Lost in Translation?

Abstract

According to neo-Russellianism, in a sentence such as John believes that Mont Blanc is 4000 m high, any other proper name co-referring with Mont Blanc can be substituted for it without any change in the proposition expressed. Prima facie, our practice of translation shows that this cannot be correct. We will then show that neo-Russellians have a way out of this problem, which consists in holding that actual translations are not (merely) a matter of semantics, but also make an attempt at preserving some pragmatic features of the sentences to be translated. We then turn to translations that only preserve the semantic characteristics of the sentences and we argue that, although these translations are unable to show that neo-Russellianism is incorrect, they still show that it relies on some theses that seem to have no justification, apart from saving neo-Russellianism itself.

It is not clear that the notion of translation is amenable to any rigorous treatment, in the sense that necessary and sufficient conditions could be stated for a sentence in some language to amount to a translation of another sentence in a different language. Nonetheless, it seems possible to rely on translation in order to draw some conclusions pertaining to semantics. In fact, in many authors translation has played a role in establishing what truth and meaning are in general, or what the meaning is of some particular bit of language.

In this paper, we will focus on proper names as they occur in the context of propositional attitude predicates, as is the case with a sentence like

• (1) John believes that Mont Blanc is 4000 m high,

and on something current translations seem able to tell us about them. According to Russell, and to the many defenders of what are nowadays called Russellian propositions, (1) expresses something that has as a component neither a private idea of Mont Blanc, nor a sense containing a mode of presentation of the mountain, but the mountain itself, that object out there in reality. Neo-Russellians are then Millian about proper names: what the name “Mont Blanc” contributes to what (1) expresses is not mediated by a sense or anything else and is simply its referent. Thus, according to neo-Russellians, any other proper name co-referring with “Mont Blanc” can be substituted for it without the Russellian proposition expressed changing in the least. Prima facie, as we will point out, our practice of translation seems to show that Millianism cannot be correct. We will then show that Millians have a way out of this problem, which consists in holding that actual translations are not (merely) a matter of semantics, but also make an attempt at preserving some pragmatic features of the sentences to be translated. We then turn to translations that only preserve the semantic characteristics of the sentences and we argue that although these translations are not able to show that Millianism is incorrect, they still show that it relies on some theses and distinctions that seem to have no justification, apart from saving Millianism itself.

1 Introduction

It is no easy matter to say what translation is. It is not even clear that the notion is amenable to any rigorous treatment. How are we to settle, e.g., whether the mapping provided by Google.Translate.com from and into a number of languages satisfies the minimal requirements to count as a translation at all—possibly a second rate one, but a translation nevertheless? It seems that very few generalizations concerning translation are entirely safe. One thing that seems almost indisputable is that two declarative sentences related by way of translation ought to share their truth-value. This is Saul Kripke’s Principle of Translation: “if a sentence of one language expresses a truth in that language, then any translation of it into any other language also expresses a truth (in that other language)” (1979, 139). However, even this principle might not hold strictly and universally, even though it appears to hold good in a large majority of cases.¹ In addition to their truth-values, prima facie it appears that other features of the original sentence and its components ought to be preserved in translation, such as reference and linguistic meaning. However, it has been shown long ago that principled reasons exist that make this impossible in general. Willard Quine pointed out that, for example, some mass nouns in some language can only be translated into other languages by using count nouns, which differ from the former in reference (1960, 90–95). Also, it was shown by William Hart (1970) and Tyler Burge (1978a), concerning self-referential sentences, that translation can preserve either reference or self-reference, but not both. Taking as an example the self-referential Italian sentence

• (2) Questo enunciato non è un teorema,

we can translate it into English either as

• (3) (2) is not a theorem,

which preserves the references, but is not self-referential, or as

• (4) This sentence is not a theorem,

which preserves the self-referential aspect of (2), but in which reference is made to (4), and not to (2). If the self-referential sentence occurs within an informal exposition of Gödel’s incompleteness results, since self-reference is crucial for the proof, it is clear that in this particular case self-reference could take precedence over reference.

