In his work "The Significance of Quine's Indeterminacy Thesis," Michael Dummett explores Quine's philosophy. In the beginning of section 3, he states: "Indeterminacy of translation proper involves that two equally acceptable schemes of translation from a language L into a language M might carry a given sentence of L respectively into distinct sentences of M to which a single speaker of M would ascribe different truth values."

He goes on to clarify that the scenario he discusses is not one involving "soft" (my term) sentences, whose truth value can vary depending on the speaker, but rather a case of pure formal divergence in translations.

If this is indeed the case, I fail to understand how it is possible. I comprehend how the two translation schemes can render one sentence of L into two sentences of M with different meanings, while the dispositions regarding them remain invariant (this is my understanding of Quine's view). However, I struggle to see how these two sentences can also have two different truth values.

When the original sentence is presented to the native speaker of the object language L, they might either assent or dissent from it, enabling the translators to determine its truth value. Hence, if the two translations differ in truth value, there should be a way to dismiss one as incorrect while retaining the other as a valid translation.

What aspect am I overlooking in Dummett's argument?

  • 3
    Might it be just as simple as a word in the sentence having contradictory meanings, such as cleave, which can mean to split or to stick together? You can take the meaning of a sentence contains the word 'cleave' in one of two mutually contradictory ways even without translating it. Oct 26, 2023 at 19:49
  • 2
    Of course if L was written by you M could be fully determined, but if you're not its author its meaning indeterminism immediately kick in and is related to the famous rule-following paradox: Kripke writes that this paradox is "the most radical skeptical problem that philosophy has seen to date"...no course of action could be determined by a rule...But if there cannot be rules governing the uses of words, as the rule-following paradox apparently shows, this intuitive notion of meaning is utterly undermined.... Oct 27, 2023 at 6:31
  • @MarcoOcram this could work as a solution! but if this is the case why does Dummett complicate the whole section so much, instead of just stating this simple idea? Could it be pure mannerism?
    – Amit Hagin
    Oct 27, 2023 at 6:38
  • @DoubleKnot to my understanding these are two distinct arguments for the absence of meaning. In that sense, indeterminacy is a stronger argument than the rule-following paradox: According to Kripke no rule forces 2 speakers to follow it in the same manner, and therefore no rule can determine meaning to begin with. Yet, even if the 2 speakers follows one rule in exactly the same way, according to the indeterminacy thesis there will be 2 different translations to their behavior.
    – Amit Hagin
    Oct 27, 2023 at 6:40
  • 2
    The phrase 'there are no examples' quoted above makes me immediately cynical about the value of the concept. Oct 27, 2023 at 7:53

3 Answers 3


Your response is a standard initial one to Quine’s thought experiment. We naturally assume that the translator is making some kind of imposition onto his observed subject language community, and that indeterminacy is a matter of working out where you went wrong, rather than an actual feature of truthful translation as such. But the problem does go a little deeper than simply getting meanings wrong, because the dispute isn’t just about meaning but also about method.

This is the crux of the Quine/Dummett disagreement, because Dummett is a Logical Empiricist. For Dummett, contradictory translation is something that happens in practical translation projects, and is very constructive when it reveals something fundamental about what we’ve presumed about our host linguistic communities. We might have to do a deep review of some of our more foundational meaning posits to get to the point of a sound interpretation - maybe “is” works differently in this language! But eventually, subject to sufficient revision and contact with reality, we will reach a place of methodological well-foundedness such that a model of “what is meant” can be drawn upon as a way of making sense of what any given speaker is saying.

For Quine, the point isn’t the presence of contradictory meanings as such but rather that we have to recognise in our current interpretive paradigm that there is no privileged consistent ground from which to carry out the translation. We can’t be certain what methodological variations in our linguistic meta-practices are themselves influenced by our home language, and even reflexive linguistic interpretation can’t be trusted as a guide to our own interpretive resources. As Wittgenstein says, “whereof we cannot speak, thereof we must be silent” - so if we are through linguistic convention condemned to silence about our own linguistic practice, then it cannot serve as a unique island of established Truth to ground the interpretation of other languages. Indeterminate translation is a perpetually lurking threat given our linguistic methodology, because of the dynamics between language practices. In other words, the presumption that the real metatheory is just a revision of our own is an unfounded assumption, and might lead us down some very misguided directions.

It’s a reasonable challenge, and humanity’s history of e.g. Eugenics suggests this is not a purely theoretical idea. Humans have gone down some really screwy scientific paths driven by cultural contingencies, and some amount of methodological humility is tremendously important (perhaps a current example of this might be replicability and scientific publication culture). But if we follow Popper down the falsificationist path, the idea is that when we encounter error, we see it as a chance to revise, not a skeptical dissolution of the whole enterprise. No, we don’t have a privileged foundation of truth, and perhaps we have no real justification to say we ever will, but we do know some things that don’t work, and as we subject ourselves to greater scrutiny, we can at least work in the direction away from what we know to not work.

