Does analytic philosophy’s naturalism, combined with its conception of language as in some sense rule-governed lead to the paradoxical thesis that, roughly, we don’t know what we mean? Briefly, does what Livingston calls the “structuralist” project, combined with the idea that a naturalistic, external approach to language is the correct philosophical method, lead to a paradoxical failure to match language as we live in it? Paul Livingston argues in “The Breath of Sense: Structure and the Paradox of Origin,” as well as in his book, Philosophy and the Vision of Language, [1] that it does.

There is much to admire in this essay and in Livingston’s book. The book, especially, takes an unusually wide perspective on the history of philosophy in the last century, using a very wide knowledge of analytic philosophy, but writing in the light of a similarly broad knowledge of Continental philosophy. Livingston gives a persuasive account of many analytic thinkers, and by and large accurately describes and illuminates the trains of thought that have led analytic philosophy down the paths it has taken. I will focus on the essay included in this journal, while occasionally mentioning topics and arguments that occur in the book.

For all the virtues of this essay, there are questions to raise about some of Livingston’s central contentions. I question whether Quine’s argument is in fact a consequence of naturalism and structuralism, whether contemporary naturalism is committed to anything like Quine’s thesis, whether analytic philosophy has abandoned its previous insistence on the centrality of language to philosophy because of Quine’s aporia, and whether the aporia, in some form, is a paradox or rather a discovery.

Quine’s “Ontological Relativity”[2] was the culmination of his on-going critique of Carnap, as Livingston observes. In talking of “critique” we should remember that Quine agrees with Carnap on some central points. Carnap and Quine are both committed to denying that there are necessities that rest on the natures of things. That is, both are anti-essentialists. Quine’s critique of the analytic-synthetic distinction in “Two Dogmas of Empiricism”[3] can be viewed as a kind of deconstruction of Carnap’s claim to renounce essentialist metaphysics and to be thoroughly empiricist. Part of the general positivist project was to treat all necessities as “mere” linguistic matters, i.e., as resting on meanings. Quine’s point is thus that Carnap is still committed to assigning essences to meanings. To be a non-essentialist, post-metaphysician, then, one must be an empiricist about meanings. Quine’s empiricist substitutes for meanings, clearly, are stimulus meanings.

“On What There Is”[4] is another part of Quine’s anti-essentialist re-focusing of philosophy. The underlying thesis, developed more clearly in Word and Object,[5] is that, rather than thinking of beliefs as attempts to correctly represent beings, we should think of beings as organizational devices, posits. Being, as it were, supervenes on truth, rather than vice-versa. This is a view that Carnap endorses in “Empiricism, Semantics, and Ontology,”[6] which, while it comes after Quine’s article, continues in the spirit of the logical pluralism of Carnap’s Logical Syntax of Language.[7]

Word and Object and “Ontological Relativity” develop the apparent consequences of denying meanings. If we take as our fundamental empirical basis the stimulus inputs to an organism, thus apparently eschewing meanings as entities that have natures in themselves and therefore construct meanings out of what is given in empirical experience, we find that there is no clear sense in which we can tell the difference, even from the inside, between talking about undetached rabbit-parts, talking about instances of rabbithood, and talking about rabbits. Objectively, that is, on the basis of every possible experience, choice of ontology is arbitrary and relative to the interpreting background language.

We should note that Quine retains some of the suppositions of Carnap’s “Empiricism, Semantics and Ontology” in “Ontological Relativity.” In particular, he retains the notion that ontology is a structure of posits that organizes a given empirical domain. That is, he retains the distinction between observations and the organization of observations, the given that allows variation in the “ontology,” the non-given structuring concepts.    

As Livingston observes, Quine’s conclusion in “Ontological Relativity” was reasonably taken to be a paradoxical conclusion. Surely something is missing in the empiricist account if it implies that we cannot tell whether we are talking about rabbits or undetached rabbit parts. For Livingston, Quine’s result shows that something in the practice of using language is not captured by a structuralist account of language that takes an outside, “what are the data” point of view. The naturalist, empiricist approach to language, that takes a structuralist approach to language misses something every language-user experiences.[8] Livingston’s project is to diagnose what has gone wrong with naturalism’s “structural”[9] account of language and to suggest lines of correction that stem from Heidegger’s thought.

