Indeterminacy of Translation as a Hermeneutic Doctrine
Lorenzo Peña
Foot note 1
publ. in Hermeneutics and the Tradition
ed. by Daniel O. Dahlstrom
Washington: American Catholic Philosophical Association, 1988
pp. 212-24. ISBN 0-918090-22-9
Quine's Indeterminacy of Translation thesis is grounded in language-verificationism -- the claim that language contains no unverifiable relation. Quine's thesis issues in a hermeneutic doctrine to the effect that exegetical disputes are empty, any construal's correctness depending on the choice of a translation manual alone. Gadamer, on the other hand, maintains that a sense is bestowed upon a text by the interpreter's understanding horizon, the comprehension thus secured amounting to a merging of horizons. Translation, according to Gadamer, is to be conceived as a recreation (Nachbildung). While there are differences between Quine's and Gadamer' s views, I emphasize their deep agreements. They both are in the end committed to embrace sense-relativism and so to regard clashing interpretations as correct upon adequate choices. That relativism is found fault with; for it entails that philosophical deliberation is futile, insomuch as a philosophical system can be construed in such a way as to be equivalent to another, when what was going on was pondering on which of them, if any, was true. The paper concludes by rejecting the assumptions Quine's and Gadamer's hermeneutic doctrines of sense-indeterminacy are based on.


  1. Introductory Remarks
  2. The Main Reason for ITT
  3. The Indeterminacy of Reference
  4. The Nature of Semantic Relativity
  5. The Problem of Settling upon a Text's Real Sense in Contemporary Hermeneutics
  6. Gadamer and Quine's Views on Translation
  7. The Hermeneutic Difficulties of the Indeterminacy Thesis

§0.- Introductory Remarks

This paper's aim is to set up a reading of Quine's Indeterminacy of Translation Thesis (ITT henceforth) which shows its significance as a hermeneutical principle. My conclusion will be that Quine encounters hermeneutical difficulties akin to the ones which be set hermeneutical methodologies put forward by Gadamer, Coreth and other people working within the continental tradition.

§1.- The Main Reason for ITT

The ITT says that, once two languages, L and L', are given, for any sentence of L, p, unless p belongs to a restricted set of sentences which is carefully marked off, there are alternative ways of giving respective sentences of L' as translations of p, each of those ways being correct, even though such translations may well fail to agree in truth-value, let alone in «meaning». What does `translation' mean in such a context? Only this: a sentence, p', in L' is a (right) translation of a sentence, p, in L iff it is right for a speaker of L' to report an utterance of p by a speaker of L by asserting that he has said that p'.

Thus, while Quine is sometimes taken to reject all intensional notions and esp. indirect quotation or «believing that», in general, what indeed seems to be his own opinion is that only unrelativized «believing that» is to be waived. For, since there are, for a sentence p of L, several correct ways of translating it into L' and thus several nonequivalent ways of assigning beliefs to an utterer of p -- assuming him to be sincere -- , not only do we lack any ground for relinquishing all concept of «believing that» and «asserting that», but, what is more, the very existence of those alternative ways of rightly translating a sentence enjoins us to conclude that we cannot do without those very same concepts -- otherwise there would be no way of rightly reporting a man's words except through direct quotation.

Quine has over the years put forward a number of reasons for countenancing the ITT. But in a nutshell what seems to me the main reason is this one. Language is nothing else but verbal behaviour; hence there is nothing to language -- nothing to either syntactic or semantic or pragmatic relations -- which is not amenable to a study of linguistic behaviour; in other words, any linguistically relevant relation has to be liable to observation in the way patterns of linguistic behaviour are.

Now, any inquiry into linguistic behaviour is unable to settle upon one of the different ways of translating a sentence. Suggesting that even so there nevertheless is just one correct translation would amount to positing something in language beyond linguistic behaviour, something immaterial for any study of such a behaviour.

Of course you may decide that sentence p refers to a situation s, which is referred to in your own language by sentence p'; so p translates as p', any other non-equivalent translation being incorrect; you may add that those relations are not immaterial for the study of linguistic behaviour since translation is part and parcel of that behaviour.

