Willard Van Orman Quine’s Word and Object is a work of analytic philosophy about how language hooks onto the world—and about how little of that hooking-on we can actually verify from the outside. Published in 1960 and dedicated to Rudolf Carnap, it belongs to the tradition of logical empiricism while quietly dismantling some of that tradition’s central assumptions. Quine writes as a Harvard philosopher addressing other philosophers, but the book’s famous thought experiments—especially the scene of a linguist trying to translate a wholly unknown language—have travelled well beyond the seminar room.
The book opens with a confession and a metaphor. We know physical things only through the impacts they make on our nerve endings, yet ordinary talk about desks and stones is not something we could ever replace with a more basic vocabulary of pure sensation. Otto Neurath’s image governs everything that follows: scientists and philosophers are sailors who must rebuild their ship plank by plank while staying afloat on the open sea. There is no dry dock, no neutral vantage point outside our inherited conceptual scheme from which to justify it. Quine’s target in the first chapter is the philosopher who pretends otherwise—who claims to strip away physicalistic conjecture and bare the raw sense data underneath. Even that stripping, he points out, relies on scientific knowledge (optics, anatomy) at every step.
From the Jungle Linguist to the Indeterminacy of Translation
Front Matter and the Preface
The book opens with a series title page for “Studies in Communication,” edited by Leo L. Beranek, Roman Jakobson, and William N. Locke, and a title page identifying Willard Van Orman Quine as Edgar Pierce Professor of Philosophy at Harvard. The copyright is 1960 by the Massachusetts Institute of Technology, and the dedication reads “To Rudolf Carnap, Teacher and Friend.” Two epigraphs set the tone: Otto Neurath’s image of scientists as sailors rebuilding their ship on the open sea without docking it, and James Grier Miller’s “Ontology recapitulates philology.”
In the Preface, Quine lays out his central claim: language is a social art learned through intersubjectively available cues, so linguistic meanings can only be construed as dispositions to respond overtly to socially observable stimulations. This limitation makes translation systematically indeterminate—the main theme of Chapter II—and the indeterminacy reaches even to what objects a term is true of; semantics of reference makes sense only when directed at our own language from within. He then sketches the chapters: anomalies and conflicts implicit in our referential apparatus (IV), remedies in the spirit of modern logic (V, VI), and imputing existence (VII). He lists the lectures where portions were presented—the six Gavin David Young Lectures at Adelaide in June 1959, lectures at Tokyo, the Howison Lecture at Berkeley, five at Stanford, the Royaumont colloquium, his presidential address to the Eastern Division of the American Philosophical Association, and 1956–57 lectures at Princeton, the Institute for Advanced Study, Columbia, and Pennsylvania. He mentions his Harvard philosophy-of-language course given ten times since V-J day, and courses at Oxford as George Eastman Visiting Professor and the A. T. Shearman Lectures at University College, London. Overlapping publications are cited, including “Le mythe de la signification,” “The scope and language of science,” “Quantifiers and propositional attitudes,” and “Logical truth.” He acknowledges sabbatical support from Harvard, the Institute for Advanced Study, the Ford Foundation, and the Rockefeller Foundation, and thanks Donald Davidson, Burton Dreben, and many named critics including J. L. Austin, Alonzo Church, Nelson Goodman, H. P. Grice, C. G. Hempel, Roman Jakobson, Thomas Kuhn, Hilary Putnam, and P. F. Strawson. The preface is signed at Stanford, California, June 3, 1959.
