Word and Object by Willard Van Orman Quine

First published: 1960

Type of Philosophy: Epistemology, logic, philosophy of language

Context

Word and Object is W. V. O. Quine’s magnum opus, the most complete expression of his views in a single place. It was written when he was at the height of his philosophical powers, roughly between 1955 and 1959, and continues the themes of his earlier articles in From a Logical Point of View (1953), of which the two most famous are “On What There Is,” which discusses criteria for ontological commitment, and “Two Dogmas of Empiricism.” The first dogma is that there is a clear distinction between analytic sentences (which are true by virtue of their meanings) and synthetic truths (which are made true by facts). The second dogma is that each meaningful sentence is reducible to an equivalent sentence, all the terms of which refer to immediate experience.

As these two articles make clear, Quine is equally interested in the problems of ontology and language, problems that he thinks are intertwined. Quine elaborated and refined the views of Word and Object in Ontological Relativity and Other Essays (1969). In the title essay of the latter book, which constitutes the first of the John Dewey lectures given at Columbia University in 1968, Quine admits his debt to Dewey. In Word and Object and other essays, he expresses his debt to Charles S. Peirce and his commitment to a kind of pragmatism. By his own admission, then, Quine is in the mainstream of traditional American philosophy.

Words and Language

Word and Object consists of three projects: One concerns words, one concerns objects, and one concerns the conjunction of words and objects. The first project is an attempt to give empirical foundations to language, to explain the human use of language in terms of human behavior and the perceptual environment. Quine restricts the theoretical terms of the explanation to these two because, he claims, they are the only available resources for the evidence upon which human beings learn language; thus Quine is very much concerned with reconstructing how a person—typically, but not invariably, a child—might come to learn a language. The second project concerns the classic problem of metaphysics: What kinds of objects are there? What really exists? Quine’s short answer to these questions is that there are two kinds of objects that really exist: physical objects and sets or classes of objects. These first two projects come together in his discussion of the kind of language that is appropriate for expressing what there is. According to Quine, it is science that says what there is, and the language for science, what he calls “a canonical notation,” is first-order predicate calculus with identity.

Quine’s most famous or infamous thesis about language is what he calls the “indeterminacy of translation.” The thesis is this: Two systems of translating one language into another can be devised such that each system is compatible with all the speech dispositions of those who know the language, yet the two systems are not equivalent. Quine develops his thesis in the course of describing the situation with which a linguist would be confronted when first coming upon a culture wholly alien to his or her own. How can the linguist correlate sentences of his or her own language with sentences of the native speaker? That is, how can the linguist come to translate between his or her own language and that of the native? This is the problem of radical translation.

Radical Translation

Suppose a rabbit hops by and the native says, “Gavagai.” The linguist might plausibly guess that the utterance means, “There’s a rabbit,” or “Look at that rabbit.” Of course the linguist might be wrong; in order to determine that, the linguist has to test his or her guess or hypothesis by interrogating the native in some way. However, how can one do this? One way is to say “Gavagai” the next time a rabbit appears and observe the reaction of the native. The linguist wants to see whether the native will assent to or dissent from the utterance. Assuming that a linguist can ask a native whether a given sentence is appropriate, Quine defines “affirmative stimulus meaning” as the class of stimulations that would prompt assent, “negative stimulus meaning” as the class that would prompt dissent, and “stimulus meaning” as the ordered pair of the two. Further, two utterances are stimulus-synonymous just in case they have the same stimulus meaning; that is, when they would produce assent or dissent in the same situations. Although the notion of stimulus meaning is well defined, the linguist is still faced with a cluster of problems, to which Quine is attentive. Which utterances of the native are to count as assent and which as dissent? Given that “evok” and “yok” are the utterances expressing each, which is which? Another problem is that a native will not always be willing or able to respond to the query. His or her glimpse of the object may not have been long enough to allow a response. Therefore, in addition to the assents and dissents there will be some lack of response. The native will sometimes make mistakes; perhaps he or she was looking in the wrong direction or attending to the wrong object. Alternatively, the native might lie. Because of all of these possibilities for skewed results, stimulus synonymy is not what is ordinarily meant by “synonymy.”

This partial catalog of the linguist’s problem is not meant to imply that the linguist’s task is impossible. The point is rather to indicate what difficulties one faces in learning a wholly alien language, what resources are available to learn it, and the strategy the linguist will employ in matching utterances with behavior. Given enough data, time, and imagination, the linguist will surely succeed in writing a manual of translation.

