First Draft, 28/4 -98
1. Introductory Remarks: Quine, Wittgenstein and Davidson on Philosophy and Linguistic Diversity
Despite their very different ways of doing philosophy, Quine and the later Wittgenstein appear to have much in common. The criticism of mentalist and platonist conceptions of meaning; the emphasis on language as a public, social art; the rejection of the claims of traditional conceptual analysis; the disapproval of any substantial account of truth; these are all central themes in both Quine’s and Wittgenstein’s writings. Such similarities may give the impression that the lessons to be learned from the two thinkers are basically the same.
However, I agree with those who hold this consonance to be only superficial. #1 If one takes a closer look, it becomes clear that the differences between Quine and Wittgenstein go far deeper than the similarities. For example, even if both philosophers emphasise the public use of language, their notions of ‘public use’ are very different. Quine’s conception is strictly behaviouristic: for him, overt language use is basically the emission of sounds and marks from human organisms situated in a purely physical environment. Wittgenstein, on the other hand, treats language use as a species of ordinary human action that take place in everyday, irreducibly social settings.
Another difference, which is perhaps even more fundamental, is Quine’s and Wittgenstein’s antagonistic conceptions of the role and character of philosophy itself. Quine firmly rejects any attempt to draw a sharp line between factual inquiries on the one hand, and philosophical or conceptual inquiries on the other. According to Quine, good philosophy is continuous with systematic, scientific theory-building. The following passage is representative:
All these luminaries and others whom we revere as great philosophers were scientists in search of an organized conception of reality. Their search did indeed go beyond the special sciences as we now define them; there were also broader and more basic concepts to untangle and clarify. But the struggle with these concepts and the quest for a system on a grand scale were integral still to the overall scientific enterprise. The more general and speculative reaches of theory are what we look back on nowadays as distinctively philosophical. What is pursued under the name of philosophy today, moreover, has much these same concerns when it is at what I deem its technical best. [Quine 1981:191]
In an opposite vein,Wittgenstein is careful to separate philosophy from science. Already in the Tractatus, he disclaims the notion that philosophy deals with what is ‘more general and speculative’, declaring instead that “[o]ur problems are not abstract but perhaps the most concrete that there are.” [#5.5563] In his later thought, he goes one step further and rejects consistently all systematic theory-construction within philosophy. In an often quoted passage, he writes:
It was true to say that our considerations could not be scientific ones. […] And we may not advance any kind of theory. There must not be anything hypothetical in our considerations. We must do away with all explanation, and description alone must take its place. And this description gets its light, that is to say its purpose, from the philosophical problems. These are, of course, not empirical problems; they are solved, rather, by looking into the workings of our language, and that in such a way as to make us recognize those workings: in despite of un urge to misunderstand them. The problems are solved, not by giving new information, but by arranging what we have always known. [Wittgenstein 1974:#109]
One important aspect of the difference between Quine’s and Wittgenstein’s conceptions of philosophy, is their opposite attitudes towards linguistic diversity. According to Wittgenstein, philosophical problems arise because we go astray in the ancient city of language. To find our way about, we have to remind ourselves of how the streets that we habitually walk every day actually look, and how they are connected to each other. Naturally, this will involve reminding ourselves of the more or less detailed differences between various buildings and blocks. Hiding or obliterating such differences only makes the necessary orientation more difficult, or even impossible. Indeed, it might have been precisely such obliteration which made us go astray in the first place. Being able to resist the inclination to impose uniformity on language is therefore, according to Wittgenstein, a central philosophical virtue.
Quine’s viewpoint is quite different. He wants philosophers to be like city-planners who replace old, irregular housing areas by new, uniform blocks. In a scientific spirit of systematicity and simplification, Quine thinks we should dispense with all “quirks of usage that we can straighten.” [Quine 1960:158] Allegedly, philosophers should not try to give a wholly faithful description of actual language use, but rather improve language by fitting it into an austere “canonical notation” that employs only the constructional resources of first-order predicate logic. [1960:226ff.] Only such “regimentation” - a process of “coax[ing]” and “trimming” that may even require “some torturing” - makes it possible to perform the supposedly necessary “clearing of ontological slums.” [1987:157; 1960:180, 275]
The construction of theoretical models within science is a process that usually involves far-reaching simplification and idealisation. Deviant experimental data are adjusted, and sometimes even ignored. Quine’s dismissing as “waste cases” or “don’t-cares” the parts of natural language to which his regimented notation fails to make justice, is therefore fully in line with his general view of philosophy as a scientific endeavour. [1960:182] And Wittgenstein’s opposite conviction, that apparent ‘don’t-cares’ are actually what philosophers should care most about, forms a very important part of what he means by saying that philosophy and science are separate activities.