This is (if not exactly at least adjacent to) what Dummett means when he talks about “Conclusive” disagreement resolving to simple disputes of meaning - there will, eventually, turn out to be some matter of fact which separates the two theorists / the plural interpretation that gives rise to contradictory interpretations. But it doesn’t wholly resolve Quine’s point, because there remains one important puzzle - if the two contradicting evaluations are modelled as speakers who are indeed speaking the same language M, but the dispute between them is eventually resolved, what language are they speaking now, and what, exactly, was M to begin with?

  • Whenever I see those little rabbit ears, it reminds me of... TV. So, "what is meant” has to lie outside of language, which is only "fingers pointing to the moon." I agree that meaning has to converge, but language and thought may be lacking.
    – Scott Rowe
    Oct 28, 2023 at 17:02

Assume that the scenario involving "two equally acceptable schemes of translation from a language L into a language M might carry a given sentence of L respectively into distinct sentences of M to which a single speaker of M would ascribe different truth values" is plausible .

We have an original sentence S of language L that is mapped into two different sentences S1 and S2 of language M by two different translation schema.

Then you say: "when the original sentence is presented to the native speaker of the object language L, they might either assent or dissent from it, enabling the translators to determine its truth value. Hence, if the two translations differ in truth value, there should be a way to dismiss one as incorrect while retaining the other as a valid translation."

But this is exactly the gist of Quine's thought experiment with "Gavagai" vs "Rabbit" into Ch.2 Translation and Meaning of Word and Object (1960):

A rabbit scurries by, the native says 'Gavagai', and the linguist notes down the sentence 'Rabbit' or 'Lo, a rabbit') as tentative translation. [...] For, suppose the native language includes sentences S1, S2, and S3, really translatable respectively as 'Animal', 'White', and 'Rabbit'. [...] He [the linguist] is thereupon in a position to accumulate inductive evidence for translating 'Gavagai' as the sentence 'Rabbit'. The general law for which he is assembling instances is roughly that the native will assent to 'Gavagai?' under just those stimulations under which we, if asked, would assent to 'Rabbit?'; and correspondingly for dissent.

Then the long discussion about "stimulus meaning" follows, ending in the well-know example:

For, consider 'gavagai'. Who knows but what the objects to which this term applies are not rabbits after all, but mere stages, or brief temporal segments, of rabbits? In either event the stimulus situations that prompt assent to 'Gavagai' would be the same as for 'Rabbit'. Or perhaps the objects to which 'gavagai' applies are all and sundry undetached parts of rabbits; again the stimulus meaning would register no difference. When from the sameness of stimulus meanings of 'Gavagai' and 'Rabbit' the linguist leaps to the conclusion that a gavagai is a whole enduring rabbit, he is just taking for granted that the native is enough like us to have a brief general term for rabbits and no brief general term for rabbit stages or parts.

See Dummett's comment (page 362): "As is well known, Quine's general thesis about translation is that there are certain criteria of correct translation whose application can be determined by observation of the linguistic behavior of the speakers of the language, but that, within the limits imposed by these criteria, many different translation schemes may be possible between which there is no objective criterion which will select any one as alone correct."

In conclusion: is this scenario plausible? is it useful?

Dummett's objection seems to be: Quine's theory is not a "full" theory of meaning, but just a presentation "admittedly in metaphorical form, [of] a model of language (page 362)."

See page 374: "Indeterminacy of meaning does not entail indeterminacy of translation: if an expression of L is indeterminate in meaning, then the only strictly correct translation of it into M will be by an expression which has an exactly matching indeterminacy. Of course, M may contain no such expression: but Quine and his supporters have insisted that the indeterminacy thesis is not a mere assertion of the banal observation that exact translation may sometimes be impossible".

And see page 395:

I have tried to show that, unless reinforced by the inextricability thesis, the indeterminacy thesis is quite implausible.

But it is useful to consider SEP's entry on Quine and Indeterminacy of Translation: "we need to distinguish two kinds of indeterminacy. Quine introduces the general idea of indeterminacy, in Chapter Two of Word and Object (1960), without explicitly distinguishing the two, but subsequently comes to treat them quite differently. The first is indeterminacy of reference: some sentences can be translated in more than one way, and the various versions differ in the reference that they attribute to parts of the sentence, but not in the overall net import that they attribute to the sentence as a whole. "

This seems not to be the case discussed above, because it does not affect the "overall net import that they attribute to the sentence as a whole" that must include truth value.

"The second kind of indeterminacy, which Quine sometimes refers to as holophrastic indeterminacy, is another matter. Here the claim is that there is more than one correct method of translating sentences where the two translations of a given sentence differ not merely in the meanings attributed to the sub-sentential parts of speech but also in the net import of the whole sentence. [...] This claim involves the whole language, so there are no examples, except perhaps of an exceedingly artificial kind. There is also nothing resembling a proof; in some late works, indeed, Quine refers to it as a “conjecture”."