One reservation about the above account is whether in fact Quine’s paradoxical result actually follows from a structuralist naturalism that eschews essentialism. One reason to think perhaps not is that Donald Davidson, who owes a great intellectual debt to Quine, modifies the Quinean position on anti-essentialist grounds and reaches a much more innocuous conclusion. Davidson extends and purifies Quine’s anti-essentialism while maintaining a completely “structuralist” account of what it is to have a theory of meaning. Davidson’s critique of “Ontological Relativity” occurs among other places in “On the Very Idea of a Conceptual Scheme.”[10] There Davidson argues that even Quine succumbs to the metaphysical temptation to take beings as prior to truths. The “ontological relativity” Quine espouses presupposes a given domain that can be sorted in various ways. But such a domain, in this case the domain of stimulations, is a domain of beings taken to be an unproblematic given. Quine’s “indeterminacy of translation” starts with an empirical given, stimulations, and, relative to that given, shows that incompatible alternative ontologies are equally grounded. Davidson realizes that this abandons the idea that objects are posits and re-introduces essentialism about the domain sorted by various predicate-schemes.

Davidson thus develops the consequences of being a non-essentialist naturalist. With Wittgenstein, and following and modifying Quine’s Word and Object, Chapter 2, he takes agreement on what is  the case to be a fundamental starting point for communication and therefore for language. Briefly, Quine’s argument in Chapter 2 of Word and Object applied Wilson’s “Principle of Charity”[11] only to “hard data,” the given sensory inputs that were the foundation of his empiricist theory. Davidson, in abandoning the given, de-privileges agreement on “sense-inputs.” The Principle of Charity thus becomes for Davidson the interpretive principle, “maximize agreement generally.” Davidson, having done much of his early work in game theory, of course takes “maximize agreement” to apply to desires as well as beliefs.[12]

From Davidson’s perspective then, generalized ontological relativity/indeterminacy does not obtain. If we are interpreting another by seeking agreement, then agreement on whether there are rabbits is to be sought. Thus, Quine’s application of the principle of charity as introduced in Chapter 2 of Word and Object is generalized so that general ontological indeterminacy is no longer a consequence of the “externalist” conception of how to understand meaning common to Davidson and Quine.[13]  

Another significant difference from Quine’s account of radical translation is Davidson’s never-abandoned view that the proper form of a theory of meaning is a truth-definition, a recursive account of the truth-conditions of any sentence in a language. A truth-definition in effect uses a sentence of the interpreter’s language to express the meaning of the interpretee’s language. A theory of meaning for a given speaker is a theory that correctly generates Tarskian biconditionals. The evidence for the correctness of such a theory is completely empirical—roughly the matching of respective webs of belief and desire. The meaning of a sentence, though, is just the properly-generated biconditional. So, what is the meaning of “Joe is a frog?” A complete answer is “Joe is a frog” is true if and only if Joe is a frog. This is a “structuralist” conception of the semantics of a language.

As Davidson’s thinking on language matures, he realizes what linguists have long recognized, that “languages,” as practices fixed by determinate sets of rules obeyed by groups of people who speak the same language, are a myth. Any two people speak different languages in the “strict” sense—each of us has at least slightly different vocabulary and different syntax. “English” is a label for a vague group of dialects each of which is a vague group of idiolects.  Davidson does indeed give up on language as a structure of rules that govern usage among a group and over time, but he retains the idea that at any moment, a person’s idiolect is to be characterized by a truth-definition.

“The philosophers’ notion of language”[14] is that of a body of rules shared by members of a group, a body that imposes norms on their linguistic behavior.  Davidson rightly regards this notion as unnecessary for the explanation of communication, and problematic in making, for instance, Finnegans Wake not linguistic.  Davidson makes interpretation the central linguistic skill.

Davidson does give up on the conventionalist conception of language. I do not see, contra Livingston, that he gives up on the conception of language as capturable by a truth-definition based on publicly available information. It is true that Davidson has nothing like “rules of use” that helpfully describe the application of any predicate terms or illuminate the inner structure of concepts. The “rules” of the application of the term “game” to a practice, for instance, consist of the advice, “Apply ‘is a game’ to a practice if and only if it is a game.” Whatever would go into the “analysis of concepts,” that is, figuring out what applications by an interpretee suffice to justify interpreting a term as “game,” are evidence for the interpretation, not the meaning.

Davidson’s conception of language-interpretation treats it as a special case of action-interpretation.  A person says or writes something for a purpose. That purpose requires taking into account how the other person will interpret the act. Thus Davidson treats metaphor and other figures rhetorically—a person says one thing, intending that the audience understand something else.  Sentences with truth-conditions are equipment to be used for many purposes other than asserting that what is the case is the case.

The notion of purposive speech-actions also explains the sense in which Davidson continues to think of language-use as having something to do with norms. The norms in question are not “obligations” in the sense in which Brandom,[15] for instance, treats linguistic norms. No one is necessarily legitimately upset if a speaker violates a linguistic norm, in Davidson’s sense.[16]   Rather, the norms are practical guidelines for successful communication. You use words in familiar ways in order that your audience will interpret you as you wish to be interpreted. If you are explaining something to undergraduates, it is counter-productive to use “decimate” in its etymological sense. It is not that “decimate” as “reduce by a tenth” is incorrect English or violates a rule, rather it is a poor way of communicating this to this audience. Such practical guidelines, Davidson holds, suffice to yield the wide-spread general agreement in speech-patterns and word-meanings one finds in groups of language-users.

Through all of these adaptations, continuations, and purifications of Quine’s projects, Davidson continues to be a naturalist, an externalist about meaning, and to hold that there is widespread, albeit not ubiquitous, indeterminacy of radical interpretation. That is, there are occasions when what we mean is indeterminate even to ourselves. Davidson surely remains a structuralist, in Livingston’s terms, in virtue of taking the form of a theory of meaning to be a recursive truth-definition.

Davidsonian indeterminacy is often practically innocuous. Here is a simple example of semantic indeterminacy, not one of Davidson’s own:[17]   When I say “The Bronco’s crushed the Patriots,” there are two ways of understanding the predicate “crushed,” in the sense that there are two non-equivalent truth-definitions for our idiolect. On one interpretation, “crush” on this occasion means something like “thoroughly defeated.” On this interpretation, “crush” has two predicate clauses, etymologically related, but distinct. On the other interpretation, I am knowingly using a sentence that is false in order to communicate something about the game’s outcome. Why is this indeterminate rather than clearly metaphorical? Since the metaphor is so tired, my intention cannot be to illuminate the event. This utterance is to all intents and purposes just a synonym for “thoroughly defeat,” you might say. That is, pragmatically, these interpretations come to the same thing.[18]

Is this indeterminacy Davidson retains a feature of his retaining analytic philosophy’s “structuralist” naturalist tradition? It is hard for me to see that that is so. The core idea that generates indeterminacy is the denial of essentialism, which entails the denial of idiolect-transcendent meanings. But you don’t have to be a naturalist or an analytic philosopher to have reasons to deny essentialism, and thus transcendent meanings. And thinkers outside the analytic tradition who do deny essentialism have been led to similar conclusions about indeterminacy (or undecidability). Derrida, in particular, seems to have reached similar conclusions about indeterminacy as a consequence of a similar view about language, that there is no notation that avoids the arbitrariness of the signifier. Derrida expresses this idea, among other ways, by arguing that writing is prior to speech. I take this to mean that writing does not conceal the fact that there is no natural connection between a word and what it means, whereas speech, especially the inner speech of thought, does conceal that.

The indeterminacy thesis, either in the version of Davidson or that of Derrida, is a consequence of a thorough-going anti-essentialism, the thorough-going realization that, as Davidson said, “the beginning of wisdom is the realization that all sameness is relative to a predicate.”[19]   On that perspective, language is all-important, since there is no domain of beings to be sorted into groups. On this conception, the method of “metaphysics”[20] is to work out the ontological commitments of the truths we know by getting at the logical forms of kinds of sentences. Even if we each speak a different idiolect, so that there is no “language in the philosopher’s sense,” there is guaranteed to be enough agreement on what is true that investigations of logical form can tell us that, for instance, there are events. The argument is something like this: There are an infinity of inferences we are in a position to correctly make; such inferences are only possible given that we have an ontology of events, so in fact events exist. The fact that there is no “language in the philosophers’ sense” does not mean that “structuralism,” in Davidson’s sense of there being a passing theory in the form of a truth-definition, is not still appropriate.

In my view, as a participant in the sea change, what has led very many analytic philosophers to cease to think of language as the key to all philosophical knowledge was not coming to believe that Quine had shown the inadequacy of the naturalist-structuralist project, but rather a startling change in widely-accepted semantics. Kripke’s lectures in 1970, which became Naming and Necessity, made essentialism respectable. In fact, Kripke’s metaphysics was close to that of Aristotle in its central opinions. It became a mainstream view[21] that nature is objectively divided into natural kinds and that predicate-extensions were right or wrong depending on how nature in fact is. Even though the intuitive considerations about how the reference of names and general terms worked involved an essentially linguistic investigation, the consequence was that the previous fifty or so years of eschewing metaphysics ceased. Once essentialism became the norm, metaphysics could return to pre-Kantian questions and methods. So, analytic metaphysicians using essentially intuition and a priori arguments could raise and answer questions such as what really are the kinds of things in the world, how many beings are there, and is the past real? Theories of the basic make-up of the world are advanced—that the world consists of hunks of matter,[22] that the only genuine entities other than micro-particles are organisms,[23] and the like. In short, pseudo-problems of philosophy are now being seriously asked once again, as Carnap, Neurath, and Schlick, not to mention Quine and Davidson, spin in their graves.

The style of argument and writing remained that of the analytic philosophy tradition: Carefully laid-out sequences of argument, no premise left unturned, and the like. But the Quinean assumptions about the proper way to use linguistic considerations in doing philosophy were dropped by a very large portion of American analytic philosophers. Of course, language was and continues to be of great interest, and the project of giving truth-conditional semantics for natural languages is an extremely active research
project.[24]

In my opinion, philosophers were swept into essentialism by intuitive considerations that have counter-considerations. For instance, the argument that the lectern could not have been made of ice ignores the fact that the very stuff of the lectern, the constituent quarks and electrons, could have been configured in a way that made the lectern mostly water. In brief, as I argue elsewhere, all of the examples of de re necessities Kripke gives are not necessities about the entities themselves. So, I think that becoming essentialists was overly-hasty, and that in fact Davidson’s continuation of the Quinean line of thought is on the right track.[25]

Finally, does the aporia that does in fact arise with a thorough-going anti-essentialism call for a different approach to language? Might it perhaps rather be a discovery about language, that the feeling that when we engage in thought and inner speech, we know something about what we mean that is not available to the outside is an illusion?

To my mind, the search for Being is a mistake, not because Western Philosophy has been deluded by the idea of a subject confronting an array of objects, but rather because both Being and “the beings” are posits, in Quine’s sense—something we, as finite users of an infinite language, have to posit. If we did not have singular terms and predicates, and thus think and speak of the world as a world of beings and their features, we language-users would not be able to say and think the infinity of things we do in fact say and think. The fact that any way of positing beings leaves gaps and indeterminacies is just the way it goes—not a defect, but part of the very possibility of speaking and thinking. To expect something else is to lapse into the kind of nostalgia Nietzsche famously falls into in “On Truth and Lies in the Extra-Moral Sense”:

What then is truth? A mobile army of metaphors, metonyms, and anthropomorphisms--in short, a sum of human relations, which have been enhanced, transposed, and embellished poetically and rhetorically, and which after long use seem firm, canonical, and obligatory to a people: truths are illusions about which one has forgotten that this is what they are; metaphors which are worn out and without sensuous power; coins which have lost their pictures and now matter only as metal, no longer as coins.[26] [26]

Nietzsche finds language lacking in something—something that could be characterized as “real connection with Being.” If, as Davidson would have it, ontology is something that we impose on what is the case in order to think in a language with a finite vocabulary, then this lack is not a genuine defect at all.   

The importance of language to human existence does not get slighted on this account. Davidson’s central idea is an account of the intimate connection between language and human existence as human, his version of Heidegger’s Dasein. For Davidson, language is action that is made possible only by social relations. Only in a communication-situation, that is, a second person taken to have a view of the world, can there arise the difference between how things are for me and what is the case, that is, the notion of
truth.[27] For exactly similar reasons to Wittgenstein’s, Davidson denies the possibility of an absolutely private language.[28] Given that the very notions of belief and desire require the notion of truth, and that truth is a predicate of sentences, being a rational agent at all requires possessing a language and recognizing other agents.[29] Thus for Davidson, the very possibility of belief, desire, and intentional action, the whole realm of “mental events” rests on language. The fundamental notion of the intentional, the mental, and thus the human, is truth, the linking notion that connects beliefs, desire, action, reason-giving, and reason-having—briefly the whole domain of understanding events as actions and understanding organisms as persons. The structure of Truth is the structure of Dasein, not the structure of Being.[30]