Quine’s and Wittgenstein’s different attitudes towards philosophy and linguistic diversity are not just matters of temperament or style. On the contrary, these attitudes lie at the very heart of the two philosophers’ ways of thinking. What we have here are two entirely different diagnoses of what philosophical problems are and what it means to solve them. Nothing is more wrong-headed, according to Quine, than to start from a sharp distinction between scientific and philosophical, or between factual and conceptual, problems. He regards all such distinctions as varieties of the untenable dichotomy between analytic and synthetic sentences. Consequently, he argues that a philosophical problem can be properly solved only if treated as a scientific one. Good, fruitful philosophy is possible only if the philosopher aquires “the robust state of mind of the natural scientist who has never felt any qualms beyond the negotiable uncertainties internal to science.” [1981:72]
Wittgenstein’s diagnosis is the converse. In fact, he thinks the kind of assimilation of philosophy to empirical science that Quine suggests is the most important source of metaphysical abberation:
Philosophical investigations: conceptual investigations. The essential thing about metaphysics: it obliterates the distinction between factual and conceptual investigations. [1967:#458]
Indeed, there are good reasons to believe that Wittgenstein would regard Quine’s claim to limn “the true and ultimate structure of reality” as a paradigmatic instance of metaphysical nonsense. [Quine 1960:221] #2
Donald Davidson’s philosophy is strongly influenced by Quine. Davidson shares the belief that there is no clear distinction between the conceptual and the empirical, and that, consequently, there is no sharp line to be drawn between philosophical and scientific inquiry. Moreover, Davidson is no enemy of theory-building within philosophy. On the contrary, he holds the construction of systematic theories of meaning for natural languages to be a central philosophical task, and, in line with Quine’s predilection for extensionality, he believes that such theories should use only the simple formal resources of first-order predicate logic to describe the logical structure of and connections between sentences. However, Davidson’s philosophy also incorporates elements with a seemingly Wittgensteinian ring. In particular, he does not propose to regiment or otherwise modify language, but only to describe it as it is actually used, in all its diversity. Davidson writes:
Like Quine, I am interested in how English and languages like it (i.e., all languages) work, but, unlike Quine, I am not concerned to improve on it or change it. (I am the conservative and he is the Marxist here.) I see the language of science not as a substitute for our present language, but only as a suburb of it. [Davidson 1985:172]
So, even if Davidson conceives of philosophy as a theoretical enterprise that is continuous with empirical science, he claims not to strive for an idealised and simplified picture of language. On the contrary, he stresses “the importance, if one is interested in the semantics of natural languages, of looking for a theory that fits all the relevant phenomena.” [1985:173; italics added.]
In this paper, I shall argue that Davidson’s philosophy suffers from an irresolvable tension between these two tendencies. Either you give a theoretical, systematic account of language, and put up with the fact that this is an imposed description which gives only a very simplified and partly distorted picture of actual usage and competence. Or, you give a description that involves no such simplifications and distortions, and acknowledge the fact that this description can neither be complete, nor fulfil the scientific ideals of simplicity and systematicity. Davidson’s wish to have it both ways is fundamentally misguided.
This tension becomes especially prominent in Davidson’s later (post 1984) writings. In his earlier papers, he seems to take for granted that natural languages are what Tarski called ‘languages with a specified structure’, and that all speakers of a natural language share an ability that can be represented by one systematic theory of meaning. In his later writings, however, this presupposition is problematised. Davidson starts to emphasise that “actual linguistic practice is only loosely related to any fully and precisely specified language, with phonetics, semantics and syntax made explicit,” [Davidson 1994:2] and pays much attention to various ‘quirks of usage’, such as malapropisms, nicknames and slips of the tongue. In effect, Davidson gives up the notion of natural languages as languages with a specified structure, and argues instead that ordinary usage and communication involves truly creative and unforeseeable elements.
As a consequence, one would perhaps expect Davidson to give up the notion of systematic meaning altogether, thereby rejecting root-and-branch one of the grand ideas to which he owes his philosophical reputation, namely, the idea that meaning-theories should assume the form of Tarskian truth-definitions. This is not a step Davidson takes, however. Even if he abandons the idea of representing the ability of speaking, say, English, by means of one systematic meaning-theory, he does not give up entirely the aspiration towards theoretic systematicity. What he rejects is rather the idea that one theory is enough. In other words, the insight that actual linguistic practice is far from systematic is not straightformardly countered with a corresponding de-systematisation at the theoretical level; rather, Davidson chooses to represent each speaker as armed with a huge set of systematic meaning-theories that are constantly revised and modified. (This point is developed in section 2 and 3 below.)
