Issue #19 -- December 2007. Pp. 79-87
Incommensurability and Interpretation
Copyright © by Anthony D. Baldino and Sorites
Incommensurability and Interpretation
by Anthony D. Baldino
According to the defenders of incommensurability, many important competing scientific theories, theories that replace each other through scientific revolution (e.g. Ptolemaic and Copernican astronomy, Newtonian mechanics and Einsteinian relativity), are said to be incommensurable. The theories in question are committed to radically different central or fundamental principles; the principles of one theory undermine or exclude those of its rival; they thereby undermine or exclude the whole of the rival theory built upon those fundamentals. That is, when the fundamental principles of one theory are in place, those of the other theory cannot be, and all of the concepts and so-called facts that have their foundation in these displaced principles are rendered absurd or meaningless -- the theories are thus said to be incommensurable with one another.
Davidson's argument against non-intertranslatable languages and incommensurable conceptual schemes is based on his theory of meaning, which in turn rests on his conception of radical interpretation. Meaning is given in the production of a truth theory for the agent being interpreted. The truth-theory is produced via radical interpretation -- the interpreter must proceed only upon the empirical evidence of which sentences under which circumstances are held true by the interpretee. Since Davidson does not think that under such constraints the interpreter can assign beliefs to the interpretee independent of assigning meanings to her utterances, the production of a truth-theory must proceed by simultaneously producing the interrelated constructs of belief and meaning. Such a truth theory, which will take the form of T-sentencesFoot note 1, will be accepted as an interpretation of the sentences of the interpretee if the totality of T-sentences optimally accounts for all the empirical evidence we have assumed access to. Such a theory, then, does not produce meaning sentence by sentence but holistically; a sentence is given an interpretation by the right hand side of the relevant T-sentence if the interpreter knows the empirically adequate truth-theory to which that sentence belongs.
A fuller description of Davidson's theory of meaning will, of course, have to say more about how we go about producing such a truth-theory, that is, how beliefs and meanings are to be produced simultaneously and the related question of how a truth-theory is judged empirically adequate. But with even just this much said about radical interpretation, we might glimpse a speedy dismissal of incommensurability. We might be tempted to use an argument Davidson uses explicitly against the skeptic for a quick refutation of the possibility of concepts of one point of view being incommensurable with those of some other viewpoint.
Davidson writes:
The point here is that meaningful error presupposes a foundation in truth, but essentially the same point is sometimes put by Davidson more explicitly in terms of the project of interpretation, that is, that meaningful disagreement between interpreter and interpretee must have a foundation in agreement.
A skeptic may or may not be committed to meaningful error, and so perhaps this is a good anti-skeptical argument. But the incommensurabilist is absolutely not committed to meaningful disagreement -- it is exactly that that the incommensurabilist wants to deny -- holders of incommensurable theories are doomed to misapprehend and talk past one another. And so Davidson's argument, stated as baldly as that, cannot refute the incommensurabilist. The incommensurabilist denies both meaningful agreement and meaningful disagreement between holders of incommensurable theories, and so it does not matter to him that the latter entails the former.
It would seem that if the radical interpretation approach to meaning is to be the source of an anti-incommensurabilist argument, we must look more deeply into what constitutes that approach to reveal that argument. In particular, we must pay more attention to what conditions on an acceptable truth-theory must be imposed and what must be done in the process of interpretation to meet those conditions.
The principal difficulty confronting the radical interpreter concerns the interdependence of belief and meaning that was mentioned earlier. As a result of this interdependence, the interpreter has to find a way, given the evidence of which sentences an agent holds true under which conditions, to arrive at both a theory of meaning and a theory of belief. Davidson's familiar proposal for accomplishing this is to hold one of these unknowns constant: we hold belief constant and solve for meaning. Davidson writes, `We cannot take even a first step towards interpretation without knowing or assuming a great deal about the speaker's beliefs.' (Davidson 1984, 196)
The equally familiar principle by which we manage the task of holding belief constant is what Davidson calls the `principle of charity' whereby we attribute beliefs to the agent so that we minimize unexplained error on the part of the agent. Given our empirical evidence which consists in assent and dissent to sentences in particular circumstances, we begin to generate correlations between sentences and truth conditions by `assigning to sentences of a speaker conditions of truth that actually obtain (in our own opinion) just when the speaker holds those sentences true' (Davidson 1984, 196) and, presumably, do not obtain when the speaker holds the sentences false. This policy is `to be modified in a host of obvious ways' (Davidson 1984, 152) -- the goal is not the absurd one of making error or disagreement impossible, nor is it to simply maximize agreement and minimize error, but it is to `countenance( ) error where it can be best explained' (Davidson 1984, 318). A truth-theory is an acceptable interpretation for a speaker, Davidson thinks, if it satisfies certain formal constraints and if it accounts for the empirical evidence of assent and dissent under the charitable constraint outlined.