Self-referential sentences seem to show something else about translation. Is it really as clear as is generally assumed, that a satisfactory translation can be given for every sentence of a language into any other language? Gottlob Frege distinguished various components in indicative sentences: the thought, the assertion, and a third component, over which the assertion does not extend, and that is “[w]hat is called mood, fragrance, illumination in a poem, what is portrayed by cadence and rhythm” (1956, 295). He then raised doubts on the translatability of poetry because these aspects of the third component “make the translation of poetry very difficult, even make a complete translation almost always impossible, for it is in precisely that in which poetic value largely consists that languages differ most” (ibid.). Perhaps it is not just poetry that is not translatable. Even though no general and fully-fledged theory of translation exists, as far as we are aware, it is widely held that the business of translation is to preserve the thought, in Fregean terminology, or the proposition expressed. It is also widely assumed that propositions are language-independent. Stephen Schiffer, for instance, puts it as follows:

[A proposition] is mind- and language-independent in two senses. First, its existence is independent of the existence of thinkers or speakers. That eating carrots improves eyesight was not brought into existence by anything anyone said or thought. Second, that eating carrots improves eyesight can be expressed by a sentence of just about any natural language but itself belongs to no language. (2011, 535)

Being language-independent, propositions ought to be expressible in any expressively adequate language whatsoever. But self-referential sentences give us reason to doubt that this is in general the case. Since thoughts are composed of the senses of sentential components, and senses uniquely determine referents, it clearly follows from the received view that referents are to be preserved. On the other hand, self-reference is crucial for Gödel’s incompleteness results and seems to be part of the thought expressed by

• (2) Questo enunciato non è un teorema.

But then, since there is no sentence in English (or in any other language) that can preserve both the references made in (2) and its self-referential aspect, it seems that we should conclude that the thought express by (2) is tied to Italian.

Another prima facie reason for looking askance at the view that translation can, and must, preserve thoughts, comes from Kripke’s puzzles about belief. On the received view, the thought that London is pretty is expressed both by the English sentence

• (5) London is pretty

and by the French

• (6) Londres est jolie.

If any two sentences exist that are related by way of translation, (5) and (6) are such. But it is far from clear that this can be so. Suppose, with Kripke, that Pierre is a competent speaker in both French and English but that he learnt English by the direct method, without using any translation of English into French (Kripke 1979, 143–144). Pierre seems to believe that London is pretty when he is taken as a competent speaker of French, and also to believe its negation, when he is taken as a competent speaker of English. The same can happen to any pair of sentences belonging to different languages that are related by way of translation to one another, and to any subject who is competent in both languages without being bilingual, i.e. without realizing that either sentence translates the other (which is perfectly compatible with her twofold competence).²

Are we to conclude from self-referential sentences and from the case of Pierre either that thoughts are not language-independent, or that they are not preserved in translation, or that there are sentences in one language that cannot be translated into another? Not necessarily. For example, one might perhaps hold that self-reference, despite appearances, does not really concern the thoughts expressed. Nonetheless, surely these cases show that translation is not an easy matter. Preservation of the thought, moreover, if it is a condition on translation at all, is clearly not a sufficient one. If we are to believe Frege (1956, 294) and many after him, the interrogative sentence

• (7) Is London pretty?

and the corresponding indicative one, i.e. (5), express the same thought. But it would be absurd to claim that (7) might be translated into French as the indicative sentence (6), even though they allegedly share the same thought.

