incompatible theories equally compatible with the evidence is correct. The claim of indeterminacy in interpretation, as we see it, rests centrally on Davidson view of the concepts of the theory of interpretation being purely theoretical.
However, our objection to the possibility of radical interpretation does not rest solely on this. We also aim to raise a difficulty that is more internal to the project, though, as it turns out, one that is ultimately connected in an important way with the issue just raised, as Simon brings out.
It will be helpful to begin by laying out a principle which our argument can be seen as resting on:
(3) If there can be genuine indeterminacy of theories of interpretation, it must be possible for this to make sense from the standpoint of an interpreter of another speaker.
Why impose this requirement? First, it would be odd if we maintained that there was indeterminacy of interpretation, but when we actually engaged in interpretation, we could not accept that that was so. But, second, it should follow from Davidson’s own fundamental assumption that the most basic methodological stance we can take with respect to language and related matters is the standpoint of the radical interpretation. If indeterminacy cannot emerge from that standpoint, we should, by that methodological principle, reject it.
What we argue is that there is at least one stage in the interpretation of another speaker, in accordance with the procedure which Davidson describes, where there are different starting points for projecting a truth theory for the speaker’s language, which cannot be taken by the interpreter as saying the same thing about what sentences in the object language mean. The basic difficulty arises because the interpreter must confirm correlations between conditions in the speaker’s environment and what sentences the speaker holds true in response. This gives rise to sentences of the form (L). At the next stage, one assumes (subject to correction in the light of more evidence and holistic considerations) that this provides a target interpretive T-sentence for a truth theory for the speaker’s language of the form (T) (since the speaker holds true s iff he believes that p ands means that p).
(L) S holds true s att iff p (T) s is true iff p
But the causal structure of the world being what it is, there will be many different correlations or sets of correlations to choose from. And that means, from the point of view of interpretation, many different incompatible starting points, for from the interpreter’s point of view, the different correlations are with different states of affairs, and so assigning meaning on the basis of one will be to say a sentence in the speaker’s language means something different from assigning it on the basis of another. This is underdetermination, not indeterminacy, from the point of view of the interpreter, subject to one caveat, which I come to in a moment. Invoking our principle above, we should conclude that, if these starting points yield theories that are empirically equivalent, radical interpretation is impossible—again subject to one further caveat.
What is the caveat? Well, how could we make out that despite how things look from the interpreter’s standpoint with respect to the differences between the theories he deploys what we really have is indeterminacy? We would have to maintain that each of the theories kept track of the same facts equally well. We could make sense of this by appeal to Davidson’s analogy with measurement theory if we held that what is going on is that the structure the interpreter uses to keep track of the facts of the matter with respect to the speaker, like the numbers in application to physical quantities they are used to keep track of, is essentially richer than the phenomena which the interpreter is keeping track of. When we use the numbers to keep track of temperatures, for example, we do not use all of their features. For the Centigrade and Fahrenheit scales, there is no sense given to an absolute origin. One can choose to assign 0 to any temperature. Then one can choose to assign 100 to any other temperature to fix the relative positions of all temperatures with respect to one another. Since sameness of differences can be physically made sense of with temperatures, we get a scale which is unique up to the choice of origin and a linear factor. There are different ways of mapping numbers onto temperatures because a subset of the features of the number is used in keeping track of the physical phenomena we are concerned with.
This would make sense of indeterminacy from the interpreter’s standpoint, despite the point made above. But this move has a fatal flaw. The problem is that it requires the assumption that the interpreter’s language can mark distinctions that cannot be marked in the speaker’s language. But radical interpretation is supposed to be possible with respect to any language. A speaker of the interpreter’s language would possess a language as rich as the interpreter. That fact could not emerge from the standpoint of radical interpretation if we employed this maneuver. And so radical interpretation fails, and the one maneuver that could rescue it fails as well because it cannot be applied to the interpreter’s own language.