Why does Davidson draw this seemingly far-fetched conclusion? Why not simply give up the idea of systematicity altogether? Or, alternatively, acknowledge that systematic accounts of language are always simplified and partly distorted pictures of real-life linguistic practice? Towards the end of the present paper, I will speculate that this has to do with another tension within Davidson’s philosophy, namely, his vascillating between treating meaning-theories as things that competent speakers in some substantial sense actually know and make use of, and viewing such theories merely as modelling the practical abilities of speakers. Surely, the official policy - at least in his later writings - is the latter, more modest conception:
To say that an explicit theory for interpreting a speaker is a model of the interpreter’s linguistic competence is not to suggest that the speaker knows any such theory. […] claims about what would constitute a satisfactory theory are not […] claims about the propositional knowledge of an interpreter, nor are they claims about the details of the inner workings of some part of the brain. They are rather claims about what must be said to be a satisfactory description of the competence of the interpreter. We cannot describe what an interpreter can do except by appeal to a recursive theory of a certain sort. [Davidson 1986:163; original emphasis.]
It is far from clear, however, that Davidson sticks to this modest attitude in practice. In fact, I will suggest that his resistance to giving up the idea that linguistic meaning must be systematic, probably reflects a much more substantial conception of the role of meaning-theories than he is willing to admit.
In order to clarify the issues that I have just sketched, I shall consider in some detail Davidson’s well-known 1986 paper, “A Nice Derangement of Epitaphs.” This paper is often referred to as a land-mark in Davidson’s philosophical development, since it introduces many themes that are central to his later philosophy. In the next section, I will give a brief summary of the main arguments in that paper. Thereafter, I will try to develop my criticism of Davidson’s way of handling the phenomenon of linguistic irregularity.
2. “A Nice Derangement of Epitaphs” - A Summary
“A Nice Derangement of Epitaphs” constitutes an attack on a certain picture of meaning and communication; a picture, says Davidson, which is prevalent among philosophers and linguists. His often quoted and boldly stated conclusion is that “there is no such as a language, not if a language is anything like what many philosophers and linguists have supposed.” [1986:174]
The picture that Davidson is attacking consists of three fundamental assumptions:
(1) First meaning #3is systematic. A competent speaker or interpreter is able to interpret utterances, his own or those of others, on the basis of the semantic properties of the parts, or words, in the utterance, and the structure of the utterance. For this to be possible, there must be systematic relations between the meanings of utterances.
(2) First meanings are shared. For speaker and interpreter to communicate successfully and regularly, they must share a method of interpretation of the sort described in (1).
(3) First meanings are governed by learned conventions or regularities. The systematic knowledge or competence of the speaker or interpreter is learned in advance of occasions of interpretation and is conventional in character. [1986:161]
On this received picture of language, a prerequisite for understanding an utterance is that one is already prepared for it in advance, so to speak. No room is left for truly innovative yet understandable usage; the principles (1)-(3) imply that there are no real imagination involved neither in speaking to a person nor in understanding what he or she says. Linguistic competence is depicted as something entirely regular, and understanding is conceived as the mere application of a rigid, pre-established machinery for interpretation.
But of course, actual language use isn’t like that. In many forms of real-life communication, inventiveness and imagination do play a crucial role. Consequently, counterexamples to the received picture are easily found. Experimental literature provides extreme counterexamples. However, we do not have to go to literature; our ordinary, pedestrian usage is also filled with irregular, unforeseeable, yet meaningful and understandable uses of words. The sort of counterexamples that Davidson focuses on are malapropisms - more or less ludicrous or funny uses of words, such as ‘We’re all cremated equal’ and ‘We need a laugh to brake up the monogamy’. Davidson says,
[m]alapropisms introduce expressions not covered by prior learning, or familiar expressions which cannot be interpreted by any of the abilities so far discussed. Malapropisms fall into a different category, one that may include such things as our ability to perceive a well-formed sentence when the actual utterance was incomplete or grammatically garbled, our ability to interpret words we have never heard before, to correct slips of the toungue, or to cope with new idiolects. These phenomena threaten standard descriptions of linguistic competence (including descriptions for which I am responsible). [1986:162]
So, Davidson thinks the received picture of language must be modified. Which one of the above principles (1)-(3) does he reject, then? Answer: number (3). According to Davidson, phenomena like malapropisms show that our ability to understand each other cannot be learned wholly in advance of the particular occasions of communication. There are no common rules which we first agree upon, and then apply to particular cases; there are no conventions which cover and determine in advance all meaningful uses of words. Davidson concludes:
We must give up the idea of a clearly defined shared structure which language-users acquire and then apply to cases. And we should try again to say how convention in any important sense is involved in language; or, as I think, we should give up the attempt to illuminate how we communicate by appeal to conventions. [1986:174]
Now, Davidson makes not only this critical point. He also sketches a new, alternative picture of communication and language use. In brief, this picture is as follows. To begin with, Davidson keeps the idea that speakers are, in some sense, equipped with systematic meaning-theories that enable them to produce and understand linguistic utterances. What he rejects is the notion that all speakers share one, static theory that they apply to particular cases. Instead, Davidson credits each speaker with a whole bunch of theories, none of which is shared with anyone else for more than a short period of time. Furthermore, these theories are not supposed to be static and unchanging; on the contrary, they are continually revised and adjusted, from one particular occasion to another. As language users, we constantly make up the rules as we go along, so to speak.