Quine helps us understand how the linguist will proceed with his or her job of translation beyond those utterances whose use is most closely tied to observation by explaining how the linguist might move from translating observation sentences like “Gavagai” to truth-functional sentences. The linguist comes to translate a linguistic element as expressing negation when and only when adding it to a short sentence causes a native speaker to dissent from a sentence previously assented to; the linguist comes to translate a linguistic element as conjunction when and only when it produces compounds from short component sentences that the native is disposed to assent to when and only when he or she is also disposed to assent to the components separately. The qualification “short” is added to guard against the native’s becoming confused by a sentence of extreme length. Also, it applies only to the language-learning situation; once the terms are learned, there is no restriction on the length of the sentences to which the terms are applied.

After the observation sentences, the truth-functional ones, and some other related sorts are translated, how does the linguist proceed? Roughly, he or she divides the sentences he or she hears into those segments that are often repeated; these are counted as the words of the language. The linguist’s task is then to correlate these words with words of his or her own language in such a way that the correlation conforms to the translation of the earlier sentences. Quine calls these correlations “analytical hypotheses.” A further constraint on analytical hypotheses is that stimulus-analytic ones, those sentences that receive unanimous assent among the natives, should, if possible, be correlated with sentences that are stimulus-analytic for members of the linguist’s own speech community; mutatis mutandis for stimulus-contradictory sentences. The parenthetical “if possible” is an escape clause. It is not always possible, without sacrificing the simplicity of the analytical hypothesis, to match a stimulus-analytic sentence of the natives with one of the linguist’s community. It may be necessary, in the interests of simplicity, to translate a stimulus-analytic sentence of the natives as “All rabbits are people reincarnate.” Such translations, however, are a last resort. By the principle of charity, one should always avoid attributing absurd or bizarre beliefs to foreigners.

There is another problem, or rather another result, of the thought experiment involving radical translation. The most fundamental relation between language and the world, the relation of reference, is infected by a kind of indeterminacy, which Quine calls “the inscrutability of reference.” Suppose that one linguist has determined that “Gavagai,” whatever its other uses, translates “rabbit” when it is used as a term. It remains a possibility that a second linguist, acting on the very same evidence as the first, will determine that “Gavagai” translates “rabbit stage” and a third that it translates “undetached rabbit part.” Each of these preferred translations is consonant with all the empirical evidence, yet the references of “rabbit,” “rabbit stage,” and “undetached rabbit part” are different. In other words, there is no one correct answer to the question, “What does Gavagai’ refer to?” Reference is inscrutable.

The situation of radical translation implies that the translator has a language and attempts to correlate the sentences of his or her own language with the sentences of a foreign language. In this regard, the problem of radical translation is different from the situation that infants are in when they begin to acquire language. However, there is an important respect in which infants are in the very same situation: They have the very same resources available to them as the linguist does. Like the linguist, infants must learn their language on the basis of perceptual stimulation and human behavior and must construct and test hypotheses about what an utterance means just as the linguist, but self-consciousness is not essential to the learning process.

The babbling of human beings during the end of their first year of life becomes transmuted into an incipient language by selective, positive reinforcement. Among the randomly produced verbal sounds of the infant will be “mama” and “papa,” which for the infant have no significance. They acquire significance when its mother and father reward the infant for producing those vocal sounds. Like a chicken learning to pull a lever for a pellet of food, a child first acquires language. The comparison of a child with a chicken is neither facetious nor unfair. Quine’s model for the first steps of language acquisition is a stimulus-response model, and he approvingly refers to the work of his Harvard colleague B. F. Skinner in this regard. In an oblique response to the criticisms of Noam Chomsky, who trenchantly criticized Skinner’s work, Quine concedes that, in addition to the stimulus-response mechanism, such innate forces as the natural tendency for an infant to smack its lips in anticipation of nursing and thereby to utter “mama” and a “basic predilection for conformity” play some role in a total causal account of infant language acquisition.