How is this problem related to the assumption about the theoretical character of the concepts of the theory of interpretation? Here is what we think: that assumption must be consistent with what emerges from the standpoint of the radical interpreter, on the assumption that the evidence the concepts are to keep track of is purely behavioral. That is, it can’t look as if there is genuine underdetermination from the interpreter’s standpoint if that assumption is true. But it does look as if there is genuine underdetermination from the interpreter’s standpoint. Therefore, the assumption (relative to the evidence noted), is incorrect.
Simon draws attention to a connection between this argument and our rejection of the view that the concepts deployed in interpretation are purely theoretical concepts, in the sense we have noted. The problem we noted was that we deploy the concepts from the first person standpoint as well, and it is clear that they are not there deployed on the basis of how they keep track of patterns of behavioral evidence. The first person standpoint shows up in the argument we have just given as well. For it is the interpreter’s first person standpoint on his own meanings which gives rise to the view that the speaker he is interpreting does not mark in his language distinctions as fined-grained as those that can be marked in the language of the interpreter. And so in a way the fundamental difficulty for Davidson emerges as the fact that the third person point of view presupposes a first person point of view, that of the person who takes the third person point of view toward another. When the person taking the third person point of view toward another applies concepts to himself in a way other than on the basis of adopting toward himself the third person point of view, his deployment of them to others has to respect the distinctions he marks in his first person deployment of them.
Piers offers a different, and important, argument against the possibility of radical interpretation, namely, that the Principle of Charity, which enjoins us to find the speaker’s beliefs to be about what prompts them, is incompatible with the degree of indeterminacy which results from the restrictions on the evidence with which the radical interpreter must work. Piers suggests that indeterminacy with respect to the contentof beliefs and other propositional attitudes itself entails irrealism or instrumentalism about beliefs. I think that this is not enough, because Davidson can make sense of an invariant content of a given belief across different interpretation schemes, on the model of measurement theory (though as indicated above this move has its problems too). That is, as long as he can argue that in different schemes the same belief with its content is kept track of with different sentences, each of which is acceptable because, as in the use of numbers in keeping track of physical quantities, the structure they are a part of is rich enough to be mapped in different ways on to the phenomena. However, Piers also argues that between different schemes of interpretation there is no reason to expect that it is the same states of the speaker that will be treated as interpretable, and this, I think, raises a further and deeper difficulty. If a disposition grounded in the speaker’s physical constitution may under one scheme be treated as contentful but not under another, then there will not be a scheme invariant set of beliefs and other propositional attitudes for a speaker. Thus, there is no fact of the matter, for example, about whether a given objective state of a speaker is a belief or desire or whether or not it is even a propositional attitude. But as the Principle of Charity seems to presuppose that there is a fact of the matter about the speaker’s beliefs, one of the fundamental constraints on interpretation on Davidson’s view seems to be in conflict with the extent of the indeterminacy the evidence and totality of constraints admit.
I agree with Piers about the seriousness of this problem. One response might be that there are enough constraints given the evidence to determine an invariant set of states as the set of a speaker’s propositional attitudes, but I do not myself see how to make good on this. Supposing that the constraints and data available to the radical interpreter fail, as Piers argues, to determine a unique set of states as the set of the agent’s propositional attitudes, it looks as if Davidson must embrace an instrumentalist view of propositional attitude talk which sits uneasily not just with the Principle of Charity but with much else that he has to say. Yet let me make one suggestion designed to relieve a bit of the pressure on Davidson’s position by suggesting a way of construing the use of the Principle of Charity which makes it consistent with this result, even if it does not make the result consistent with everything else that Davidson says. The idea is to treat the Principle of Charity as a holistic requirement in the following sense. We aim to keep track of behavioral dispositions. There are different theories that will do this equally well, which even differ on which dispositions are treated as attitudes, and so there are no constant propositional attitude nodes across different acceptable theories. Nonetheless, we impose on each acceptable theory the following holistic constraint: it makes most of the speaker’s general beliefs and beliefs about his environment true. So the idea is that we impose the Principle of Charity as a holist constraint on each acceptable theory, without presupposing identity of beliefs across different interpretation theories. Understanding its application in this way, the Principle of Charity is not in conflict with the consequence of the indeterminacy arising from the constraints on the interpreter’s project.