This is how Davidson describes what goes on in successful communication:
the interpreter comes to the occasion of utterance armed with a theory that tells him (or so he believes) what an arbitrary utterance of the speaker means. The speaker then says something with the intention that it will be interpreted in a certain way, and the expectation that it will be so interpreted. In fact this way is not provided for by the interpreter’s theory. But the speaker is nevertheless understood; the interpreter adjusts his theory so that it yields the speaker’s intended interpretation. [1986:166]
Now, Davidson complicates this basic story by introducing a distinction between ‘prior’ and ‘passing’ theories. The exact details of this distinction are never made entirely clear, but the main lines are as follows. A ‘prior’ theory is the theory that I, as a competent language user, supposedly bring with me to a particular conversation. This is not the theory I actually use during that particular conversation, however; remember, Davidson’s main critical point was that it cannot be decided beforehand how particular utterances are to be understood. Rather, starting from my ‘prior’ theory, I construct a ‘passing’ theory that is “geared to the occasion” [1986:168]; and this construction I perform on that very occasion. Passing theories are single-use, throwaway articles that are made and applied simultaneously. Indeed, as the passage quoted above suggests, the construction or modification of a passing theory might even take place after my hearing some particular utterance the interpretation of which is supposed to be accomplished by means of the theory. On Davidson’s model, if I hear an utterance that does not fit the passing theory I happen to be using, I simply modify the theory until it yields the correct interpretation.
Finally, a very important point. Of course, invoking a multiplicity of meaning-theories that are contantly revised does not, by itself, account for the innovative, non-mechanical features of communication. For, if the construction and revision of meaning-theories were themselves conceived as entirely regular procedures, Davidson’s alternative view would have no better chance of accounting for the genuine irregularities of language than the received picture he claims to reject. Consequently, a crucial part of Davidson’s alternative picture is his claim that the construction and modification of passing theories do not follow precise, mechanical rules:
there are no rules for arriving at passing theories, no rules in any strict sense, as opposed to rough maxims and methodological generalities. A passing theory really is like a theory at least in this, that is derived by wit, luck, and wisdom […]. There is no more chance of regularizing, or teaching this process than there is of regularizing or teaching the process of creating new theories to cope with new data in any field - for that is what this process involves. [1986:173f.]
3. Meaning-Theories and Scientific Models
For those mainstream linguists and philosophers of language who are deeply convinced that the principles (1)-(3) above capture what language is and must be, Davidson’s criticism probably seems both radical and provocative. On the other hand, to those of us who feel uncomfortable with all these principles, considered as philosophical theses about language and communication - i.e., those of us who do not believe that successful communication presupposes that speakers somehow share a systematic meaning-theory - to us, Davidson’s criticism seems not radical enough. Indeed, from our point of view, Davidson’s discussion seems more like a desperate and unwarranted attempt to save as much as possible of the received picture.
I have already hinted at what I take to be the basic problem of Davidson’s view. Let me now state this problem once more, this time fully explicitly. Davidson proposes to give a faithful account of the linguistic abilities of competent speakers. He also points out, rightly, that the concrete manifestation of these abilities - i.e., actual linguistic practice - is not systematic, but studded with quirks and irregularities. Since Davidson subscribes to the claim that linguistic meaning is essentially tied to public practice, one would expect him to draw the straightforward, Wittgensteinian conclusion that faithful descriptions of linguistic abilities can only have the form of piece-meal, non-systematic remarks, and that it is just false to represent speakers as being armed with systematic meaning-theories. However, this is not Davidson’s conclusion. Somewhat like those old astronomers who desperately tried to save the Ptolemaic system by increasing the number of epicycles, Davidson tries to compensate for the irregularities of actual linguistic practice by multiplying the number of systematic theories that the speakers are supposed to have. There are, to begin with, all the passing theories that the they make use of during their lives. But furthermore, they are allegedly armed with a great number of prior theories; for, as Davidson says, “an interpreter must be expected to have quite different prior theories for different speakers.” [1986:171] In other words, I have one complete prior theory for conversations with my wife; another complete theory for conversations with my father; a third one designed for my barber; and so on and so on. This is not all, however. To repeat, Davidson adds that all these theories are often revised and adjusted. And these revisions and adjustments are not supposed to be limited to some small part of a theory. Since Davidson is a radical holist, he rejects the idea of such local adjustments. According to his holist conception, when a word changes its meaning, or takes over the role of some other word or phrase, or when a new word is invented, this has repercussions on the theory as a whole. Consequently, the alleged revisions must be quite complicated.