This second kind is the one discussed by Dummett.

And see page 379-on, where the discussion shifts from "general" language to so-called theoretical terms: "It may be replied that there is no need of any elaborate argumentation to establish that an indeterminacy of translation must reflect an indeterminacy of socially agreed meaning. For, the reply may run, it is only theoretical sentences that are indeterminate with respect to translation: and the whole argument for the indeterminacy thesis rests on the premiss that even the totality of all true observation statements leaves theory underdetermined (where a statement is an ordered triple of a sentence, a place and a time, and is true if an observer would have assented to it had he been at that place and time). Hence it is assumed from the outset that, for any theoretical sentence, there can be no means agreed on by all speakers of recognising it as conclusively established."

  • 1
    I'm always amused by how American English has words from other languages that take on very specific meanings, which they didn't have in the original language. For example, 'shampoo' is soap for washing hair, but originally just meant 'soap'. Strangely, most of these words have to do with personal hygiene: toilet, tampon, douche... English adds words to specify more closely, other languages seem ok with misinterpretations I guess. German just splices words together, which is interesting. There are probably German words that become entire English sentences.
    – Scott Rowe
    Oct 27, 2023 at 10:58
  • 1
    I understand how two different schema can translate 'Gavagai' once into 'Rabbit' and once into 'Rabbit's Parts'. However, if there is a sentence in L involving the word 'Gavagai' from which the native dissents, translator A dissents, but translator B assents (which is the sentence Dummett assumes exists), then it's clear that translation B is not good enough. Perhaps it translates 'Gavagai' itself correctly, but then some other parts of the sentence are not translated correctly. A difference in truth value necessarily implies a difference in dispositions, and that is my problem.
    – Amit Hagin
    Oct 27, 2023 at 11:14
  • @AmitHagin - agreed, and this can be IMO the gist of Dummett's objection: indeterminacy of translation is only indeterminacy of meaning. See pag.366: "the situation thus described must be one into which native speakers of L can also get. That is to say, the thesis of indeterminacy of translation constitutes merely a picturesque way of expressing a thesis that could be stated by reference to the language L alone." A fact is certain: human communication is full of misunderstanding. Oct 27, 2023 at 11:25
  • 1
    @MauroALLEGRANZA, but then what is the point of the whole argument? if out point is as clear as that, didn't Quine see it by himself? it seems like Dummett is attacking an idea, which to begin with nobody holds...
    – Amit Hagin
    Oct 29, 2023 at 8:25

You don't need to consider translation to see the underlying problem. Even among native speakers of the same language there is no guarantee that what is uttered by one person conveys its intended meaning when it is heard by another.

Just as we cannot be sure that your experience of 'yellow' say, or the smell of a rose is identical to mine, likewise we cannot be certain that your understanding of certain words is identical to mine. With words that represent everyday objects, such as 'dog', I can be reasonably confident that when I say there was a stray dog in my yard, you will not suppose I am talking about a cat, or an elephant. However, my mental image of the stray dog might be quite unlike yours. With more abstract words, the inherent problem is that the only way in which I can explain the nuances I associate with a word is to describe them with other words which don't have the same nuances. It is rather like trying to explain a particular shade of green in terms of other shades of green. If you would like to think of it as a Venn diagram, you cannot always find a set of words whose meaning exactly overlays the meaning of another word without gaps or potentially misleading differences.

What I have said so far applies to individual words- the scope for misunderstanding increases enormously when words are formed into sentences or longer prose. On Philosophy SE you will find many cases in which an answer receives comments from members of the site who have clearly misunderstood its intended meaning. It seems evident, therefore, that a passage of text may have conflicting truth values, since the meaning ascribed to it might vary from person to person even the they are all native speakers of the language in which it is expressed.

You can then introduce another level of complexity by considering the case in which a passage of text written in one language is read by someone who is not a native speaker of it, and clearly the scope for confusion increases accordingly.

Returning to Quine, there are some obvious criticisms that can be made of his approach. We don't acquire our native tongue by reading a dictionary or asking people to spell out what words mean. His rabbit example is facile. If a child, having studied their picture book, is still uncertain whether 'rabbit' refers to the whole animal depicted on the page or some un-disassembled part of it, their linguistic doubts will come to be resolved when they have seen the word used in a variety of contexts.

  • 1
    I never thought two different meanings can be assign to one piece of language. I claimed that if the two translations are good, they might indeed give two different meanings to one sentence, but they can't imply two different behaviors regarding the same sentence. In your Qualia example, the meaning might differ, but the behaviors will be invariant. If there is a behavioral difference, then there will be a way to know which translation is correct and which is incorrect.
    – Amit Hagin
    Oct 30, 2023 at 11:13

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .