Quine's Word and Object: Indeterminacy of Translation

Michael BrenndoerferUpdated March 18, 202513 min read

W.V.O. Quine's 1960 masterwork introducing the indeterminacy of translation and inscrutability of reference, challenging traditional assumptions about meaning and language.

PhilosophyAll chapters
Reading tools Notes & comments
Notes & comments
Part 8 of 12Logic, Language, and Interpretation
Chapter 105Translation and its indeterminacy › What the linguist can learn from a rabbit
Reading the full chapter · 7 sections · 13 min7 sections · 13 min

All 7 sections · 13 min read

Translation and its indeterminacy › What the linguist can learn from a rabbitFull chapter · Sections in this chapter1 / 7 ↓

What the linguist can learn from a rabbit

A rabbit runs by. Someone says “Gavagai.” A linguist records the sounds and provisionally writes down “Rabbit!” The next test is to repeat the unfamiliar utterance on a suitable occasion and learn whether the speaker accepts it. Several such trials could give the linguist good reason to translate the whole remark as “Look, a rabbit.” W. V. O. Quine's question is whether the same trials already tell the linguist how an expression inside the utterance refers to a particular kind of object. Quine, Word and Object, chapter II, §§7 and 12.

Quine introduces this example in Word and Object (1960) to investigate radical translation. The linguist is to construct a translation without available bilingual speakers, a prior manual, or a known relation between the languages. Ordinary translators have learned much that this person still has to work out. Taking those resources away makes it easier to distinguish what the observations warrant from what the translator contributes in organizing them. The example will eventually require distinctions between the acceptability of a whole utterance, the reference of expressions within it, and an entire system of translation. Word and Object, chapter II, §§7–8 and 12–16.

How evidence starts to accumulate

The earliest work starts with utterances that visibly connect to a situation. Someone says a particular thing when a rabbit crosses a path. The linguist tries the expression again, leaving room for the other person to assent, dissent, or fail to give either response. Recognizing these responses is itself an achievement. The linguist has to identify ways people accept and reject proposed utterances, including which behavior belongs to replying and which belongs to reacting to the interruption. In Quine's setup this is an empirical matter. When a reply turns up consistently in circumstances where the speaker would be expected to agree, the linguist has a working hypothesis about how assent is expressed. Word and Object, chapter II, §7.

Is one glimpse of a rabbit enough? Suppose the speaker says “Gavagai” twice, once when a rabbit is clearly present and once after something stirs in the grass. They may have seen rabbits coming out of that patch earlier. The linguist may have seen only movement and concluded too quickly that the expression applies to something else. The reverse error is also possible. The linguist might expect agreement when an animal comes into view, and the speaker may have a reason to take it for something different. Quine treats these as problems about collateral information. What someone takes from a current scene can depend on what they already know. Word and Object, chapter II, §§9–10.

Occasion sentences can receive different verdicts as the occasion changes. “It is raining” can be affirmed one day and denied the next without any change in the speaker's knowledge of the language. For observation sentences, agreement between competent speakers tends to be closely tied to the scene they are jointly situated to observe. Quine treats observationality as a matter of degree. The same speaker may be much more dependent on previous information to respond to “Bachelor!” than to respond to “Red!” Seeing a person is quite enough to bring the question of bachelorhood to mind, but the relevant information may have been acquired long before that moment. Word and Object, chapter II, §§9–11.

It matters that the linguist can distinguish an unhelpful sample from an utterance that is intrinsically less revealing in this way. Looking at rabbits from several angles and across several occasions helps the linguist find out when the remark is appropriate. Observing a person every time a team member says “Bachelor!” is much less likely on its own to reveal why the verdict is appropriate. A theory of translation has to make room for both patterns. The first makes a useful foothold precisely because the second is not always available on those terms.

Stimulus meaning

For Quine, a sentence's stimulus meaning for a given speaker at a time is an ordered pair. One part comprises the stimulations that would prompt assent when the sentence is queried. The other comprises those that would prompt dissent. Some stimulations would prompt neither. The account is about dispositions to respond and can include several senses, not just vision. Word and Object, chapter II, §8.

The sets in that definition are not a list of the things an expression names. Seeing a rabbit in light and seeing it in shadow might both prompt assent. Seeing a fox instead might prompt dissent, if the conditions are right. Many episodes can be compared once we have a sense of what portion of experience belongs in the test. Quine's definition also fixes a maximum duration for stimulations and allows a person's disposition to change over time. These specifications let him ask when an English utterance and an unfamiliar utterance serve the same purpose as occasion sentences. They do not assume we already know how the speaker classifies individual objects. Word and Object, chapter II, §§8 and 12.

The sentence and the term

Suppose “Rabbit!” and “Gavagai!” have closely matching stimulus meanings. In circumstances where a rabbit is visible, a part of a rabbit that is still attached is visible too. So is a temporal stage of the rabbit's life. A decision to correlate the whole utterances does not settle which of these objects a term can be true of if it occurs inside a longer sentence. It also leaves open where the word or construction that plays the role of a term begins and ends. That is why the experiment starts from the whole utterance rather than treating the first sound the linguist records as a word with an assigned object. Word and Object, chapter II, §12.