The significant point about this principle of charity is that, since it is `not an option, but a condition of having a workable theory, it is meaningless to suggest that we might fall into massive error by endorsing it' (Davidson 1984, 318). Charity is not an option because it is that principle which supplies the truth-theory with empirical content. It is only by some such principle that we can link what a speaker says with what we observe in the world (the conditions of utterance). This very basic methodology of interpretation instructs the interpreter that most of the time he may take it to be the case that what the speaker holds true is true, and it is by virtue of this methodological principle that the interpreter can `take into account the causal interaction between world and speaker in order to find out what the speaker means, and hence what he believes' (Davidson 1986, 332).
The argument against conceptual schemes based on the principle of charity is then this: since charity is necessary to interpretation, not optional, shared agreement and conceptual contact between any two language-users is thereby necessary as well, since charity guarantees agreement and contact. The incommensurabilist appears to be confronted by the following uncomfortable dilemma: if he denies the principle of charity, he severs any ties between the interpretee's language and the observable features of the world, and it becomes rather mysterious how the person who uses such a language could have ever learnt the language in the first place; if he accepts the principle of charity as necessary, he accepts shared agreement as necessary and can no longer make his radical semantic claims.
This argument seems far more promising than the first simple one that disagreement rests on a foundation of shared agreement. This one seems to ensure agreement and, a fortiori, conceptual contact between interpreter and interpretee and thus seems to block the very possibility of large-scale failures of conceptual overlap and to render incoherent the very notion of incommensurable conceptual schemes. And it does so without appeal to meaningful disagreement, something the incommensurabilist explicitly denies.
However, I think if we look more closely we will see that this is not the case, and that, in fact, the two arguments resemble each other more closely than it may first appear and may ultimately be rejected for very similar reasons.
Whenever we approach a language to be interpreted, the argument we have been considering tells us that, in order to succeed, we have no choice but to employ the principle of charity and thus that we can never be led into failure because we have employed that principle. It appears to follow from this that there can be no learnable language that does not share substantial agreement and conceptual contact with the language we come to the interpretation with, and thus, it seems Davidson has the argument that he claims he is after in `On the Very Idea of a Conceptual Scheme', one that has as a conclusion that `translatability into a familiar tongue (is) a criterion of languagehood' (Davidson 1984, 186), and, therefore, that there can be no languages that are not intertranslatable.
If one is convinced, as I am, that Davidson's account of meaning is essentially correct, is one also forced to abandon the thesis of incommensurability as untenable? Must one deny Davidson's principle of charity to be an incommensurabilist? I do not think so.
It might seem that the incommensurabilist must deny charity since charity implies agreement, which in turn implies conceptual contact. But note that interpretation in violation of charity is interpretation without shared agreement and therefore interpretation that posits wholesale disagreement between interpreter and interpretee. Davidson's compelling argument prohibits this, but that is not a problem for the incommensurabilist because, just as we had reason to point out before, the incommensurabilist is absolutely not committed to interpretation that posits wholesale disagreement. According to the descriptions given by Kuhn and Feyerabend, it is not that holders of incommensurable points of view disagree with one another -- rather, they do not understand each other. But if they do not understand one another, they have failed to interpret one another. And if they have failed to interpret one another they cannot have possibly interpreted each other in violation of the principle of charity. They have simply come up short in their effort to produce an interpretation -- they cannot clearly make out what the other is trying to say, what the concepts they are using can possibly mean. The incommensurabilist describes wholesale incomprehension not wholesale disagreement, not interpretation that violates the principle of charity.
The dilemma we set for the incommensurabilist is a false one -- the incommensurabilist does not need to deny the very plausible principle of charity described by Davidson and thus does not cut off language from the world and rob words of all possible empirical content. Holders of incommensurable points of view do not come up with uncharitable interpretations of one another. They fail to interpret; they do not grasp what the other means at all. Why do such failures of interpretation take place? Ironically, maybe, they take place for a very Davidsonian reason. Interpretations, in cases of incommensurable theories, are impossible because in such cases the principle of charity cannot be satisfied. And, as Davidson tells us, nothing could count as a successful interpretation if it violates the principle of charity.