However Quine would look upon your manoeuvre as begging the question. Somehow or other translation is not part of the most normal or typical linguistic behaviour; moreover, since what is at issue is whether or not translation is determinate, you cannot just assume that it is, since there is no reason for supposing it is ensuant upon linguistic behaviour as it takes place outside translation itself.

Should you reply that, pari passu, one could conclude that there are neither waves nor particles, since study of physical phenomena enjoins upon us neither to positing waves nor to positing particles, either postulation being handily sufficient on its own -- provided some further adjustments are made in each case -- Quine would retort that there is no parallel: physics concerns itself with physical entities and there is no reason either for saying that particles or waves are not such or for regarding the world as being indeterminate as to whether it has particles or waves -- that far-fetched indeterminacy would in fact be hard to word in a precise way.

Likewise and more generally: even though scientific theories are underdetermined as a regards available evidence, that alone is no sufficient ground for us to say that, e.g., the world is indeterminate as between theory T and theory T', in such a way though that, if p, p', are respective conjuncts of the (finite) sets of axioms of T, T', then «p' or p» should nonetheless be true. No, nothing is to be gained through such a device. For, if the situation is as just described -- underdetermination of T, T' by available evidence -- either course of choosing T alone or T' alone would be simpler, and so better, than countenancing a theory axiomatized through a disjunctive axiom with both disjuncts thought to be nontrue.

(That would amount to countenance a non-prime theory, a theory, that is, asserting a disjunction but unable to assert either of the disjuncts; which would anyway force us to jettison classical logia and embark upon some kind of supervaluation semantics or some other device of that ilk.)

ITT has nothing to do with that. The principle of excluded middle entails that either there are particles or there are not -- and, if, somehow «by definition», a world without particles would be a world containing just waves, then in virtue of excluded middle we'd have that either there are particles or there are waves.

When a disjunction is true, one of the disjuncts at least is to be true. Nothing of the sort happens concerning translation. When there are two ways of reporting someone's words, say ascribing to him either asserting that p or asserting that p', what emerges is not our concluding that he said either that p or that p'. No, what emerges is that we are to dispose of a nonrelativezed notion of «asserting that».

No application of excluded middle can force upon us espousal of «He said either that p' or that p»; not even an application helping itself to a (plausible or useful) «conceptual necessity» or «meaning postulate» like the one through which we had concluded that if there are no particles, there are waves.

For one thing, although in order to set the doctrine in the clearest and simplest way we have momentarily supposed that there were just two alternative available translations, the number may well be infinite.

For another, nothing appears in virtue of which failing to assert the value of one of those translations would entail asserting the value of another translation, even if there were no available alternative translation; nothing except the assumption that in each case the speaker asserts something definite, which is just what is at issue.

§2.- The Indeterminacy of Reference

Indeterminacy of Reference is an immediate corollary to ITT. Even though Quine has worded indeterminacy of reference as `the inscrutability of reference', that expression seems to me unfortunate pragmatically implying as it does that there is some thing which we would be right to scrutinize could we do so. If reference were determinate, translation would be, too -- even should we be unable to satisfy ourselves that this or that translation was the correct one.

Several paragraphs earlier (in §1) I have spoken as if reference were a relation between a sentence and a «situation», which of course for Quine's it is not. In fact Quine takes as primitive the relation of satisfaction. An object o satisfies a formula p iff, when o is given as a value to the first free variable in p, the formula comes out true.

(That doesn't intend to be a technical or precise definition. In fact, Tarski-wise truth is to be defined through satisfaction rather than the other way round; moreover satisfaction is explained by Quine not for an object, but for a sequence of objects. But all that is outside our point.)

In his nontechnical writings, Quine has worded his semantic doctrine through the expression of being the value of a variable. I think some misunderstandings have followed, e. g. that for Quine something is the value of a variable regardless of the truth or falsity of the formulas the variable features in. No, for Quine an object o, is the value of a variable, x, featuring free in a satisfiable formula, p, iff when we give o as a value to x p comes out true. (In other words: iff any sequence whose first component is o satisfies p provided x is the only free variable in p; adjustments can be made for the case the latter proviso does not hold.)

Of course, Quine is interested not just in categorial truth but also in hypothetical one: if you profess theory T, thus asserting any of its theorems, p, you may be said to refer to (or be committed to) object o insomuch as o alone satisfies p. (Thus even though a closed formula is satisfied by all things, you don't refer to the Statue of Liberty by saying that there are squirrels.) Which means that, if your theory is true, if p in particular is true, o satisfies p -- i. e. o is a value of the first free variable in p.

There are clouds surrounding that at first blush clear and simple notion of referring-to and of being committed to: one of them concerns the either extensional or purportedly intensional nature of such a commitment; another, the problem raised by the case of several objects shared by all models of a theory without there being any one of them uniquely referred to by the theory -- uniquely satisfying a certain theorem of the theory; and so on. Quine himself has refined his theory on ontological commitment and reference in order to cope with some of those issues.

Other authors have also endeavoured to deal with some other difficulties, as e.g. the purportedly intensional nature of ontological commitment -- to my mind unsatisfactorily, since intensionalization is a most unQuinean way of handling matters. Be it as it may, I'm not going to dwell on those issues here.

What I want to point out is that ontological commitment is for Quine relative: as relative as translation is. For the core of ITT is precisely that there is no uniquely way of assigning to expressions and formulas of a language respectively objects and truth-conditions as that which those expressions refer to and that which those sentences convey.

It is safe to say that a name n refers to an object o iff the corresponding Quinean predicate «n-izing» is such that the open formula «x n-izes» is satisfied by o. For any other predicate, f, we shall say objects of some kind are referred to by f iff at least one of those objects satisfies «fx» while nothing satisfies «fx» unless it is of the kind in question. Let's now take a very simple example.

Michel says in French `Les lapins sont jolis'. I may rightly take it that what he says is to be translated as follows: `Rabbit hood is included in prettiness', where rabbithood and prettiness are taken to be intensional attributes while inclusion is understood extensionally.

A Quinean extensionalist would maintain that what Michel says is false, since there is no rabbithood. Of course Michel's utterance might all the same be regarded as pragmatically truelike -- a notion which may set defined in an easy-going way; but that is by the way.

What now matters is that my translation of Michel's words is as good as your translation according to which he has said that rabbits are pretty, which we both deem true. What objects does Michel's expression `les lapins' refer to? According to my translation, that French expression -- at least as Michel uses it in that context-is a proper name n, whose Quinean transform would be the predicate «being identical to n»; which, upon being juxtaposed to a free variable, results in an open formula which can be satisfied by nothing else but rabbithood, if at all. (In fact, being an extensionalist, I go on to maintain that nothing does satisfy such a formula.)

Is Michel really committed to intensional attributes simply because of my deviant way of translating what he says? Well, that is just a bad question. For, if we were right on ITT, ontological commitment is relative: it is, independently of any viewpoint, neither true nor false that Michel is committed to attributes. Were ontological commitment absolute, reference, satisfaction and truth-conditions would be absolute, too; and so would be translation -- at the very least it would be determinate up to truth-conditions, which is a lot.

§3.- The Nature of Semantic Relativity

What exactly is the nature of the relativity enjoyed by reference and other semantic relations? Two candidates are to be rejected. First, the idea that the relativity -- or indeterminacy -- in question amounts to some situation which can be expressed through a conditional statement; something like this: «If manual of translation M is correct, then...».

That will not do. For one thing, in such a case translation would be determined -- up to truth conditions at least. For another, the adjective `right' in the protasis is misused, since there is no sense in which a translation manual is «correct» as an assertive sentence is, that is to say to the exclusion of alternatives: if «p» is right, «not-p» is wrong, while the «rightness» or «correctness» of a translation manual doesn't rule out rightness of alternative translation manuals.

The second candidate we ought to dismiss is the idea that the relativity in question is some kind of dependence on tacit premises, or presuppositions, or assumptions. The reasons are quite similar: a premise is something taken to be true; a translation manual is neither true nor false.

Withal, were the relativity or indeterminacy in question nothing else but correctness of alternative reasonings depending on respective sets of assumptions, translation would not be accurately said to be indeterminate -- nor would reference either, of course; for then things could be straightened out and de-relativized by setting upon the right assumption -- or anyway upon an assumption we took to be right.

What seems to me to constitute Quine's view on the relativity of reference could be briefly articulated like this. A speaker is rightly taken to refer to some object, or to convey through his statements some particular truth-conditions, only relatively to two chosen parameters: a language wherein the indirect quotation is made, and a translation manual

Let's now ignore or disregard the former. As for translation manuals, they are stipulated, la id down, not asserted. A translation manual is not a set of sentences of the form «`e' refers to o» or «The truth-conditions of «p» are that so and so be the case», or anything of that sort. Such sentences involve the by now junked notion of determinate, nonrelative reference and truth-conditions. No, a translation manual is rather a set of imperative statements like that: `Let «e» go into «o»', and so on. A translation manual is a set of rules. (Of course you can view rules as functions rather than either injunctions or allowances.)

Accordingly, all semantic relations are, according to Quine, indeterminate or relativized in the way just pointed out: no semantic relation obtains independently of a choice; the choice in question is that of both a target-language and translation manual Interpreting a text is nothing else but translating it from one language into another -- even, of course, if those languages happen to be the same.

So all discussions concerning how to interpret an author are misguided: provided a very few requirements are met, any such interpretation is right -- right, that is, depending on the choice of a translation manual You can go on debating about whether Plato said or «intended to say» this or that. But the dispute is empty.

The same thing applies to Quine himself of course. Quine may try to «clarify» his views by laying down that some of his own former utterances are to be read this way or the other; but anyhow what he is thus doing is nothing else but choose some particular translation manual (most times a non-homographic one), which is neither better nor worse than any other. No one has the privilege of being alone entitled to prescribe translation manuals, even when his own writings are concerned. For else translation and reference would, or might, be determinate.

§4.- The Problem of Settling upon a Text's Real Sense in Contemporary Hermeneutics

It seems to me apposite to try and apply Quine's ITT to some hermeneutic problems as they have emerged in historic-philosophical scholarship. New analytic approaches have be en recently set up and developed to a great many philosophical texts, esp. to those of ancient philosophers.

Analytic readings of the Presocratics, of Plato, Aristotle and so on have by now acquired wide currency. That has given rise to some sharp reactions. It seems clear to some people that the new interpretations make violence to the texts under consideration, by having them address questions their authors never brought up and by wording their claims in terms their authors cannot even have conceived of, as such terms are fraught with «connotations» borrowed from contemporary philosophy.

On the one hand, (almost) everyone agrees that no interpretation of a philosophical text is (philosophically at least) interesting unless it shows:

  1. the relevance of what the author had to say -- and, we hope, somehow managed to say in the thus interpreted text -- to problems which are not peculiar to him or his times;

  2. that the thus interpreted doc trine can be understood as an available -- and, at least at first blush, or up to a point, sensible -- alternative to other approaches to the same issues. (The latter demand is perhaps too strong. It would be more cautions to be content with the thus interpreted doctrine's being possessed of something or other to recommend it, some kind of buttressing evidence or argument.)

On the other hand, though, some boundary is to be drawn, since what surely cannot be indulged in is a boundless application in historical scholarship of conceptual frameworks quite foreign to the authors we try to read.

Some interpretersFoot note 2 emphasize the austerity principle: they insist that you must forbear attributing to the philosopher you read assumptions such that there is no special reason to think they were obvious to him, or inferences he did not make in so many words, or commitment to entities which he did not explicitly posit. As Gluck puts it (op. cit., p.395) you ought to read the texts instead of reading into them.

The opposite trend is represented by people who, like Jonathan Barnes or David Charles,Foot note 3 try to assess the philosophical significance of the interpreted philosophers' claims by comparing them with solutions put forward today, thus preferring to stress constraints like explanatory coherence or reading each text in the light of others by the same author.

Now, while working with the latter kind of approach broadly construed, H.N. Castañeda, in his interpretations of Plato, Leibniz and other past philosophers, has nevertheless la id emphasis on hermeneutic requirements the most striking of which is that every text is to be read and understood on its own; what he demands is a Darwinian method, a piece-meal approach, taking each text as a separate exegetical unit; while Castañeda recognizes it is difficult to determine what is to count as a text-unit in each case, he thinks that such a kind of approach is all in all better.

Let us now consult one of the greatest hermeneutic thinkers, - Hans-Georg Gadamer. Like Emerich Coreth and Paul Ricoeur, he insists that our readings or construals are unavoidably conditioned by our fore-understanding horizon, i. e. by our set of concepts, problems, assumptions, which it would be futile to try sidestepping. While they all preach to strenuously attempt to catch the sense or senses of what the philosophers said and somehow or other plunge into their own understanding horizon, they nevertheless warn us that it would be of no use to strip ourselves of our conceptual framework, since, were it possible, it would thwart all our own efforts to grasp a sense at all.Foot note 4

Gadamer has therefore stressed the positive role of fore-judgments. Doubtless, he also sets up as an hermeneutic goal what he calls a fusion or blending of horizons -- the interpreter's and the interpreted one's. But such a blending -- which Coreth and others have deemed an impossible target to attain and one which would not constitute real understanding -- not only does not rule out the persistence of our own understanding horizon, but even it rules it in, it calls for that persistence.

Gadamer hence says (ibid. p.374) that there can be no interpretation which is right in or by itself: every interpretation amounting to, or at least supervening upon, a blending of horizons, each of the two thus merged horizons is to persist up to the point they become one. It is then out of the question to approach the other's horizon from nowhere, so to speak.

§5.- Gadamer and Quine's Views on Translation

Gadamer has shown how all the hermeneutic enterprise is mediated by language. Every translation is an interpretation.Foot note 5 A translation is the fulfilment of a sense that the translator bestows upon the text offered to him.

Granted, Gadamer says (ibid.) that the hermeneutic problem proper is outside translation, since he thinks that when people speak with one another they do not translate: so the hermeneutic problem is not that of attaining a good mastering of a language, but that of reaching an agreement about the thing which is implemented or carried out by means of language.

Through linguistic intercourse what is at issue is an understanding not so much of what the other has said as of him who says it: we try to set or catch his meaning (Meinung), which does not boil down to a particular sense of his words. However, such comprehension is a recreation (Nachbildung): we can only reach the philosopher's meaning -- what he me and -- by approaching the sense or senses of his words. Now, Gadamer recognizes (pp. 363 ff) that in practice things are somehow more complicated: for, even if I understand the other's language, I cannot reach the blending of our two horizons if my own conceptual framework, minted as it is in my language, remains outside my dealings with his words and his problems.

Thus, I am at the end compelled to fall back on translation. and every translation proceeds by overhighlighting überhellung). This is why the translation's output is more clear than its input, a distance the interpreter is painfully aware of.

Let us compare Gadamer's claims with Quine's. Both stress that translating entails interpreting and even that it is nothing else but interpreting. Both reject the idea that the translated text's sense is something independent of the translator's approach (cf. Gadamer, p.448).

So, both of them recognize some kind or other of translation indeterminacy. They both denounce neutrality as an illusion: it is not by, so to say, approaching a text without tools taken from your own gear that «the real sense» of the text is going to make itself manifest to you.

Beyond those agreements, though, there are no less important discrepancies. For Gadamer a text has many senses, it is full of senses, and its senses somehow are combined or heaped up on it. For Quine a text's sense is unique -- except in cases of plurivocity, which for him lie outside what is now at issue -- once a language target and a translation manual have been chosen.

(That the translation's output is in turn indeterminate as regards its own sense is a quite different matter: the given text's sense is relative neither to its output, nor to the output's own sense or senses, but to (the choice of)the tools used in reaching that output.)

For Gadamer there is some case of comprehension in which no translation is needed -- indeed all translation is ruled out at that case, since what is there going on is direct understanding through conversation in the same language. For Quine all understanding is translation, no difference of principle separating out the communication through what at first blush appears to be one same language from any other communication: for the claim that both communicants mean the same by the same words is, not debatable, but in fact a meaningless claim, ignoring as it does the ITT: that you mean the same as I do by saying something is neither true nor false unless and until a translation manual has be en given, and then it is either true or false as regards that translation manual

For Gadamer there is some further entity, the Meinung of the text's author, which is to be grasped through the catching of a sense of the text. Nothing of the sort is posited by Quine. What is more, Quine's philosophical principles would indeed rule out any such entity at all.

So Quine would have nothing to do with Gadamer's understanding someone beyond -- even though by way of -- understanding what he says. Furthermore, Quine cannot be prepared to accept a fusion or a welding of horizons: the very notion would sound odd to him, to say the least; and he would stress that, if by an understanding horizon we mean a conceptual framework comprising both a language -- which can be thought to include a set of statements which are «analytic» in the weak sense that they have be en learnt by learning the language -- and a translation manual, then understanding horizons cannot merge.

A conceptual framework is made to correspond to another through translation manuals, but cannot be blended either with or into it: it cannot do it outside translation -- since there is no other way for two such frameworks to become counterparts of one another; and it cannot do it by means of translation either, since translation does not impinge upon the two systems' respective identity -- quite on the contrary, the very reality of translation calls for each of those systems to be there (remember that, even when the source-language and the target-language are, as we would normally say, one and the same, their real identity, which requires meaning identity of each word, is neither true nor false on its own, but only true or false upon some particular translation manual or other).

Finally, Quine cannot accept Gadamer's idea of an agreement on the thing itself concerned by the ongoing communication. For, according to Quine, there is no unique or determined relation of concerning or referring to. Indeterminacy of translation carries in its wake indeterminacy of reference, of ontological commitment and of ontology itself (the thesis of ontological relativity). Gadamer seems to think instead that, even though sense is not uniquely determined, the semantic relation of referring to or being about is.

On that point, Quine seems to me more coherent, since it is hard to understand how the being-about relation can be determined if sense is not: should one message have different senses for different translation manuals -- which of course Gadamer somehow seems committed to but never recognizes explicitly in so many words -- there would, or at least might, be as many things concerned by the message; unless a postulate is la id down to the effect that all the different senses a message may have are «about» the same thing or things -- an extremely implausible postulate unlikely to be justified in any reasonable way.

§6.- The Hermeneutic Difficulties of the Indeterminacy Thesis

In spite of differences, Quine's and Gadamer's hermeneutic doctrines share, as we saw in the previous section, a number of points. This is why they face similar difficulties.

The main blemish those doctrines labour under is paradoxically ensuant upon an important advantage they have over such as countenance sense determinacy. Remember the unmanageable difficulty brought up above, in §4: to what extent are we entitled to read a text by somehow reading into it -- i. e. by reading it through our conceptual glasses?

Were such a procedure warranted, the path would be open to unbridled arbitrariness and wantonly subjectivistic interpretation. But else, we turn out to be deprived of something we cannot do without; for it is me who wants to understand, which I cannot do if my act of purportedly understanding is cut off from everything I normally think, if it has to take place through terms bearing no relationship to the ones I am acquainted with.

Such a problem seems to find a Solomonic, easy-going solution by means of Quine's ITT. Quine would tell us that there is no fact of the matter over whether, say, Plato did or did not accept states of affairs in his ontology. He accepted them upon some translation manuals, while he didn't upon other translation manuals.

Is that the end of the matter? Perhaps not. We could discuss respective merits of those translation manuals. Not the merit of correctness, which they all have anyway, but virtues of simplicity, elegance, compliance with some useful stipulations -- useful in order to so arrive at translation outputs fitting more nicely into our accepted history of philosophy schemes.

(Considerations of such a kind have be en taken advantage of and knitted into some objections levelled against the ITT: after all some translations seem in the end to be more plausible than others, more in accordance with the remainder of our whole linguistic behaviour, don't they? Quine could reply that, even if there is such a gradation, even the worst-off translation manuals are utterly honourable ones -- or, put in less colourful terms, perfectly compatible after all with linguistic behaviour as a whole.)

What is more, Quine might even look upon such hermeneutic disputes as confirming evidence accruing to his own case for ITT. Either way you regard things, there is no clinching argument for or against a construal of Plato which avails itself of currently fashionable concepts, or terms, and addresses issues that (or as) are being debated now a days. He would add that language contains no unverifiable relation, since language is just linguistic communication, linguistic behaviour -- and something unverifiable about such a behaviour sounds like a contradictio in adjecto.