From all these considerations it seems that we should doubt that in general necessary and sufficient conditions could be stated for a sentence in some language to amount to a translation of another sentence into a different language. Translating might well be a matter of negotiation, without any solution in sight that clearly is the only correct one or even the best possible one, if there is one at all. But even though the notion of translation is hardly amenable to rigorous treatment, this does not necessarily preclude it from being a useful tool in assessing some theses that pertain to semantics, and in fact it looms large in philosophy of language. Tyler Burge (1978a), Alonzo Church (1950), Donald Davidson (2001, 17–36; 93–108; 125–140), Michael Dummett (1981, 371–381), Frege (1956), David Kaplan (2004, §3), Kripke (1979), Quine (1960, 26–79; 211–216) and Alfred Tarski (1983), to name but a few, have employed the notion of translation, and what they intuitively took correct translation to be, in drawing some of their conclusions pertaining to semantics. In many of these and other authors translation has played a role in establishing what truth and meaning are in general, or what the meaning is of some particular bits of language like proper names and clauses occurring in the context of propositional attitude predicates, such as the clause “that Mont Blanc is 4000 m high” occurring in

• (1) John believes that Mont Blanc is 4000 m high.

In the rest of this paper, we will focus on these particular bits of language and on something translation seems able to tell us about them.

As the label suggests, propositional attitude sentences such as (1) are generally taken to express the holding a relation between a subject and a proposition. In the particular case of (1) the received view is that it expresses that John is related to the proposition that Mont Blanc is 4000 m high. But what is this proposition? Quite famously, Bertrand Russell maintained that “in spite of all its snowfields Mont Blanc itself is a component part of what is actually asserted in the proposition ‘Mont Blanc is more than 4000 m high’” (1904, 169). Thus according to Russell, and to the many defenders of what are nowadays called Russellian propositions, (1) expresses something that has as a component neither a private idea of the mountain, nor a sense containing a mode of presentation of the mountain, but the mountain itself with all its snowfields. Neo-Russellians are then Millian about proper names: what the name “Mont Blanc” contributes to what (1) expresses is simply its referent. According to neo-Russellianism, as we will discuss it here, then, any other proper name co-referring with “Mont Blanc” can be substituted for it without the Russellian proposition expressed changing in the least.³Prima facie, as we will point out, our practice of translation seems to show that Millianism cannot be correct (§2). We will then indicate how neo-Russellians can react to this problem. In a nutshell, they can say that translation is not (merely) a matter of semantics, but (also) of pragmatics (§3). We will then turn to translations that only preserve the semantic characteristics of the sentences to be translated and we will claim that, although these translations cannot show that Millianism is not the correct semantic account of proper names, they still show that Millianism relies on some theses and distinctions which do not seem to have a rationale, apart from that of saving Millianism itself (§4). The general conclusion will be that, however obscure the notion of translation might be, it might still be a useful tool in assessing semantic theories.

2 Practical Canons of Apt Translation: Problems for neo-Russellianism?

In order to see that, on the assumption that a correct translation preserves the proposition expressed, neo-Russellianism is at odds with our current practice of translation, let us take the following passage from Frege:

• (8) der Gedanke des Satzes “der Morgenstern ist ein von der Sonne beleuchteter Körper” verschieden von dem des Satzes “der Abendstern ist ein von der Sonne beleuchteter Körper”. Jemand, der nicht wüsste, dass der Abendstern der Morgenstern ist, könnte den einen Gedanken für wahr, den anderen für falsch halten. (1892, 32)

Forget for the time being the first sentence in which two quoted sentences occur, and let us focus instead on the second sentence. In it, two different co-referential names are used, i.e. “Abendstern” and “Morgenstern”.⁴ Quite predictably, also in the many translations of (8), two different names for Venus are used. All translations in English, for example, go as follows:

• (9) the thought in the sentence “The morning star is a body illuminated by the Sun” differs from that in the sentence “The evening star is a body illuminated by the Sun.” Anybody who did not know that the evening star is the morning star might hold the one thought to be true, the other false. (1984, 162; 1997, 156)

It is pretty obvious why all translators opted for this strategy: an English speaker reading

• (10) the thought in the sentence “The morning star is a body illuminated by the Sun” differs from that in the sentence “The evening star is a body illuminated by the Sun.” Anybody who did not know that the evening star is the evening star might hold the one thought to be true, the other false

instead of (9) would simply be unable to understand Frege’s point. We can moreover note that surely Black, Beaney and all other translators in the various languages in which Frege’s piece has been translated knew that the morning star is the evening star. Furthermore, it does not really make any difference whether they agreed with Frege or not. Even a neo-Russellian who, as such, endorses Millianism, would use two different names.