Let me finally turn to the questions that Simon raises at the end of his remarks.
If the authors are right, almost everything Davidson wrote on meaning, truth, language and reality is flawed. (The same is true, of course, of all interesting philosophers.) What, then, should we make of Davidson? What are the lessons to be learned? How does Davidson’s project fit into the bigger picture of 20th century philosophy?
These are large questions, and I can only say some modest things here. But the first thing to be said is that there is much to be learned from Davidson’s work even if we are right that the project fails to meet its ambitions. For in exploring in detail why an ambitious transformative project like Davidson’s ultimately fails we inevitably gain a much deeper understanding of the domain which it aims at illuminating. Think of Kant’s project in the Critique of Pure Reason. One does not have to suppose it to be successful to recognize how much it has deepened our understanding of reason and the mind-world relation. Second, we can say some thing about what can be pursued independently of full success in the project. I think that Davidson is right that the standpoint of interpreting another speaker is methodologically fundamental to meaning and related matters. If we are right, he was wrong to think that it could be carried out from the very special position of the radical interpreter. But this does not mean that we can free the project of understanding language from the third person standpoint. An important follow up project is to ask what additional constraints must be supplied in order to show how interpretation can succeed on the basis of public evidence, and perhaps some additional assumptions – fundamentally, I think, that the speakers we set out to interpret are like us in certain psychological respects which helps to narrow down the range of acceptable theories of interpretation. It must still be the case that the essentially public character of language constrains understanding of meaning and language and the propositional attitudes, and much of what Davidson says about this must still be true. What we want to do now is to rethink how that fact shapes the concepts we deploy in interpretation while we free ourselves from the overly stringent requirements of radical interpretation. In addition, we regard Davidson’s work in the theory of meaning, and in particular his exploitation of a truth theory in pursuit of a meaning theory, as of fundamental importance. The lessons of this have yet to be fully grasped, we think.
Finally, what is Davidson’s place in 20th century philosophy? Though Davidson rejected traditional empiricism and its identification of sensory experience as the source of meaning and the epistemological foundation for knowledge, his philosophy is nonetheless a development out of the broadly empiricist movement in 20th century analytic philosophy, and to understand its historical significance it is important to understand how it grows out of this background. In that tradition which stretches back through Quine to Carnap and the Logical Positivists, and thence to the early Wittgenstein, Russell and Moore, Quine marks an important turning point. It is in terms of Quine’s rejection of certain key doctrines of the Logical Positivists, and further changes Davidson made on the foundation those provided, that we can see the steps by which, through a series of changes internal to the tradition, Davidson turned the traditional picture upside down, giving primacy to a fully objective, even quotidian, the third person point view, rather than, as in the empiricist tradition, the first person point of view.
Carnap was the greatest influence on Quine, just as Quine was on Davidson. We see Quine’s work most clearly in seeing it grounded in Carnap’s work, though developing from in part by way of, the rejection of certain key theses in Carnap. Similarly, we can see Davidson’s work most clearly by seeing it as grounded in Quine’s work, though developing in part from the rejection of certain characteristic features of Quine’s view. Carnap saw sensory experience as the source of meaning and as the epistemic foundation of our theories of the world, and he accepted a sharp distinction between analytic and synthetic truths. Quine rejected the analytic/synthetic distinction of traditional philosophy as insufficiently clear, and, in rejecting it, he rejected as well the allied distinction between the a priori and the a posteriori, a distinction which Carnap had aligned with that between the analytic and synthetic. With this Quine gave up the view that philosophy was fundamentally distinct from the sciences and could provide an a priori foundation for them. He saw philosophy rather as continuous methodologically with the various disciplinary sciences and distinct largely in working with broader categories. In this case, philosophical theories become subject to the constraints which govern the development of scientific theories generally.