Beginning with Ordinary Things
The Contents list seven chapters with numbered sections. Chapter I, “Language and Truth,” opens with §1, “Beginning with Ordinary Things.” A desk manifests itself by resisting pressure and deflecting light; physical things are known only through effects at our sensory surfaces, yet common-sense talk of them proceeds without sensory explanation. Entification begins at arm’s length; things glimpsed, not glimpses, are the primordial points of condensation. Each of us learns language from others through observable mouthing under intersubjective circumstances, so public, conspicuous, nearby things are named first. Talk of subjective sense qualities is derivative—colors described as orange or heliotrope, smells as like rotten eggs; one identifies sense data by reflecting them in external objects, as one sees one’s nose best in a mirror. Philosophers from Berkeley onward try to strip away physicalistic conjectures and bare the sense data, yet even recapturing the data depends on sidelong glances into natural science—concluding visual data are two-dimensional from the ocular surface, from illusions of paintings and mirrors, or from the necessary surface interception of light; audition’s pitch and loudness from knowledge of frequency and amplitude. The motivating insight that we know external things only through impacts at nerve endings is itself based on general knowledge of physical objects. A philosopher may imagine an ideal “protocol language” evidentially prior to physical talk, but this misleads: immediate experience will not cohere autonomously; references to physical things hold it together and give our main access to past sense data, since actual memories are mostly traces of past conceptualization or verbalization, not past sensations. Conceptualization on any considerable scale is inseparable from language, and ordinary language of physical things is about as basic as language gets. Neurath’s boat image is invoked: philosopher and scientist rebuild plank by plank while afloat. Verbal perversity lies in denying that familiar physical talk is largely understood as it stands; “understood,” “real,” and “evidence” are too ill-defined to withstand such punishment. Dr. Johnson demonstrated the reality of a stone by kicking it; Johnsonian usage is our starting point. Some philosophers overdo this, treating ordinary language as sacrosanct and ignoring its disposition to evolve; scientific neologism is linguistic evolution gone self-conscious, and philosophy is not essentially distinguished from good and bad science. Certain existence idioms may take on key significance and be favored as existence affirmations “strictly so-called”; one could even end up denying existence to ordinary physical things in that refined sense.
Synonymy, Analyticity, and the Limits of Behavioral Definition
Quine opens §14 by stretching “synonymous” beyond its etymological home in names to cover sameness of meaning generally, refusing any sharp line between word and phrase—even “Gavagai” may or may not parse into parts, depending on the analytical hypotheses adopted. He then sorts synonymy into a broad and a narrow grade. Broadly, two sentences are synonymous when assent and dissent travel together strictly because of word usage rather than the run of the world; Quine prefers talk of assent and dissent to talk of truth values because it keeps the account anchored in verbal behavior. Carnap’s narrower “intensional isomorphism,” with its part-by-part correspondences, can be built on the broad relation: first equate parts by analogy of role within synonymous wholes, then define narrow synonymy through homologous parts. Length and complexity can sway assent, but §11 already sketched how to absorb that. For occasion sentences the notion surfaces as intrasubjective stimulus synonymy, best socialized, since only verbal habit plausibly explains concomitant variation across all stimulations; community-wide collateral information counts as a determinant of that habit. Standing sentences resembling occasion sentences, like “The Times has come,” still fare well under stimulus synonymy, though the less variable the sentence the sparser its stimulus meaning and the poorer the approximation. Lengthening the modulus of stimulation enriches stimulus meanings and tightens synonymy, but at the cost of scrutability—under a month-long modulus no one can say what to expect from fantastic sequences, and the subject may revise theories unpredictably. Stimulus synonymy on an optimum modulus approximates sameness of confirming and disconfirming experiences, as Grice and Strawson suggest; their own proposal, and Perkins and Singer’s, founder on the interconnections of sentences, since experience bears on many sentences only indirectly and systematic adjustments can absorb a recalcitrant datum. Grice and Strawson’s definition, recast with “every sentence S,” collapses into stimulus synonymy, and a conjunctive variant fails faster still. Either route yields strictly intralinguistic synonymy, though the auxiliary language need not be ours, since conjunction and the material conditional are translatable. Intrasubjective sentence synonymy is interdefinable with analyticity: sentences are synonymous just when their biconditional is analytic, and a sentence is analytic just when synonymous with a self-conditional. Tradition gestures at three nested categories—analytic, a priori, necessary—none defined by detectable verbal behavior. Defining analytic sentences as those affirmed “come what may” comes to nothing unless “what may” is circumscribed; read as “come what stimulation may,” it yields virtually stimulus analyticity, a debt Quine owes Davidson. Socialized stimulus analyticity covers “There have been black dogs” as readily as “2 + 2 = 4,” so it is a behavioristic ersatz, not a reconstruction. The intuition of analyticity is real enough—denials bewilder us as ungrasped foreign sentences do, and the effect can be gradual—but it will not support a sweeping epistemological dichotomy; “semantical rules” and “meaning postulates” merely assume the notion in disguise.
Analytical Hypotheses and the Indeterminacy of Translation
Section 15 gathers what passive observation and querying yield: observation sentences translate, truth functions translate, stimulus-analytic and stimulus-contradictory sentences are recognizable, and intrasubjective stimulus synonymy of native occasion sentences can be settled when raised—though those sentences themselves remain untranslated. To pass these bounds the linguist segments utterances into recurrent parts, compiles native words, and hypothetically equates them with English, conforming to the four established results. Such are analytical hypotheses, begun tentatively long before the four checks are complete and guiding the choice of examples—indispensable to the fourth, since without indirect hints there is almost no telling which non-observational pairs go together.