In sum, Davidson thinks the irregularities of actual linguistic practice forces him to ascribe to each one of us an enourmous amount of complete, systematic theories of meaning that are frequently subject to fairly sophisticated revisions. Now, isn’t this an extremely far-fetched conclusion? Doesn’t it show just how very far Davidson is prepared to go in order to save whatever can be saved of the principles (1)-(3)? In other words, isn’t it a clear indication that he is still held very much captive by the very picture of language he claims to reject?
One may try to defend Davidson by arguing that these rhetorical questions are based on a failure to take into due consideration his scientific aims. “The theory of interpretation,” says Davidson, “is the business jointly of the linguist, psychologist and philosopher.” [1984:141f.] Presumably, the philosopher’s putative task in this co-operative project is to provide a basic framework for empirical research, while linguists and psychologists are supposed to carry out this inquiry in concrete detail. Now, the Wittgensteinian notion of piece-meal, non-systematic descriptions is obviously quite useless as a strategy for empirical research within psychology and linguistics. After all, the purpose of science is not to produce mere enumerations of particular data, but to go beyond these particulars by fitting them into more general and systematic patterns. Therefore, despite its extraordinariness, Davidson’s viewpoint might seem like a better alternative than Wittgenstein’s if fruitful science is the goal, since it does not abandon entirely the scientific virtues of generality and systematicity.
But this rejoinder fails. It is true, of course, that Wittgensteinian piecemeal description is not a strategy to be recommended in science. However, Davidson’s program hardly fares any better. As a matter of fact, I know of no linguist, psychologist or cognitive scientist who has produced any valuable empirical results by adopting the kind of viewpoint Davidson proposes. And this lack of empirical results seems to be no coincidence. Let me explain why.
Suppose to begin with, and for the sake of argument, that the old, received picture of language that Davidson rejects were actually true; that our linguistic practices, albeit highly complex, were entirely regular and static, and that the abilities of competent speakers were therefore possible to model by means of one, systematic meaning-theory. Now, it is clear that carrying out this modelling in actual detail would require an enourmous amount of empirical research. Teams of field-linguists would probably have to work for several years in order to gather data enough for the construction a complete theoretical model of the envisaged sort. Their task would not be Sisyphean, however; thanks to the presumed regularity and constancy of the practices under investigation, their patient empirical research would amount to a detailed mapping of actually present structures.
With respect to Davidson’s alternative picture, however, the issue of empirical verification becomes much more problematic. Indeed, it is difficult to see how the kind of modelling he envisages could even get off the ground. The basic trouble has to do with the fact that if empirical data are to be of any help in detailed theory-construction, a certain amount of stability at the theoretical level must be presupposed. The theory which the data are supposed to confirm or falsify cannot be assumed to be in constant flux, but must be held constant long enough to guarantee that a sufficiently rich data base can be collected. As was indicated by the previous, ‘static’ scenario, collecting data enough to construct just one theory of meaning would probably require several years of intense field research. But now, if meaning-theories are supposed to be very short-lived, then data will have to be collected much faster. Consequently, the available data base will inevitably be very sparse, and so its usefulness for theory-construction will be practically nil.
To repeat, the problem is to get off the ground. If we were allowed to assume previous knowledge about what theories a certain speaker is armed with at a certain time, it would perhaps not be impossible to account for later modifications and constructions. If the speaker makes or understands an utterance where a word is given some unexpected role, we would then figure out the implications of this new role for logical relations to other words and sentences, and make the corresponding revisions in our original model. But how do we find out about the theories that a speaker has without assuming such previous knowledge? To this question, Davidson provides no answer. And I cannot see how he could. Consequently, I think Davidson’s idea of how linguistic abilities are to be modelled, is just worthless as a starting-point for concrete scientific work.
Actually, if considered only as such a starting-point or framework, the old, received picture of language and communication is much better than Davidson’s alternative proposal - despite the fact that actual linguistic practice is as irregular as it actually is. Again, it is the normal case in science that theoretical models are only idealised and simplified representations, based on a motley of more or less deviant data. The mere fact that the received picture of language fails to make justice to the various quirks of usage which, to be sure, permeates ordinary speech, is therefore per se no objection against its being used as a framework for scientific inquiry. This old picture is at least possible to handle empirically, while Davidson’s alternative framework leaves no meaningful room at all for honest empirical inquiry.