In rejecting the analytic/synthetic distinction, Quine likewise rejected as insufficiently clear the traditional empiricist notion of meaning. When he set out to replace the traditional notion with something more scientifically respectable in Word and Object, his fundamental starting point was the observation that “language is a social art.” It is this that motivates the stance of the radical translator as fundamental in reconstructing a scientifically respectable notion of meaning, or at least sameness of meaning, i.e., translation. For it follows from the essentially social character of language, Quine argued, that evidence for its acquisition and deployment must be intersubjective, and, hence, recoverable from overtly observable behaviour. The concepts involved must have their content exhausted by their role in keeping track of the evidence, and, hence, two theories which accommodate the evidence equally well must have the same content. This is the application of the traditional empiricist idea of content as exhausted by contributions to confirmation conditions, at the level of the theory. In a conservative revision of the traditional empiricist theory of meaning, which keyed content to sensory experience, Quine then keyed sameness of meaning to sameness of response to patterns of physical stimulus of the sensory surfaces. This extrudes the traditional subjective basis of content in the empiricist tradition, sensory experience, to the sensory surfaces and renders it in principle intersubjectively available.
Davidson took over from Quine the idea that the third person standpoint – the standpoint of an interpreter of another, in Davidson’s development of the idea – is methodologically fundamental in understanding meaning. He also took over from Quine the idea that, as language is intersubjective, the evidence to which the notion of meaning is responsible must ultimately be purely behavioural evidence, described neutrally with respect to what it may mean or what attitudes it may be connected with. Unlike Quine, however, Davidson did not set out to introduce a new more scientifically respectable notion of meaning, but to understanding and describe the actual workings of language. Meaning as it is ordinarily understood, he held, was a notion recoverable from ordinary behavioural evidence, neutrally described. This is connected with two important, connected, differences between Quine’s approach to the theory of meaning and Davidson’s. The first is that Davidson treats a theory of interpretation as including a full theory of a speaker as a rational agent whose propositional attitudes organize its behaviour into intelligible patterns. The second is that Davidson takes one step further the transformation of the basis of shared meaning in Quine from the subjective to the objective by pushing the shared basis of meaning out from the sensory surfaces to distal events and objects in the environment. In so doing, he rejects the last vestige of empiricism in Quine’s philosophy.
The upshot is that the basis of meaning, thought and knowledge in Davidson’s philosophy is the intersubjectively shared distal environment. Knowledge of our own minds, the minds of others, and the external world are part of the basic package that makes thought and meaning possible. This vision of our embedding in the world is as far from traditional empiricism as could be imagined. It is, in a certain direction, as this tracing of the stages of its development shows, the culmination of a line of response initiated by Quine to empiricism as represented by the Logical Positivists, and centrally by Carnap. In a very loose analogy, we may think of the line of development Carnap- Quine-Davidson as similar to that of Locke-Berkeley-Hume in the modern period, in the sense that this development in the history of ideas, in retrospect at least, appears to be almost determined by a logic internal to the philosophical problems they were responding to and their particular position in the historical development of ideas about them.
I end with a word about the structure of the problem area, the mind-world relation, to which both traditional empiricism and Davidson’s philosophy is a response. Each can be seen as taking a different starting point and thereby rendering unproblematic for itself something that emerges as problematic for the other. The most difficult challenge for empiricism is the recovery from the subjective foundation it provides an objective world as an object of knowledge. The corresponding difficulty for the intersubjective foundation Davidson offers is how to make sense of knowledge of the subjective, in the sense of knowledge of our own thoughts and meanings from the first person point of view. It is not a challenge that Davidson overlooked or failed to respond to. But part of the burden of our book is that the challenge has not been adequately met. And then the larger question is whether there is any position which can accommodate the role that the first person plays in our self-conception compatibly with making sense to ourselves of our having secured reflective knowledge of a fully objective mind-independent world.