How much success here rests on real kinship of outlook rather than linguistic ingenuity or luck, and whether the question even makes sense, occupies the next stretch. We may marvel alternately at the inscrutability of the native mind or at how much like us the native is—having muffed the best translation in one case, read our provincial modes into native speech in the other. A contrast clarifies: islanders who call pelicans their half-brothers need no analytical hypotheses to make the cultural difference objective, since they have a short occasion sentence assenting indiscriminately to any half-brother or pelican, while English is oppositely endowed. Contrasts that depend on analytical hypotheses, by contrast, threaten to become meaningless. The frequent claim that deep linguistic differences carry ultimate differences in thinking runs into indeterminacy of correlation: the farther from sentences directly conditioned to non-verbal stimuli, and the farther from home ground, the less sense it makes to call a translation good or bad.
Vagueness at the Edges of Terms
Quine begins by distinguishing vagueness from mere imprecision. A term is vague when it is unsettled how far toward yellow something can be and still count as green, where muddy water gives way to wet mud, or how far from the summit one can stand and still be on Mount Rainier. Singular terms suffer this too: a named physical object may have fuzzy boundaries in space-time. “Mountain” is vague in two directions at once—how much terrain belongs to each indisputable mountain, and which lesser eminences count at all. “Organism” shows both sorts more mildly: is ingested food part of the organism, does the individual date from conception or the severed cord, is a slime mold one thing or a colony? The first sort produces indeterminacy of count. “Big” and “little” are extravagantly vague, partly because they are relative to classes—big butterflies, little elephants—though that relativity is syncategorematic use rather than vagueness proper. Polar words can be relativized (“greener”) to narrow the spread, and vagueness does not disturb the truth values of ordinary sentences; pressure for reform comes only when a sentence hinges on the penumbra, as in legal and almanac disputes over the biggest fresh lake, the longest river, the largest city, or Colorado’s fourteen-thousand-foot peaks. Vagueness even earns its keep: an expositor can state A loosely, move to B, then touch up A without forcing the reader to unlearn falsehoods.
Ambiguity of Terms and of Syntax
Ambiguity differs from vagueness: “light” can be clearly true and clearly false of the same dark feathers. Context often settles it, and homonyms are split by lexicographers for convenience. Verbal nouns carry systematic ambiguities—process versus product (“assignment”), action versus custom (“skater”). Quine’s nearest clear test: a term is ambiguous when, from utterance to utterance, it can be clearly true or false of one and the same thing according to interpretative clues. “The door is open” shifts with the door’s motion and from door to door, yet is not ambiguous. Where ambiguity turns on syncategorematic use, the compound is the ambiguous term—”poor” may impute poverty, express pity, or suggest “badly,” and “big butterfly” means more F than the average G. “Nothing” and “nobody” masquerade as definite terms, generating the puzzles of Gershwin, Lewis Carroll, Locke, Heidegger, and Parmenides. Syntax adds its own ambiguities: inclusive versus exclusive “or”; “The violinist was poor”; and the plural subject that slides between “every,” “some,” and the abstract class, as in “Lions like red meat,” “Lions are roaring,” “Lions are numerous.”
Referential Opacity
Some contexts resist substitution of co-designating terms. “Tully was a Roman” and “Cicero was a Roman” differ inside quotation, so the name there is not purely referential. Quine calls a construction transparent when purely referential positions stay referential inside it; truth functions and predicative general terms are transparent, but “believes that” is opaque. Tom may affirm “Cicero denounced Catiline” while denying that Tully did, because he thinks them distinct. Indefinite terms complicate matters: “The commissioner is looking for someone” must be read referentially, and “Tom believes that someone denounced Catiline” splits by scope. Quantifying into opaque constructions breeds disaster—grant transparency and Tom believes everything, true and false alike. So Quine localizes opacity in the “that” and the “to,” not in “believes,” letting “Tom believes Cicero to have denounced Catiline” keep its referential grip.