If Davidson’s motive is to give a contribution to science, then his peculiar way of dealing with linguistic diversity is a failure. Indeed, the problems of empirical verification only strengthens the impression that Davidson’s picture of language and communication is a highly speculative armchair construction. So, he better come up with some other reasons for postulating all these meaning-theories. But what other thinkable reasons are there? This is the topic of the next section.
4. The Place of Meaning-Theories
What are Davidson’s reasons for maintaining that systematic meaning-theories are needed to describe the abilities of competent speakers? After all, he acknowledges and brings to the fore the fact that actual linguistic practice is a motley of more or less irregular and innovative uses of expressions, a motley that is at most loosely related to the sort of fully and precisely specified formal structures to which Tarskian truth-definitions apply. Furthermore, he emphasises that mastering such irregular practices is not just a matter of mechanical interpretation, but must involve non-systematisable ‘wit, luck and wisdom’. Now, why not go all the way, and conclude that meaning-theories have no substantial role at all to play in descriptions of what competent speakers are able to do?
In the literature, there are two standard arguments for the claim that systematic, recursive meaning-theories are needed to characterise the ability of competent speakers. One is that this ability is infinite, or open-ended: it is impossible to list, one by one, all the meaningful sentences that a competent speaker can produce and understand. The other reason is what is sometimes called ‘creativity’ or ‘productivity’: competent speakers can use and understand new sentences, i.e., sentences that they have never seen or heard before. Once the genuine irregularities of language use are acknowledged, however, the open-endedness and creativity of linguistic competence must be conceived in a way that undercuts these standard arguments for recursivity. Indeed, what can be learnt from malapropisms and other ‘quirks of usage’ is precisely that linguistic open-endedness and creativity are not systematically generated.
Consider the infinite sequence of ordinary arabic numerals. Its infinite character is regular: it is possible to state a recursive rule which unequivocally determines the whole sequence. In fact, this sequence is recursively constituted: the only possibility of giving the whole sequence is in terms of some recursive operation. By contrast, the open-endedness of natural languages is irregular. As Davidson emphasises again and again, there is no rule or set of rules that can generate the whole set of utterances that a competent speaker is able to understand, let alone rules that specify what meanings all these utterances have.
A similar distinction can be drawn with respect to novelty. A numeral, say 2987697, may be new to you, in the sense that you have never encountered it before. Still, this novelty is regular, in the sense that you recognize the numeral as generated by the recursive rule which you know beforehand and which determines the sequence of ordinary numerals. By contrast, the novelty of many utterances in natural languages is not regular in this sense. The truly innovative and unforeseeable uses of language are ‘truly innovative and unforeseeable’ precisely in the sense that they are not generated by some previously known and specified rule or set of rules.
Consequently, the two standard arguments for giving recursive meaning-theories a central role in the philosophy of language cannot consistently be used by the later Davidson. Still, recursive meaning-theories continue to play a central role also in his later philosophy of language. So, it seems as if some other argument must be lurking in the background. But what other arguments are available?
In his early writings, Davidson relied on both infinity and productivity to motivate his claim that any model of linguistic competence must be systematic. In an often quoted passage, he related infinity to the problem of learnability:
When we can regard the meaning of each sentence as a function of a finite number of features of the sentence, we have an insight not only into what there is to be learned; we also understand how an infinite ability can be encompassed by finite accomplishments. For suppose that a language lacks this feature: then no matter how many sentences a would-be speaker learns to produce and understand, there will remain others whose meaning are not given by the rules already given. [1984:8]
Of course, the later Davidson would never say anything like this. As I have repeatedly emphasised, the main point of ‘A Nice Derangement of Epitaphs’ is precisely that no pre-established system of syntactic and semantic rules can be complete: no matter how carefully we state the rules, there will always remain an unsurveyable amount of meaningful utterances that they fail to cover. In fact, the picture of language which is implicitly rejected in the last sentence in the passage just quoted, is precisely the one that the later Davidson embraces: there are no such things as natural languages with a fully and precisely specified structure.
Let us take a more careful look at this difference between the early and the later Davidson. The passage quoted above is immediately followed by some qualifications:
This argument depends, of course, on a number of empirical assumptions: for example, that we do not at some point suddenly acquire an ability to intuit the meanings of sentences on no rule at all; that each new item in the vocabulary, or new grammatical rule, takes some finite time to be learned; that man is mortal. [1984:8f.]
The two latter assumptions are somewhat peculiar. What would it mean to learn something infinitely fast? Even if man were immortal, what would it mean to say that at some point in his life he will have learned an infinite number of rules? However, what is of interest here is the first assumption: that we do not have the ability to ‘intuit the meanings of sentences on no rule a all’. Notice that Davidson describes this as an empirical assumption. So, here he regards the claim that understadning involves the employment of rules not only as a more or less fruitful façon de parler, but as a substantial hypothesis about how we actually go about understanding linguistic utterances. In his later writings, he rejects this conception. He declares that philosophers of language should eschew speculation “about arcane empirical matters that neither philosophers nor psychologists know much about,” and states explicitly that “I do not think we normally understand what others say […] by appealing to some theory of interpretation.” [1994:3] However, if one looks closer at what he says elsewhere, this rejection seems less definite. Consider the following passage:
You may notice that I do not speak of implicit knowledge here or elsewhere: the point is not that speaker and hearer has a theory, but that they speak and understand in accordance with a theory - a theory that is needed only when we want to describe their abilities and performance. [1994:5]
This is extremely confusing. What on earth is Davidson’s reason for maintaining that speakers and hearers speak and understand in accordance with a theory? After all, he admits that linguistic practice involves essentially irregular elements. So, if he also subscribes to the ideas that meaning is public, and that speaking and understanding are therefore activities the character of which must come out in concrete linguistic practice, then one would expect him to say that language users do not speak and understand in accordance with a theory. Why does he draw the opposite conclusion, then? The only answer I can find is that he is still held captive by the idea that underneath the ‘surface’ of actual language use, there take place empirically real processes of interpretation which involve the application of strict rules. If this is true, his denial that speaker and hearer has a theory of interpretation is idle talk.
Or, look at it this way. As we have seen, the later Davidson does not abandon the idea that systematic theories are required to account for the abilites of competent speakers. However, unlike his early self, he now thinks that each such theory is very short-lived and is often constructed by the interpreter at the time of utterance. Furthermore, he says this process of theory-construction is a creative act that does not follow strict rules, but requires ‘wit, luck and wisdom’. But now, once ‘wit, luck and wisdom’ are allowed to play an essential role in the model, the question naturally arises: why not go one step further, and skip all talk of theories? Why not simply classify understanding itself as an unregularisable kind of intuitive wisdom? The only reasonable answer seems to be that Davidson assumes, just like his earlier self, that it is an empirical fact about us that we lack the ability to ‘intuit’ the meanings of utterances on no rule at all, and that rules are therefore still needed, even if they are only temporary and invented in non-systematisable acts of theory-revision and theory-construction.
Of course, Davidson would probably respond that even if he sometimes expresses himself as if he made such an empirical assumption, this should rather be understood in methodological terms, i.e., as one part in a framework for scientific modelling of linguistic competence. However, as we saw in the previous section, it requires very little thought to realise that as such a framework, his viewpoint is quite useless. If Davidson were really serious about providing a framework for concrete scientific studies, then he would never have written ‘A Nice Derangement of Epitaphs’. Consequently, his basic motive must be something else. And, again, the only remaining alternative seems to be that he makes substantial claims about what actually goes on beneath and explains the hurly-burly of real-life communication.
5. Concluding Remarks
I have tried to uncover the tension in Davidson’s later philosophy between, on the one hand, his attempt to give an account of linguistic competence that is not entirely non-systematic, and, on the other hand, his claim that this account should not simplify, idealise or otherwise distort what linguistic competence actually involves. Davidson wants to find a theoretical, systematic description that makes full justice to language, in all its diversity. I have argued that this is impossible. In the previous section, I suggested that Davidson’s failure to perceive this impossibility is connected with another tension, namely, his vascillating between regarding meaning-theories as something that actual speakers somehow know and make use of, and treating such theories as mere models of linguistic competence. I will end this paper by making som further comments on this issue.
If one’s aim is only to construct a scientific model of linguistic competence, one should not worry about making simplifications and idealisations, given that they promote systematicity and perspicuity within one’s theory. Furthermore, one should not hesitate to use special, technical terminologies and systems of classification, as long as those terminologies and classifications suit one’s scientific interests. It is irrelevant that most speakers do not know these technical terminologies, as long as one’s aim is not to describe what communication involves from the perspective of the communicator herself, but rather to predict and control linguistic behaviour as that behaviour is conceived within one’s own scientific framework.