So the incommensurabilist can wholeheartedly agree that interpretation cannot be successful unless the interpreter satisfies the principle of charity, unless the interpreter can use what he observes to be true in the world to assign meanings to the interpretee's utterances. Indeed he can also concede that whenever the principle of charity is met in a satisfactory way a successful interpretation is achieved. But, importantly, he need not concede because it has not been shown that, whenever an observer uses what he observes to be true in the world to assign meanings to utterances, the interpreter has satisfied the principle of charity or succeeded in interpretation. The incommensurabilist may hold that a successful interpretation depends on the use of what is observed to be true in the world, but he need not also hold that all interpreters are capable of making the same observations -- in fact that is just what he denies in stating the incommensurabilist position. (Surely it is wise to deny this. Although these are clearly not cases of incommensurability, it is still the case, for example, that I cannot make the same observations as someone who understands cricket while watching a match with him, and that someone who does not read music or comprehend principles of tonality cannot make the same observations when looking at a musical score as someone who can.)
A successful interpretation surely rests on shared agreement. It simply does not follow from this that the interpreter, without going well beyond the language he came to the interpretation with, has the conceptual resources to agree with the interpretee, and it does not therefore follow that the interpreter is able to give an interpretation from that original language. That is precisely what the incommensurabilist has been denying from the start, and the present argument does nothing whatever to block such a denial.
It is important to see that the response I offer on behalf of the incommensurabilist does not deny that the principle of charity is a necessary constraint on interpretation nor that any interpretation that adequately satisfies this principle (and certain other formal constraints) is a successful interpretation. The incommensurabilist merely denies that, in all instances, conceptual contact between interpreter and interpretee is sufficient for the would-be interpreter to describe the truth conditions of the interpretee's utterances -- that is, the interpreter lacks the conceptual resources to successfully meet the constraint of charity. In cases of incommensurability, one may use what one observes to be true of the world to assign meanings, one may look as best one might for shared agreement, but one will never succeed in satisfying the principle of charity, will never succeed in minimizing unexplained error. One may, to be sure, do the best one can to minimize unexplained error with the conceptual resources one has, but if the points of view are incommensurable in the way Kuhn and Feyerabend imagine, such an `interpretation' will be just a great mass of unexplained error -- the alternative point of view appears absurd, almost as if it were not a point of view at all.
In short, just because it has been shown by Davidson that you cannot possibly succeed without charity, it simply does not follow that you always have the concepts required to be truly charitable. In a case like this, the only way you could truly minimize error, truly satisfy charity, would be to experience a conversion of the sort the incommensurabilist describes, that is, to go native with respect to the concepts of the interpretee. In such a case only absurdity is detected when you obstinately insist on matching your interpretee's comments with observations about the world at the time of utterance that you are able to make only from within the confines of your original point of view. Understanding -- true interpretation -- is only achieved when the grip of your old concepts are loosened and you allow yourself to learn your interpretee's point of view from scratch, by going native, without detour through your own native tongue.
After such a conversion, successful interpretation becomes available.
After you have learned to observe the world as your former rival does, you have no trouble using what you are now able to observe to be true of the world to assign meanings to her sentences, no trouble locating a mass of shared agreement, no trouble minimizing unexplained error. The incommensurabilist has not severed meaning and interpretation from what can be observed in the world as he has been accused of doing; he merely claims that not all languages are capable of expressing the same observations, and, therefore, not all language users are prepared to make the same observations.
We are almost ready to conclude that the argument from the principle of charity neither succeeds in showing that the criterion of languagehood is translatability into a familiar tongue, nor that there could be no languages that fail of intertranslatability, nor that the semantic claims of the incommensurabilist are incoherent.
One final attempt at a critique of incommensurability from the point of view of radical interpretation, however, suggests itself based on what we have said so far. We have noted that when an interpreter confronts an incommensurable conceptual scheme or untranslatable language, the so-called language or scheme appears absurd, almost as if it were not a scheme or a language at all. Davidson has said that is `tempting' to say of such a situation that `nothing ... could count as evidence that some form of activity could not be interpreted in our language that was not at the same time evidence that that form of activity was not speech behavior.' (Davidson 1984, 185) Why should the interpreter not follow this hunch and conclude that the sounds issuing from his `rival's' mouth are not really the statement of a point of view at all -- they are merely sounds? Why, after being as charitable as he can be from his own point of view and still coming up with nonsense, should the would-be interpreter not just conclude that there was never anything there to interpret in the first place?
As to this argument, however, Kuhn, Feyerabend and I are all almost perfectly in accord with what Davidson goes on to say: `As fiat, the thesis (that a form of activity that cannot be interpreted as language in our language is not speech behavior) lacks the appeal of self-evidence.' He adds, `If it is a truth, as I think it is, it should emerge as the conclusion of an argument.' (Davdison 1984, 185,6) We all can agree that there should be an argument for such a thesis. Davidson thinks that he has produced one, and so that the thesis is true. If I am right in what I have said so far, no such argument has been produced.
To say that the thesis in question `lacks the appeal of self-evidence' is not at all to say that the thesis has no intuitive appeal: why should we persist in seeking an interpretation when all we can see, from our native point of view, is nonsense? But it seems to me that the temptation to say that we are not dealing with speech behavior in such a case only derives from the harmless fact that, if we find that the person we try to interpret is speaking of none of the things we know how to speak about, we are left unable to even clearly imagine what he or she is talking about. The fact that we cannot imagine what our would-be interpretee is talking about is, as I say, harmless for Kuhn and Feyerabend since that is exactly what they think the meeting of incommensurable points of view is like. So the fact that we cannot imagine what they are speaking about could not be considered proof that they are not speaking about anything without obviously begging the question against the incommensurabilist. Moreover, not only do Kuhn and Feyerabend acknowledge the fact that one is tempted in these cases to conclude that one is dealing with mere nonsense, they exploit the fact to explain the stubborn resistance scientists of the old-guard typically display in the face of scientific revolution. The temptation may be natural (often only the greatest of scientists succeed in overcoming it), but there has been no argument that has successfully shown that succumbing to such a temptation would always be justified.
All that has been said in this essay up till now still leaves open the question of whether there actually are alternative conceptual schemes or languages that are not intertranslatable. It seems to me, though, that if there is any hope of finding such schemes, the history of science would be one of the more likely places to uncover them. How far do we really think the Aristotelian would get toward interpreting what is said in the modern physics laboratory if he limited himself to matching what is said there to what he knew how to say before he entered the time-machine that transported him there? How confident ought we be that there is enough in common between the schemes in question for the Aristotelian to build modern concepts out of the ones he came to the interpretation with?
The conclusions reached thus far in this essay indicate that if we wanted to answer these last two questions we would have to consider the actual contents of the Aristotelian and modern points of view. That is, we cannot rule out, via some a priori argument, the very possibility of alternative conceptual schemes, and so, the incommensurabilist's commitment to such a possibility does not render his view of scientific theories untenable.
Of course, if we are going to say that there may be alternative conceptual schemes, that there are schemes that the radical interpreter has no access to while he remains within the conceptual confines of his original point of view, we should be prepared to give an account of how the stymied interpreter could go about grasping those foreign points of view -- if we could not, we would begin to wonder how those who are already comfortable with that point of view ever got that way. The incommensurabilist does have such an account, one that I have alluded to earlier: incommensurable points of view are not grasped via translation but by a process which can be called `going native' -- they are learned from scratch without attempting to translate them into one's native scheme.Foot note 2
But this recourse to the notion of going native might appear problematic within the larger context of this paper since the expressed goal is to show that nothing in the radical interpretation model militates against the thesis of incommensurability. It might be thought that, once we admit that to grasp the meaning of some foreign sentences it may be necessary to go native, we have simply abandoned the radical interpretation model. For upon that model, all there is to meaning is what is given in the radical interpreter's truth-theory, and that theory is constructed by correlating interpretee sentences with truth conditions given in the language of the interpreter. If the view is that all there is to meaning is given in the truth conditions that the radical interpreter must give in his own language, what place is there for going native once the view is adopted?
This apparent tension, however, rests on a fairly trivial confusion. The radical interpretation account is meant to be a theoretical account of meaning or linguistic competence -- what a speaker of a language knows in knowing how to speak her language is what is given in the truth-theory of the radical interpreter. But the radical interpretation account, obviously, is not and was never meant to be an account of how we come to know a language. The radical interpreter must have a language to interpret into, and we all come to know at least our first language without a language to interpret into. In fact, not only do we not learn our first language from the point of view of the radical interpreter, we rarely learn any language or interpret any speech behavior from that point of view. Radical interpretation, if actually carried out, would be an immensely arduous affair, and in most ordinary circumstances we interpret others, whether they speak our native language or some foreign language we are trying to learn, by making use of all sorts of linguistic conventions, shortcuts which the radical interpreter cannot allow himself. But, to repeat, admitting this fact does not strip from the radical interpretation account its ability to fulfill its purpose, that of giving an account of what meaning is, not of how we come to know how to produce and understand sounds and signs that have meaning.
But if radical interpretation is silent regarding how we come to know a language, going native is not incompatible with that account. That is, to be in a position to be a radical interpreter, one must know a language, and so some language must be learned outside the process of radical interpretation, learned `from scratch, as a child learns' (Feyerabend 1987, 266). If the radical interpretation model is compatible with a child's learning from scratch, it cannot be incompatible with an adult's (one who already has a language) learning that way -- and indeed it was never meant to be incompatible with this.
Nothing the incommensurabilist says prevents him from agreeing that all there is to linguistic competence is what is given in the truth-theory of a successful radical interpretation (one that meets the empirical and formal constraints). Nothing the advocate of radical interpretation holds should prevent him from recognizing that not just anyone can succeed in the enterprise of radical interpretation. Surely, for instance, someone without a language would first have to learn a language from scratch in order to succeed, and there is nothing essential to the radical interpretation model which prevents one from holding that someone without the appropriate native tongue would also inevitably fail if she took up the stance of the radical interpreter without first going native with respect to the language interpreted. Once a language is learned by the would-be interpreter from scratch or by going native, the meaning of the sentences of that language are readily given in the homophonic translations which the interpreter may now give as the truth-theory for the language learned.
Davidson's main target in his compelling essay `On the Very Idea of a Conceptual Scheme' is the third dogma -- the scheme/content distinction. A central premise of that argument involves Davidson's also compelling principle of charity as a constraint on interpretation. Davidson goes on to assert, as a sort of corollary of the main argument, that the principle of charity also contravenes the thesis of incommensurability as advocated by Kuhn and Feyerabend.
This last move, I have tried to establish, is a misstep. But this misstep does not impinge on the main thrust of Davidson's essay -- it does not vitiate the argument against the third dogma. That is because incommensurability does not imply any objectionable scheme/content distinction. Holders of incommensurable theories have different concepts. These concepts are supposed to somehow exclude one another, but in all of the ways Kuhn and Feyerabend have tried to explain how this happens, they never try to say that they exclude each other by carving up some content differently by using different schemes.
In fact, Davidson never really accuses the incommensurabilist of making this mistake. His argument against the incommensurabilist, as I say, stands apart from the main argument against scheme/content. It is the principle of charity that is supposed to undermine incommensurability and, as I have tried to show, that component of Davidson's discussion is flawed. If this is so, then the denial of the scheme/content distinction, the principle of charity, and incommensurability can all coexist.
This is an important result in that it might help reconnect some segment of philosophy of science with a segment of philosophy of language. Many greeted incommensurability as a radical, intriguing, and powerfully informative conjecture about the history of science. But misgivings, like Davidson's, about the thesis's semantic integrity did a lot to cast a dark shadow of doubt over it and restrained much of the enthusiasm that first greeted the thesis. Over time, interestingly, some in the philosophy of science, leaving aside the semantic doubts, have come back to refer to and discuss incommensurability positively. Often enough, the thesis is referred to as a way of shedding light on some specific scientific impasse or controversyFoot note 3. Those who use the thesis as this kind of elucidative tool do so by simply ignoring the semantic disrepute that the thesis had fallen into in some circles. In this essay, I have tried to establish that instead of ignoring the thesis's suspect reputation, we should challenge it.
Anthony D. Baldino
Rutgers University
Anthony [dot] Baldino [at] gfigroup [dot] com
[Foot Note 1]
A sentence of the form, `S is true if and only if R', where `S' is the sentence to be interpreted and `R' is the interpretation.
[Foot Note 2]
Feyerabend describes going native: `Analogies, metaphors, negative characterizations, bits and pieces of cultural history are used to present a new semantic landscape with new concepts and new connections between them. Historians of science proceed in a similar way, but more systematically. Explaining, say, the notion of `impetus' in sixteenth- and seventeenth-century science, they first teach their readers the physics, metaphysics, technology, and even the theology of the time: in other words, they too introduce a new and unfamiliar semantic landscape, and then show where impetus is located in it.' (Feyerabend 1987, 266.)
[Foot Note 3]
A very compelling instance of this as it relates to quantum mechanics can be found in Albert 1992.