Millianism is the thesis that co-referential proper names are everywhere interchangeable not only salva veritate but even salva significatione: as Kripke phrases it, according to Millians, “the proposition expressed by a sentence should remain the same no matter what name of the object it uses.” (1979, 127) Thus, at least prima facie, it seems that according to neo-Russellianism it should make no difference whether we use two occurrences of “the evening star”, or occurrences of two different names. But it does seem to make a difference. Why, then, are we regularly using two different names?

One might think at this point that this is in a way a special example, from which nothing general can be concluded, because this case has to do with translating a philosopher who in fact thought that Millianism is incorrect. One might think that translators homogeneously opted for the strategy they opted for in order to respect Frege’s thesis that “the morning start” and “the evening start” do not have the same meaning. Moreover, one might hold that the example is somehow sui generis also because it concerns an identity statement, i.e. “Der Abendstern der Morgenstern ist”, and identity is always puzzling and special.⁵ But in fact Frege’s own ideas and identity statements cannot be taken to be the culprits. In order to see this, take for example the following passage, in which one of the main defenders of neo-Russellianism, Nathan Salmon, explains what neo-Russellians hold:

• (11) Now there is no denying that, given the proper circumstances, we say things like “Lois Lane does not realize (know, believe) that Clark Kent is Superman” and “There was a time when it was not known that Hesperus is Phosphorus” … Recently a number of philosophers, mostly under the influence of the direct reference theory, have expressed doubt about the literal truth of such utterances in ordinary usage. If someone believes that Hesperus is a planet, they claim, then, strictly speaking, he or she also believes that Phosphorus is a planet. (1983, 81)

This passage belongs to a book that has not been translated, so we cannot rely on actual translations. Still, take the last sentence, in which “they” stand for neo-Russellians. It is pretty obvious that nobody would even dream of translating it by using only one name for Venus: it really does not seem that it is under the influence of the direct reference theory that we would claim that if a subject believes that Hesperus is a planet, then the same subject also believes that Hesperus is a planet. Even Frege would have claimed that! Salmon’s passage is by an author who does believe that co-referential names have the same meaning. Moreover, no identity statement is involved. But then why is it that in translating it we would certainly use two different, albeit co-referential, names, and not only one if, according to neo-Russellians, it should make no difference to the propositions expressed?⁶

3 A neo-Russellian Way Out: Pragmatic and Semantic Translations

As it always is when their account apparently clashes with facts about how we actually use languages, neo-Russellians would probably react to the seeming counterexamples provided by the translations above by invoking the semantic/pragmatic distinction, for example in the following way. If translation were an entirely semantic notion, they might urge, then clearly it would be sufficient to show that co-referring names do not perform exactly the same semantic functions and therefore are not everywhere interchangeable. But it is not immediately clear that facts pertaining to translation per se can show anything concerning the semantic functions of names. It is not to be doubted that the business of translation is to produce a text that gives to a competent speaker of the target language the same information that the original text gives to a competent speaker of the source language. But more than one kind of information exists. In fact Salmon himself distinguishes sharply between two kinds of translation that differ in so far as one only cares about strictly semantic information, whereas for the other “pragmatically imparted information” takes precedence over semantic information. In order to see what these two translations are, let us take the first sentence from the passage by Frege seen above:

• (8) der Gedanke des Satzes “der Morgenstern ist ein von der Sonne beleuchteter Körper” verschieden von dem des Satzes “der Abendstern ist ein von der Sonne beleuchteter Körper”. Jemand, der nicht wüsste, dass der Abendstern der Morgenstern ist, könnte den einen Gedanken für wahr, den anderen für falsch halten.