Among the most important things that the infant needs to learn are the distinctions among various types of terms. Quine distinguishes between singular terms, such as “Cicero” and “the orator who denounced Catiline,” and general terms, such as “orator” and “apple.” General terms, unlike singular terms, divide their reference among a number of objects. Definite and indefinite articles—”the” and “a(n)” respectively—and the plural ending are devices for the use of general terms in English. A person does not know how to use a general term in English if he or she does not know how to use such expressions as “an apple,” “the apple,” and “apples.” Mass terms, such as “gold” and “water,” are a kind of middle case, a kind of grammatical hermaphrodite. Syntactically, they are like singular terms in resisting indefinite articles and plural endings; semantically, they are like singular terms in not dividing their reference. However, they are like general terms in not naming one thing. The double role of mass terms extends to predication. In the subject position they are like singular terms (“Water is wet”); in the predicate position, they are like general terms (“That puddle is water”).

Physical Objects and Abstract Objects

Quine’s second project is to answer the question, “What is there?” His answer that there are physical objects (that is, four-dimensional spatiotemporal entities) and abstract objects (sets or classes of objects) is perhaps less interesting than his answer to several related questions, such as “What isn’t there?” or, to put the question more perspicuously, “Why does Quine refuse to countenance various sorts of purported entities?” On the ground of economy, Quine does not accept sense data; they are not needed for science. Physical objects cannot be eliminated from science, and they do all the work that sense data do. Sense data are not needed even to account for reports of illusions and uncertainty. Quine accounts for them with the phrase “seems that” prefixed to a sentential clause about physical objects, and he then paraphrases them away in the same way he paraphrases away propositional attitudes toward sentences. Sense data are excess baggage.

Quine’s rejection of sense data brings his standards for adjudicating conflicting claims for thinghood into high relief. Something has a claim to being an entity if it is empirically attested to or is theoretically useful. Competing claims to thinghood have to be weighed against both considerations. Sense data have empirical support but no theoretical use. Physical objects have at least some empirical support and a great deal of theoretical utility. Even if physical objects are not completely observable, or not “all there,” positing the unobservable parts involves more conceptual continuity than inventing an abstract entity. Theoretical utility also recommends classes or objects for thinghood. Classes account for numbers and numbers for mathematics. Hence, there are classes.

There are no properties or attributes because, in contrast with classes, they do not have clear identity. Classes are identical just in case they have the same members. There is nothing similar to be said for properties. The same set of objects might have two different properties; all and only creatures with hearts are creatures with kidneys, but the property of having a heart is different from the property of having a kidney.

Also, there are no facts. Like properties, facts do not have well-defined identity conditions. There is no answer to the question, “Is the pulling of the trigger the same entity as the killing of the man?” Facts are objectionable on other grounds. “Fact” is a stylistic crutch; it helps support the word “that” in some grammatical constructions, such as “The fact that he left is no excuse,” and the phrase “that fact” is a kind of standard abbreviation for a previously expressed assertion. As such, however, facts can be eliminated or altogether avoided by simple paraphrase.

Another question is, “Are objects given?” Quine says, “No.” They are posits. To call something a “posit” is not, for Quine, to be derogatory. Although some posits are bad—theoretically unjustified—some posits are really real; we posit entities of certain sorts in order to explain phenomena. For Quine, our beliefs are replete with posits. If a posit fails to explain a phenomenon or if another posit explains better, then its justification fails. However, the best explanatory posits are justified and have the status of being real. In short, for Quine, two sorts of objects have this status: physical objects for natural sciences, and sets or classes for mathematics.

Conjunction of Words and Objects

Quine’s third project is to explain how a person’s ontological commitments can be clearly expressed in language. His explanation is that what is required is a canonical notation that is clear, precise, and unambiguous. Such a notation is the first-order predicate calculus with identity. A canonical notation has two purposes. The first is that it allows for simplification of theory. It allows a person to iterate a few constructions a large number of times to the same effect as the use of a larger number of constructions a small number of times. The use of a larger number of constructions may allow for psychologically simpler constructions but not a theoretically simpler one, and that is what is demanded. The second purpose of a canonical notation is clarity. There are no ambiguities and no hedged entities in a canonical notation. Everything that is meant is up front.

Quine’s notion of philosophical explication is an important one in itself and important historically in contrast with some traditional notions of analysis. A philosophical explication does not purport to uncover or bring to light the hidden or implicit ideas of the people who use the problematic notion, and it does not purport to be synonymous with the problematic notion. In one stroke, Quine cuts the Gordian knot of G. E. Moore’s paradox of analysis. Philosophical explication is informative because it replaces the problematic notion with unproblematic notions that serve the same purpose. The notions of the explication may well be unfamiliar to and difficult for the ordinary user, but that is irrelevant. Familiarity should not be confused with intelligibility. Philosophical explication requires philosophically acceptable notions, not familiar ones.