Therefore, Quine would conclude, the very unsolvableness of hermeneutic disputes shows that they are basically empty -- even if they can go on once they have been duly modified: do not discuss over what Plato really said but over what translation manual is nicer.

Gadamer would be neither so definite nor, I am afraid, so clear as Quine over those hermeneutic controversies. Exegetical disputes are not so obviously rejected by his hermeneutic doctrine as they are by Quine's. Nevertheless, I take him to be committed in the end to conclusions quite similar to Quine's. Since a text's senses are never exhausted, since they keep on being constituted over the time and, far from existing on their own, or by themselves, acquire existence somehow by their being bestowed upon the original text by the interpreter's Nachbildungen, what we have in the last resort is nothing else but an indeterminacy situation very much like Quine's.

No unique sense is to be gathered from Plato's writings independently of our respective hermeneutical tools and keys. So, it turns out to be fine and O.K. for you to say that Plato countenanced states of affairs, but not for me. We thus leave each other in peace. Henceforth we can keep on debating, but more friendly. After all it is no longer a question of one of us being right and the other wrong. It is a question of finding out which way of recreating Plato's texts is, e.g., more suggestive, or more interesting.

Now, what might appear to be an advantage turns out to be a redhibitory flaw. Those doctrines cannot account for exegetical disputes in any straightforward way. As a consequence, philosophical debate also becomes empty.

Suppose I am mulling over the acceptance of relations as irreducible entities. Hitherto I had not admit ted them in my ontology. Now, while pondering upon the pros and cons of such an admission, I suddenly realize how futile my deliberation is: I tell myself that my revamped system positing relations would be translatable into my old one lacking them. There is always some device or other to be fallen back on for facilitating such a translation.

(After all it is possible to «reduce» all entities to natural numbers using the Löwenheim-Skolem theorem -- and so it is possible to convince oneself that one has always be en and will always be a Pythagorean, even if, as one's system is worded, it superficially countenances non-numerical entities.)

Quine has developed sharp techniques of ontological reduction (the proxy functions). and he has, as is well-known, devised some restrictions to their application, in order to shun Pythagoreanism. Well and good. But what our current line of argument seems to show is that such daintiness is misplaced.

There is no fact of the matter over whether or not one is Pythagorean, even for oneself. All depends on the translation manual chosen. If right now I am a convinced Pythagorean I can satisfy myself that it is immaterial for me to change my mind on the subject or not: either way, my future worldview can be translated into the one I have so far held on to, both as being in agreement and, alternatively -- by choosing another translation manual -- , as disagreeing on the point I was considering.

Even though it is not as easy to argue that Gadamer's theory entails results similar to those we have just mentioned -- Gadamer's approach is after all not so clear-cut and crisp as Quine's -- , it is almost obvious, yet, that much the same applies to his doctrine.

I take those unsavoury conclusions to constitute refutations of those doctrines. It may be hard to put up with unverifiable entities, either in general or in linguistic behaviour. But we are compelled to take then up, willy nilly. We may be at a loss as to how to read Plato, and may well remain forever unable to find conclusive evidence on either side of many an exegetical dispute.

However sense cannot be indeterminate. Perhaps there are sundry senses to one text, but what certainly cannot be the case is for those senses to depend on, and vary as, the reader's hermeneutic tools or keys. Semantic relations are determinate.

[Foot Note 1]

«Indeterminacy of Translation as a Hermeneutic Doctrine» was delivered to the Annual Congress of the American Catholic Philosophical Association held at Louisville (Kentucky) in the spring of 1988, thanks to a travel grant received from the Spain-US joint committee for Educational and Cultural Cooperation, to which I hereby express my appreciation.

[Foot Note 2]

E.g. Richard Robinson in his Plato' s Earlier Dialectic, Oxford, 1983; or John Glucker in his critical study of Oxford Studies in Ancient Philosophy, appeared in Philosophia, vol. 16 (1986), pp. 389ff.

[Foot Note 3]

See e.g. the latter's Aristotle's Philosophy of Action, Duckworth, 1984.

[Foot Note 4]

See Wahrheit und Methode, 2d. ed., p. 373.

[Foot Note 5]

Ibid., p.362: the German term he uses is Auslegung.