We can test the difference by pointing twice. A rabbit darts behind a tree and reappears. A linguist who had already established the words for “same” and “one” might ask whether that was one and the same rabbit. To establish those expressions in the first place, they would have to know which responses count as confirming an identity or a count. The problem is especially visible if the rabbit turns around and presents a different side. One scheme can record the same animal under a new view, while another can treat the stages or parts as different units joined by an appropriate relation. What the speaker answers about sameness may fit either scheme once the expressions for individuation have been interpreted accordingly. Word and Object, chapter II, §12 and §§15–16.

Try replacing the English word “rabbit” with “rabbit part” in a translation while leaving everything else word for word. The result would get plenty of cases wrong. For example, a remark that translated as “There are two rabbits” would be false as a claim about the total number of rabbit parts. Quine's argument requires a coordinated change that accommodates identity, number, and other expressions. The important suggestion is that translation can still get right the observable conditions in which people endorse that whole remark while differing in how it assigns values to the parts. The possibility of compensation gives the example its force. A bare substitution makes no such point. Word and Object, chapter II, §§12 and 15.

There is room to evaluate the translations using ordinary constraints. A translator can learn more expressions, test agreements in fresh situations, and aim for a coherent grammar. A manually shuffled translation that recovers only the first mention of a rabbit will generally fare badly over the rest of the language. Quine's conjecture is about schemes that meet much more demanding conditions. The reference question becomes pressing when changing the scheme as a whole leaves the judgments the experiment can test in place. An extra photograph of the rabbit cannot on its own decide the scheme in which an identity question should be read.

Building a manual beyond the first observations

An analytical hypothesis helps the linguist construct a dictionary or grammar. It can segment a recurring portion of an utterance, say what kind of expression it is, or say how its translation varies with construction or context. Some of these choices can be guided by the first results and tested against new cases. Some help the linguist decide what examples to elicit in the first place. Quine's picture is therefore more complicated than writing “rabbit” beside “gavagai” and then moving on to the next word. A good proposal has to work across the exchanges in which the expression can occur. Word and Object, chapter II, §15, pp. 68–72.

Some work can proceed without agreeing on the whole manual. Quine asks how the linguist might pick out a form of negation. If the speaker accepts a sentence, they should in suitable circumstances reject the result of putting it in that form. If they reject the original, they should be willing to accept the result of applying the form, subject to complications of ordinary use. Comparing patterns of assent and dissent helps locate the operation even while many expressions in the sentences remain unknown. Quine places logical connectives alongside observation sentences among the ordinary constraints a translator can make use of. Word and Object, chapter II, §§13 and 15.

The tests work best with simple component sentences. A speaker can find a long conjunction confusing or decline a question while remaining perfectly able to use the connective. Quine also cautions that English words such as “and” and “or” have uses beyond their clean truth-functional readings. The evidence is for a particular job a construction performs. The translator still has to learn when that job is the one it is doing. This is especially important in deciding whether a person really denies an elementary logical truth. If an interpretation of a connective makes nearly everything a speaker says plainly incoherent, a hidden difference in the use of the connective can be a better explanation. Quine treats this as a principle of translation and refers to Neil Wilson’s principle of charity in the discussion. Word and Object, chapter II, §13, pp. 56–59.

There are far fewer opportunities to check a sentence about something remote or theoretical. If a community frequently affirms a sentence that the linguist provisionally translates in terms of reincarnation, other parts of its discourse can give that result support or pressure the linguist to change it. The translation can depend heavily on how whole groups of expressions were initially assigned. Quine uses this sort of example to explain how a system can be disciplined by observations and still reach beyond what those observations determine directly. New evidence can establish that a manual makes a particular mistake without settling every decision about how to render the rest of what the speakers say. Word and Object, chapter II, §§15–16.

The scope of the indeterminacy claim

The example of the rabbit and its parts is about reference. The linguist can find evidence for roughly equivalent occasion sentences while lacking a basis for choosing what the terms within them are to count as denoting. Quine makes a further claim about rival manuals. The manuals might fit the speaker's total behavioral dispositions while producing incompatible translations of whole sentences elsewhere in the language, especially where no direct test settles what counts as the translation. Such a case would go beyond making two different assignments that coincide at the level of all translated sentences. Word and Object, chapter II, §§12, 15–16.

That last step is controversial. Early assent tests may give the same results for two incompatible accounts and then further evidence could favor one of them. Quine is claiming more than this familiar sort of limited evidence. He asks the reader to grant a fit with everything the people concerned could be disposed to say under relevant conditions. If rival manuals remain possible even then, there is a question whether a fact of correct translation still needs to be settled. He says that producing two fully worked rival manuals would take a duplicate enterprise, and argues from the character of the evidence and the procedure instead. The rabbit example opens that argument. It does not produce the pair of manuals the broader thesis calls for. Word and Object, chapter II, §§15–16, especially pp. 70–74.

Someone could grant the point about evidence and say the speaker nevertheless means something determinate. They could propose that the speaker's representation of a rabbit has a causal or psychological identity not revealed by observed agreement. Quine's challenge is to specify what the proposed difference would establish. In §16 he considers the thought that a bilingual's correlation may be realized by a neural condition. Suppose another bilingual has an incompatible correlation yet the two use each language in the same ways when considered separately. Locating a correlation in these people's brains would tell us something about how they translate and still leave open which correlation captures a further language-wide fact. That case preserves a role for psychology while pressing the demand for an account of what fixes the translation. Word and Object, chapter II, §16, pp. 72–75.

Indeterminacy also differs from the availability of several graceful phrasings in English. Translators frequently choose among expressions they regard as interchangeable for current purposes. Quine asks about manuals that support incompatible readings when their outcomes for a particular sentence are set beside one another in isolation. One answer may be appropriate only within the other commitments of its own manual. That is why he compares whole systems in §§15–16 and makes the compatibility of each with observable dispositions part of his thesis. Word and Object, chapter II, §16, pp. 72–78.

Why a translator can still prefer one manual

Quine notes practical constraints that make translation possible. A fieldworker needs a manageable set of rules that will carry over to new sentences. They may favor familiar divisions or short, consistent renderings. Early commitments lead to later decisions about what counts as a word and how sentences are put together. Two well-trained translators could make the same decisions because they aim for treatments that are easy to apply and share habits of parsing and individuation. Agreement of that kind can be reasonable and useful. The disputed inference is whether it proves that the evidence, taken on its own, singled out the entire system. Word and Object, chapter II, §16, pp. 72–75.

Return to the translator watching the rabbit. They can learn to call attention to it in the unfamiliar language. They can learn enough grammar to ask how many animals there are, what the speaker last saw, and whether it has gone back into the grass. The richer the exchanges, the more grounds the translator has to exclude errors. Quine asks at what point this achievement establishes that every part of the manual is uniquely fixed. A reader can accept the success of the conversation and still disagree with him about what makes one translation preferable. The experiment gives that disagreement a concrete form.

Further reading

Join the discussionAsk a question or share what you learned.

Comments

Add to the conversation

A question, a correction, or an example from your own work.

No comments yet. Be the first to share your thoughts!

A different angle
All writing
Philosophy

Derrida's Of Grammatology: Writing, Trace, and the Supplement

Jacques Derrida's 1967 inquiry into the relation between speech and writing, with close readings of Saussure, Lévi-Strauss, and Rousseau.

13 min read
Philosophy

Derrida's Writing and Difference: Structure, Sign, and Play

Jacques Derrida's 1967 collection of essays, with a close reading of “Structure, Sign, and Play” and companion essays on Foucault, Freud, and Levinas.

13 min read
Philosophy

Kripke's Naming and Necessity: Rigid Designation and Essentialism

Saul Kripke’s lectures on how names refer, why some necessary truths are discovered empirically, and what people and things could have been.

13 min read

About the author

Reference

Citation details

Cite or share this article.

BIBTEXAcademic
@misc{brenndoerfer2025quines, author = {Michael Brenndoerfer}, title = {Quine's Word and Object: Indeterminacy of Translation}, year = {2025}, url = {https://mbrenndoerfer.com/writing/quine-word-object-indeterminacy-translation-guide}, organization = {mbrenndoerfer.com}, note = {Accessed: 2026-11-19} }
APAAcademic
Michael Brenndoerfer (2025). Quine's Word and Object: Indeterminacy of Translation. Retrieved from https://mbrenndoerfer.com/writing/quine-word-object-indeterminacy-translation-guide
MLAAcademic
Michael Brenndoerfer. "Quine's Word and Object: Indeterminacy of Translation." 2026. Web. November 19, 2026. <https://mbrenndoerfer.com/writing/quine-word-object-indeterminacy-translation-guide>.
CHICAGOAcademic
Michael Brenndoerfer. "Quine's Word and Object: Indeterminacy of Translation." Accessed November 19, 2026. https://mbrenndoerfer.com/writing/quine-word-object-indeterminacy-translation-guide.
HARVARDAcademic
Michael Brenndoerfer (2025) 'Quine's Word and Object: Indeterminacy of Translation'. Available at: https://mbrenndoerfer.com/writing/quine-word-object-indeterminacy-translation-guide (Accessed: November 19, 2026).
SimpleBasic
Michael Brenndoerfer (2025). Quine's Word and Object: Indeterminacy of Translation. https://mbrenndoerfer.com/writing/quine-word-object-indeterminacy-translation-guide
Newsletter

Stay up to date

Get new articles, book updates, and notes on AI and technology.

Unsubscribe at any time. Privacy policy