How should we translate the two sentences mentioned in (8), i.e.

• (12) der Morgenstern ist ein von der Sonne beleuchteter Körper

• (13) der Abendstern ist ein von der Sonne beleuchteter Körper

? As many, among whom Burge (1978a, 141–144), Hartry Field (2001, 161), Peter Geach (1957, 91–92), Dummett (1981, 93–94), and Kripke (1979, 139, f. 25) have stressed, all actual translations translate what occurs within quotation marks:

• (9) the thought in the sentence “The morning star is a body illuminated by the Sun” differs from that in the sentence “The evening star is a body illuminated by the Sun.” Anybody who did not know that the evening star is the morning star might hold the one thought to be true, the other false.

While discussing how to translate passages like the one above, Salmon holds that actual translations, translating what occurs within quotes, are pragmatic translations—“preserving, for example, the point being illustrated (if necessary at the expense of preserving sense)” (1993, 166, f. 15) and preserving, more generally, the “pragmatically imparted information” (2001, 355). But, Salmon stresses, there is another bona fide translation of the first sentence in Frege’s passage, i.e.

• (14) the thought in the sentence “der Morgenstern ist ein von der Sonne beleuchteter Körper” differs from that in the sentence “der Abendstern ist ein von der Sonne beleuchteter Körper,”

which is its semantic translation, the one that preserves the “semantically encoded information” (2001, 355). Put differently, for Salmon, (14), i.e. the semantic translation, is the one that might not be able to convey Frege’s point to speakers of English, but is the one that preserves “the proposition semantically expressed” (1993, 166, f. 15) by Frege’s original passage. For in the original passage reference is made to two German sentences and if we want in the translation to preserve the propositions expressed by the German passage we have, in the English translation, to refer to those very same German sentences. On the other hand, the pragmatic translation, i.e. (9), as we saw, conveys Frege’s point to English speakers, but since in it reference is instead made to two English sentences, it expresses different propositions from those expressed by the original passage (2001, 348).⁷

Let us now put aside the issue of how to translate quoted sentences and let us go back to proper names. Neo-Russellians would probably exploit this distinction between semantic and pragmatic translation in order to explain why in translating the second sentence in Frege’s passage (8) and the last sentence in Salmon’s passage, i.e.

• (11) Now there is no denying that, given the proper circumstances, we say things like “Lois Lane does not realize (know, believe) that Clark Kent is Superman” and “There was a time when it was not known that Hesperus is Phosphorus” … Recently a number of philosophers, mostly under the influence of the direct reference theory, have expressed doubt about the literal truth of such utterances in ordinary usage. If someone believes that Hesperus is a planet, they claim, then, strictly speaking, he or she also believes that Phosphorus is a planet,

we would use two different names for Venus. They might in fact hold that surely we would use two names, but that is because in translating we are providing a pragmatic translation, whose purpose is not to preserve the propositions expressed, but to convey the points. Since a pragmatic translation concerns pragmatic information, then, firstly, it is not surprising that it might depart from mere semantic constraints, and, secondly, we cannot draw conclusions on the semantics of proper names from it. Moreover, neo-Russellians can add, from the purely semantic point of view,

• (15) If someone believes that Hesperus is a planet, they claim, then, strictly speaking, he or she also believes that Hesperus is a planet

is in fact a perfectly adequate translation of the last sentence in Salmon’s passage. For although a reader would not be able to get the point, and no matter how counter-intuitive this might seem, still (15) expresses the very same proposition expressed by

• (16) If someone believes that Hesperus is a planet, they claim, then, strictly speaking, he or she also believes that Phosphorus is a planet.

That is exactly what neo-Russellianism amounts to. Note that, according to it, the truth-values of (15) and (16) are the same.

Is this neo-Russellian move satisfactory? There is plenty to discuss about it. Recently, for example, Peter Hanks has urged that how we use proper names in translation and in conversation in general is relevant to semantics:

The English translation of the Greek name “Ελλαδα” is “Greece”, not the etymologically and phonetically more appropriate “Hellas”. “Greece is pretty” is a translation of the corresponding Greek sentence; “Hellas is pretty” is not. To say this is to classify utterances of “Greece is pretty” together with the corresponding Greek utterances, and to classify utterances of “Hellas is pretty” separately. This classification deserves the label “semantic” because it is backed by a convention among translators and bilinguals (2015, 129. Our emphasis)

The neo-Russellian notion of semantic translation relies on the idea that use or conventions among translators and bilingual are irrelevant when it comes to semantic considerations, and this is not uncontroversial. However, we will not try, here, to establish what semantics is and whether actual use can be irrelevant to it. Neither shall we discuss how exactly the alleged semantic/pragmatic divide can be drawn. No matter how sloppy and wanting the distinction between semantic and pragmatic translation is, we will nonetheless accept it for the sake of argument, and try to understand whether semantic translations can show us anything about neo-Russellianism.⁸

4 Semantic Translations: Problems for neo-Russellianism?

While discussing Millianism and translation, John Burgess notes:

in any case of apparent counterexample, the militant Millian … can simply suggest that whatever phenomena are cited by their critics pertain to pragmatics rather than semantics. (2005, 203)

But, pace Burgess, it does not really seem that neo-Russellians can explain each case that stems from translation and that seems to create a problem by merely and simply relying on the semantic/pragmatic distinction. In order to see this, we can rely on intra-linguistic translation, i.e. on translation from English into English. Let us now take the following passage, in which Salmon states his own view:

• (17) However inappropriate it may be in most contexts to say so, Lois Lane is (according to the myth) fully aware that Clark Kent is Superman … We do not speak this way, in fact it is customary to say just the opposite. (1983, 84)

Obviously, according to neo-Russellianism, (17) is true. Now take the following:

• (18) However inappropriate it may be in most contexts to say so, Lois Lane is (according to the myth) fully aware that Clark Kent is Clark Kent … We do not speak this way, in fact it is customary to say just the opposite.

(18) is, according to neo-Russellians, a semantically adequate intra-linguistic English translation of (17). But neo-Russellians (and everybody else) should admit that (18) is false, because we speak that way indeed: we do say that Lois Lane is fully aware that Clark Kent is Clark Kent.

What can neo-Russellians say? Clearly, they cannot rely on the semantic/pragmatic distinction here, because the translation we have considered is purely semantic and any consideration pertaining to pragmatically imparted information is simply irrelevant. Moreover, it cannot be claimed that preserving the truth-values of the sentences to be translated is not mandatory, again because the translation considered is purely semantic. It is precisely the business of semantic translation to preserve propositions. Given that propositions are the bearers of truth and falsity, semantic translation must preserve truth-values. While discussing his principle of translation, i.e. that if a sentence of one language expresses a truth in that language, then any translation of it into any other language also expresses a truth (in that other language), Kripke himself in fact notes that if there are violations to the principle, they only concern pragmatic translations:

Some of our ordinary practice of translation may violate this principle; this happens when the translator’s aim is not to preserve the content of the sentence, but to serve—in some other sense—the same purposes in the home language as the original utterance served in the foreign language. But if the translation of a sentence is to mean the same as the sentence translated, preservation of truth-value is a minimal condition that must be observed. (1979, 139)⁹

Arguably, though, neo-Russellians have a way of explaining the different truth-values of (17) and (18). In (17) the expressions “so” and “this way” occur and they, neo-Russellians can hold, denote some piece of language that occurs in (17) itself. By unpacking “so” and “this way”, we obtain from (17)

• (19) However inappropriate it may be in most contexts to say “Lois Lane (according to the myth) is fully aware that Clark Kent is Superman”, Lois Lane is (according to the myth) fully aware that Clark Kent is Superman … We do not speak in the way of saying “Lois Lane (according to the myth) is fully aware that Clark Kent is Superman”, in fact it is customary to say just the opposite.