Now, Davidson’s approach to language have important features in common with such a scientific approach. As we have seen, he gives a prominent place to systematicity. Moreover, he does not hesitate about employing technical notions borrowed from, e.g., formal semantics and structural linguistics. Still, there are qualifications. He does not want to impose systematicity on language, by making bold simplifications and idealisations. On the contrary, his idea is that linguistic competence is somehow really systematic. Similarly, he has a tendency to deny the fact that the technical notions he uses are technical notions at all. In fact, he seems to think that these notions are common knowledge. Consider the following passage:
The concept of a language is of a sort with, and depends on, concepts like name, predicate, sentence, reference, meaning, and truth. These are all theoretical concepts. We do not need them in order to use or learn a language; obviously they are not available to us when we are learning a first language. Where we want these concepts is in talking about speech behavior. Philosophers, psychologists, and linguists need these theoretical terms if they want to describe, theorize about, and explain verbal activities. The rest of us also have occasion to talk about talk, or write about writing, so these theoretical concepts have their place in the loose informal “theories” we all have about language. Indeed, we all talk so freely about language or languages that we tend to forget that there are no such things in the world; there are only speakers and their various written and acoustical products. [Davidson 1992:256; italics added.]
This is simply false. The concepts ‘name’, ‘predicate’, ‘sentence’, ‘reference’ and so on that Davidson mentions do not figure at all in everyday talk about language. On the contrary, they belong to quite specific scientific contexts, most of which have been invented and developed during the last 150 years. Learning these concepts is something people do only if they happen to study linguistics, logic or philosophy at some university. And such learning requires long and special training. Of course, the words ‘name’, ‘sentence’, and so forth, occur in everyday contexts; but there, their functions are quite different from the role they have within, say, formal semantics or structural linguistics.
Similarly, that ‘there are only speakers and their various written and acoustical products’ is not an obvious fact that we all tend to forget. Rather, the perspective from which linguistic utterances are viewed merely as ‘acoustical products’ and their meanings are considered as theoretical constructions, is at best a highly artificial perspective which most people have never learned to adopt, and which would seem extremely far-fetched in almost any everyday context. #4 It is indeed a perspective that has to be learned; it does not belong to what everyone knows and must realise to be true.
Davidson’s tendency to repress these quite obvious facts, is yet another manifestation of the tensions I have described in this paper. If his only aim were to give a theoretical model of linguistic competence, then why would he insist that the technical notions he uses belong to everyday talk about language? Passages like the one I just qouted indicates that he wants something more than a theoretical model. He wants a description that makes justice to language, not just as a subject for technical studies within philosophy or linguistics or cognitive science, but as something that matters to ordinary speakers in everyday cases of communication. While Quine and Wittgenstein are both aware that you have to choose between giving a scientifically acceptable account of language and linguistic competence, and giving a description that is faithful to language as it matters to engaged participants in real-life discourse, Davidson wishes to have it both ways. I hope to have shown that this wish of Davidson’s is impossible to fulfil.
References
Davidson, Donald 1984. Inquiries into Truth and Interpretation, Oxford University Press: Oxford.
--- 1985. ‘Reply to Quine on Events’, in LePore E. and McLaughlin B. (eds), Actions and Events: Perspectives on the Philosophy of Donald Davidson, Blackwell: Oxford, pp. 172- 176.
--- 1986. ‘A Nice Derangement of Epitaphs’, in Grandy R. and Warner R. (eds), Philosophical Grounds of Rationality, Oxford University Press: Oxford, pp. 156-174.
--- 1992. ‘The Second Person’, in French P., Uehling T.E. and Wettstein H. (eds), Midwest Studies in Philosophy, vol. XVII, University of Notre Dame Press: Notre Dame, pp. 255- 267.
--- 1994. ‘The Social Aspect of Language’, in McGuiness B. and Olivieri G. (eds), The Philosophy of Michael Dummett, Kluwer: Dordrecht, pp. 1-16.
Hacker, P.M.S. 1996. ‘Wittgenstein and Quine. Proximity at Great Distance’, in Arrington R. and Glock H-J. (eds), Wittgenstein and Quine, Routledge: London, pp. 1-38.
Heidegger, Martin 1927. Sein und Zeit, Niemeyer: Tübingen.
Glock, Hans-Georg 1994. ‘Wittgenstein vs. Quine on Logical Necessity’, in Teghrarian S. (ed), Wittgenstein and Contemporary Philosophy, Bristol.
Quine, W.V.O. 1960. Word and Object, MIT Press: Cambridge, Mass.
--- 1981. Theories and Things, Harvard University Press: Cambridge, Mass.
--- 1987. Quiddities, Harvard University Press: Cambridge, Mass.
Rorty, Richard 1993. ‘Wittgenstein, Heidegger, and the Reification of Language’, in Guignin C. (ed), The Cambridge Companion to Heidegger, Cambridge University Press: Cambridge, pp. 337-357.
Wittgenstein, Ludwig 1961. Tractatus Logico-Philosophicus, Routledge & Kegan Paul: London.
--- 1967. Zettel, Anscombe G.E.M and von Wright G.H. (eds), Basil Blackwell: Oxford.
--- 1974. Philosophical Investigations, Basil Blackwell: Oxford.