This view of philosophical explication introduces a certain latitude into the standard of correctness. A correct explication may not be a unique one; several nonequivalent explications may be equally acceptable, no one of which is more or less correct than the others, so long as each explication meets scientific standards and serves the original purpose. For example, it is indifferent whether one accepts Gottlob Frege’s, John von Neumann’s, Ernst Zermelo’s, Richard Dedekind’s, or someone else’s definition of number, so long as the chosen one does the job. In short, explication is elimination: Out goes the bad air of familiar but unacceptable notions; in comes the good air of intelligibility. Quine thinks that his view of philosophical explication is in line with Ludwig Wittgenstein’s doctrine that the goal of philosophy is to dissolve a problem by showing that, contrary to appearances, there really was no problem.

Principal Ideas Advanced

•Translation from one language into another is indeterminate. Two schemes of translation, incompatible with each other, might be equally adequate and acceptable.

•Two kinds of entities exist: physical objects and sets of objects.

•Entities are posited to exist if they are empirically attested to or have theoretical utility.

•First-order predicate logic with identity is a canonical notation.

•A canonical notation—that is, a logically perspicuous language—makes clear the ontological commitments of a theory.

Bibliography

Arrington, Robert L., and Hans-Johann Glock. Wittgenstein and Quine. London: Routledge, 1996. The essays in this book address the similarities and differences between these philosophers, whom the authors rank as two of the leading philosophers of the twentieth century.

Borradori, Giovanna. The American Philosopher. Translated by Rosanna Croatto. Chicago: University of Chicago Press, 1994. Conversations with Quine and other leading philosophers are recorded in an easy-to-read, question-and-answer format. The interview with Quine reveals how he views his own philosophical development and place in the history of philosophy.

Brown, Stuart, Diane Collinson, and Robert Wilkinson, eds. One Hundred Twentieth-Century Philosophers. London: Routledge, 1998. Quine’s essential ideas and influence are briefly summarized.

Clarke, D. S. Philosophy’s Second Revolution: Early and Recent Analytic Philosophy. Chicago: Open Court, 1997. The author presents a readable and useful account of the origins and evolution of analytic philosophy, especially the role of Quine in the revolution. He argues that Quine’s method of philosophical analysis marks a radical departure in contemporary philosophy.

Davidson, D., and J. Hintikka. Words and Objections: Essays on the Work of W. V. Quine. Dordrecht, Holland: D. Reidel, 1975. The essays are a response to the arguments presented by Quine in his classic Word and Object. Quine’s responses to the remarks by the essayists are also included.

Gibson, Roger F., Jr. Enlightened Empiricism: An Examination of W. V. Quine’s Theory of Knowledge. Tampa: University Presses of Florida, 1988. The premise of this book is that philosophers have not understood Quine as well as they might. Gibson’s goal is to correct the situation.

Hacker, P. M .S. Wittgenstein’s Place in Twentieth-Century Analytic Philosophy. Oxford, England: Blackwell, 1997. This text provides an account of Quine’s place in analytic philosophy and compares his ideas with those of Ludwig Wittgenstein. The author documents the role of Quine’s ideas in the decline of Wittgenstein’s influence in analytical philosophy.

Hahn, Lewis Edwin, and Paul Arthur Schilpp, eds. The Philosophy of W. V. Quine. Peru, Ill.: Open Court, 1998. This handy guide includes an intellectual autobiography, a series of essays on Quine’s work and achievements, and a bibliography.

Leonardi, Paolo, and Marco Santambrogio. On Quine: New Essays. Cambridge, England: Cambridge University Press, 1995. This collection of essays pays homage to Quine. It is also devoted to making Quine’s work better understood in Europe.

Romanos, George D. Quine and Analytic Philosophy. Cambridge, Mass.: MIT Press, 1980. The author explores the relevance of Quine’s methods to various philosophical problems. The book grew out of the author’s doctoral dissertation. What makes this work especially useful is that Quine reviewed the author’s work as it was being written.

White, Morton G. Toward Reunion in Philosophy. Cambridge, Mass.: Harvard University Press, 1956. This work focuses on solving philosophical problems in the “spirit” of Quine. It also places Quine’s work in the context of philosophers who are his contemporaries.