Regimentation, Time, and Names Reparsed
Canonical notation reduces “every F” and “some F” to quantifiers, one sufficing, as Frege showed in 1879; descriptions, class abstracts, and intensional abstracts become variable-binding operators, though class abstraction cannot eliminate attribute abstraction. No variable inside an opaque construction is bound from outside. Tense is dropped for a temporally neutral present, with “will” as “after now,” and physical objects are conceived four-dimensionally. Finally, names are reparsed: since purely referential occurrences reduce to the form “= a,” Quine treats “= Pegasus” and “= Socrates” as indissoluble general terms. “Pegasus exists” becomes “(∃x)(x is Pegasus),” false; “Socrates exists,” probably true. Namehood is relative to the project, and terms like “is drawing Pegasus” are tortured into propositional attitudes so the name keeps referential position within its immediate sentence.
Modality and the Opacity of Necessity
Why keep the operatorial form? An apparent advantage is quantifying into modal positions, since one cannot quantify into quotation. But “Necessarily 9 > 4” is true while “Necessarily the number of major planets > 4” is false; since 9 equals that number, the position of “9” is not purely referential and the necessity operator is opaque. Such opacity depends on “stubborn” objects like 9, specifiable in ways that fail of necessary equivalence. Narrowing the universe to exclude them removes the objection, legitimizing quantification by a postulate: if Fx and x alone and Gx and x alone, then necessarily all Fs are Gs. But that postulate annihilates modal distinctions—from any truth p and an object y, one could conclude necessarily p. Barcan’s and Fitch’s systems allow unrestricted quantification, unclear without the postulate; Church’s is more cautious, restricting special variables to intensional objects. Alternatively, one might pin opacity on intensional abstraction, or render modality selectively transparent as “x[x > 4] is necessary of 9.” But this yields baffling talk of necessary versus contingent attributes: mathematicians necessarily rational but not two-legged, cyclists necessarily two-legged but not rational—what of one who is both? The tradition of essence and accident is indefensible, so this route must go. Given propositional abstraction, necessity can still be defined, yielding something isomorphic to unquantified modal logic.
Propositions as Meanings
Learning “apple” or “river” involved learning sameness of re-exposure; similarly “proposition” needs a standard of identity. A proposition, not being physical, admits something analogous: identity of propositions is the question how two eternal sentences must relate for one to be the same proposition as the other. The usual answer: the sentences are synonymous, and the proposition is the sentence’s meaning—propositions being meanings of eternal sentences, not all declarative sentences, since “The door is open” stays unchanged while its associated proposition varies by occasion. An expression’s meaning is not its designated object; sentences do not designate, though words may. If identity of propositions is defined by synonymy, propositions may be identified with the class of mutually synonymous sentences; the remaining worry is synonymy of eternal sentences. For propositional attitudes the broad synonymy of earlier chapters is too broad—it would reckon all analytic sentences identical, unacceptable in belief contexts.
Attitudes Without Intensional Objects
Intensional abstraction localizes the opacity of propositional attitudes, so attributes cannot be replaced by classes, since class abstraction is transparent. Let x[Fx] be an attribute such that “Tom believes x[Fx] of a” is false; still “Tom believes x[Fx or x = a] of a” holds, given Tom’s trace of logical sense. If, unbeknownst to Tom, Fa, then the two attributes are coextensive, and the needed independence is lost under class abstraction. Classes are disqualified because their identity condition is mere coextensiveness. Taking objects with stronger identity conditions disturbs only special sentences like “Paul and Elmer agree on just three things,” whose numbers might be revised upward, even to infinity.
Quine proposes using sentences themselves, with notational identity as the identity condition: “Tom believes-true ‘Cicero denounced Catiline’,” and correspondingly for believing-true a predicate of Cicero, or a relation of Cicero and Catiline. Quotation will not fail as class abstraction did; it is a vivid opaque form reducible to spelling. The proposal extends to other attitudes: “Tom says that Cicero denounced Catiline” becomes “Tom says-true ‘Cicero denounced Catiline’.”
Explication as Elimination
Quine turns to the ordered pair as the model of what philosophers do when they “analyze” a muddled idea. Explication, as he describes it, makes no claim of synonymy and uncovers no hidden meaning; it supplies a lack. One fixes on the functions that make the unclear expression worth keeping, then devises a clear substitute that fills them. Wiener’s definition of (x, y) as {{x}, {y, Λ}} shows how little partial agreement is required; Kuratowski’s {{x}, {x, y}} is another version, and in number theory one may want pairs that never leave the natural numbers, such as 2^x·3^y. Each conflicts with the others, yet all are right, since they clash only over the “don’t-cares.” The air of paradox, Quine says, comes from supposing there is one unique correct analysis—a supposition encouraged by using the same label for every version. Explication is elimination: we abandon a troublesome expression and accomplish its purposes through other channels. Wittgenstein’s doctrine that philosophy dissolves rather than solves problems fits this well; Russell dissolves singular descriptions in favor of identity and quantifiers, and the indicative conditional in favor of a truth function. Ironically, philosophers most influenced by Wittgenstein deplore such explications as departures from ordinary usage.
Numbers and the Mind-Body Question
“What is a number?” stands on the same footing. Frege identified each natural number with a class of classes, 0 with {Λ} and n+1 with the class of all classes that become n-membered when deprived of a member. Von Neumann, playing Kuratowski to Frege’s Wiener, identified 0 with Λ and n+1 with the class of all classes identified with 0 through n. Any progression will do—any infinite series each of whose members has only finitely many precursors. Russell once added the condition that numbers measure multiplicity, but that was a mistake, since any progression can be fitted to the paraphrase. All three versions are present in our universe of values and available for selective use, so natural numbers need not be reckoned in additionally. Explication depends on compensatory revision of adjacent text: picturing Frege’s move as avoiding reference to n by recourse to N, “has … members” is not equated with “∈” but revised as “∈N.” This bears on mind and body. A defense of physicalism: talk of mental events proceeds in terms acquired through association with socially observable behavior of physical objects. If there is a case for mental events, it must be that positing them has indirect systematic efficacy in theory—but as much organization could be achieved by positing something else.
No Cosmic Exile
Quine rejects a dichotomy: pragmatic considerations of systematic efficacy are not confined to semantic ascent, nor are factual considerations about objects’ behavior confined to talk within the theory. Both are equally essential in both cases. The feeling that observation has no bearing on logic, as it does on physics, traces to misapprehensions about semantic ascent and to curriculum classifications; the truth is a matter of degree—of centrality to the theoretical structure and relevance to sets of observations. One more dodge for enjoying abstract objects without accepting them is the suggestion that such acceptance is a linguistic convention distinct from serious views about reality. But the question what there is is shared by philosophy and most non-fiction: geography describes land masses, zoology groups animals, physics abstracts ruthlessly, and pure mathematics belongs to the descriptive answer too. What distinguishes the ontologist is only breadth of categories. Scrutiny of our uncritical acceptance of physical objects or classes devolves upon ontology—making explicit what was tacit, exposing paradoxes, clearing ontological slums. There is no cosmic exile, no vantage point outside the conceptual scheme; the philosopher scrutinizes and improves the system from within, appealing to coherence and simplicity. No experiment settles an ontological issue, only because such issues connect with surface irritations through a multifarious maze of intervening theory.
The Apparatus
The book closes with its apparatus: a list of bibliographical references, restricted to works alluded to elsewhere, running from Ajdukiewicz through Zipf, and an index of terms from “About” and “Abstract object” onward.
What the Book Argues About Meaning
The argument that made the book famous runs roughly like this. A linguist dropped into a culture with no prior contact hears the natives say “Gavagai” whenever a rabbit bolts past. He writes down “rabbit” and moves on—but nothing in the observable behaviour justifies that choice over “undetached rabbit part,” “rabbit stage,” or “rabbit fusion.” Stimulus meaning, Quine’s behaviourist substitute for meaning, can tell us which stimulations prompt assent and which prompt dissent. It cannot tell us what the native is talking about. Translation is therefore indeterminate: two rival manuals of translation can each fit every disposition to verbal behaviour perfectly while assigning incompatible meanings to the same native sentences. The conflict, Quine says, is a conflict of parts seen without the wholes—and since the wholes are empirically equivalent, there is no fact of the matter about which is right.
This is not scepticism about other minds, and Quine is careful to distinguish it from the domestic analogue. When two English speakers match in all verbal dispositions, we happily say there is no sense in imagining semantic differences between them. The interlinguistic case is the same, only more visible. What the indeterminacy really exposes is the empirical slack in our own beliefs: our theories are underdetermined by all possible sensory evidence to exactly the degree that translation is underdetermined by verbal behaviour. Quine notes, with some irony, that philosophers have been far more willing to accept the domestic version than the foreign one.
Reference Without Foundations
The middle chapters turn from meaning to reference—how our language manages to pick out objects at all. Quine traces the ontogenesis of reference in the child: babbling gets selectively reinforced (“Mama” earns a smile), occasion sentences give way to general terms, and the child eventually scrambles up what Quine calls an intellectual chimney, supporting himself against each side by pressure against the others. “Same,” “another,” “that,” “an,” and the plural “-s” are learned together, each adjusted to the others until a coherent pattern emerges. Only when the child can handle “that apple,” “not that apple,” “another apple” does genuine divided reference appear—and with it the scheme of enduring, recurring physical objects.
Predication, the joining of a singular term to a general term, is where the general/singular distinction properly lives—not in how many objects a term happens to be true of. “Pegasus” is singular though true of nothing; “natural satellite of the earth” is general though true of one. Quine builds up the apparatus of relative clauses, indefinite singular terms, and descriptions, showing how each construction widens referential reach. Then he turns to the failures: vagueness (where does a mountain end?), ambiguity (“light” feathers, “light” reading), and the puzzles that arise when terms sit inside quotation marks or belief contexts. His treatment of referential opacity—the way “Tom believes that Cicero denounced Catiline” can fail to be about Cicero at all, since Tom may not know Cicero is Tully—leads to the notorious conclusion that if belief were transparent, Tom would believe every true sentence and every falsehood too. Tom believes everything. The fix is to localize the opacity in the “that” of “believes that,” not in “believes” itself.
Regimentation and the Flight from Intension
The later chapters propose a cure for these tangles: canonical notation. Quine does not ask us to speak in logical symbols; he asks only that we show the notation could do the work of every idiom it claims to handle. Quantifiers, variables, and truth functions replace the treacherous apparatus of ordinary speech. “Any” and “every” collapse into “such that”; indefinite singular terms are held to subject position; names give way to predicates. The point is not elegance but honesty about what we are committed to. When we say “there is a number between two and four,” the canonical notation forces us to say exactly what we are positing.
This is where the flight from intension begins. Quine has no use for meanings as entities, for propositions, for attributes—all the intensional furniture that other philosophers treat as real. His behaviourism is methodological rather than metaphysical: he is not denying that we have experiences, only insisting that a theory of language built on publicly observable dispositions cannot invoke private meanings to explain them. The stimulus meaning of a sentence is a disposition to assent or dissent under stimulation, and dispositions, Quine concedes, have a murky ontological status—he defers the question to a later section and never quite settles it. What matters is that the notion is behavioural, checkable, and free of intensional commitments.
Ontic Decision
The book closes by asking what all this means for ontology—for what we say exists. Quine’s answer is deflationary and, in its way, liberating. To be is to be the value of a variable: the ontological commitments of a theory are exactly the objects its quantifiers must range over if the theory is to be true. We cannot step outside our theory to compare it with reality bare; we can only compare theories with each other, asking which is simpler, which fits the evidence, which does the work we need done. The physicist who posits molecules and the philosopher who posits sense data are both engaged in the same enterprise of constructing a scheme that holds together. Quine’s preference for physical objects over sense data is not a discovery about what really exists but a judgment about which scheme is simpler and more serviceable.
This is the payoff of Neurath’s boat. We are always already afloat, always already committed to some scheme of objects, and the most we can do is rebuild carefully, plank by plank, keeping the bulk intact. Quine’s epigraph from James Grier Miller—”Ontology recapitulates philology”—is a joke, but a serious one. What we say exists is bound up with what our language lets us say.
The Book’s Afterlife
Word and Object reset the terms of several debates at once. The indeterminacy of translation became a standing challenge to any theory that treats meaning as a determinate object of study; the critique of the analytic/synthetic distinction, developed further in Quine’s earlier “Two Dogmas of Empiricism,” here receives its behavioural grounding. The doctrine of ontological commitment—”to be is to be the value of a variable”—entered the standard vocabulary of analytic philosophy. And the naturalized epistemology sketched in these pages, which treats the theory of knowledge as a chapter of empirical psychology rather than a tribunal above the sciences, set the direction for Quine’s later work and for a generation of philosophers who followed him.
The book is difficult, compact, and often maddening—Quine’s prose is dry, his examples are homely, and his arguments have a way of turning back on themselves. But it is also one of the few works of twentieth-century philosophy that genuinely changed what its readers thought they were doing. The jungle linguist and his rabbit remain, sixty years on, the standard illustration of a problem that has never quite been solved: how language reaches the world, and whether we could ever know if it missed.