Thus, neo-Russellians can add, if in (17) we substitute “Clark Kent” for “Superman”, the denotations of “so” and “this way” change. In unpacking the “so” and “this way” occurring in (18), in fact, we obtain

• (20) However inappropriate it may be in most contexts to say “Lois Lane (according to the myth) is fully aware that Clark Kent is Clark Kent”, Lois Lane is (according to the myth) fully aware that Clark Kent is Clark Kent … We do not speak in the way of saying “Lois Lane (according to the myth) is fully aware that Clark Kent is Clark Kent”, in fact it is customary to say just the opposite.

Thus, they can urge, although the occurrences of “Superman” and “Clark Kent” in (17) and (18) are purely referential as in accordance with Millianism, the names are not substitutable salva veritate exactly because the denotation of “so” and “this way” would then change. Put differently, neo-Russellians can hold that examples such as (17) are similar to Quine’s famous

• (21) Giorgione was so-called because of his size (1953, 139).

As Kit Fine (1989, 89) and François Recananti (2000, 141–143) observe, we might well take “Giorgione” as it occurs in (21) to be directly referential, as in accordance with Millianism, but nonetheless hold that the name cannot be substituted with “Barbarelli” salva veritate because “so” denotes the name occurring before it, and if we change the name, we change the denotation of “so”.

Is this neo-Russellian metalinguistic strategy for explaining (17) away entirely satisfactory? There is room for doubt. For it is not obvious that “so” and “this way” really pick pieces of language. One might in fact unpack the “so” and “this way” occurring in (17) not meta-linguistically, as in (19), but as in

• (22) However inappropriate it may be in most contexts to say that Lois Lane (according to the myth) is fully aware that Clark Kent is Superman, Lois Lane is (according to the myth) fully aware that Clark Kent is Superman … We do not speak in the way of saying that Lois Lane (according to the myth) is fully aware that Clark Kent is Superman, in fact it is customary to say just the opposite.

In order for neo-Russellians to explain these cases away, they need to hold that (19) is a better unpacking than (22). But what would the rationale for this thesis be, apart from a desire to save Millianism from apparent counterexamples? Thus this strategy seems to be, although clearly successful, somehow arbitrary.¹⁰

Be it as it may with these cases, we can move to another kind of example that seems problematic in a different way. Let us take Kripke’s disquotational principles:

the disquotational principle … can be stated as follows, where ‘p’ is to be replaced inside and outside all quotation marks, by any appropriate standard English sentence: “If a normal English speaker, on reflection, sincerely assents to ‘p’, then he believes that p.” … A similar principle holds for sincere affirmation or assertion in place of assent. (1979, 137–138)¹¹

As Kripke remarks, “taken in its obvious intent, after all, the principle appears to be a self-evident truth” (ibid.): the principle tells us that if a subject understands a sentence and is sincere, i.e. she only assents, affirms or asserts what she believes, then she believes what she assents to, affirms or asserts. It thus seems that we really cannot reject this principle. Now Kripke suggests that we can strengthen the principles above so as to obtain principles of the following kind:

There is also a strengthened ‘biconditional’ form of the disquotational principle, where once again any appropriate English sentence may replace ‘p’ throughout: A normal English speaker who is not reticent will be disposed to sincere reflective assent to ‘p’ if and only if he believes that p. (ibid.)

Neo-Russellians do not and cannot endorse the strengthened versions of the principles. For suppose that John assents to

• (23) Hesperus is a planet,

but does not assent to and, although not reticent, is not disposed to assent to, assert or affirm

• (24) Phosphorus is a planet.

This might well happen even if John is rational, because he might be mistaken on whether Hesperus is Phosphorus. Thanks to the weaker disquotational principle, since John assents to (23), we can conclude

• (25) John believes that Hesperus is a planet.

According to neo-Russellians, this means that John is related to the Russellian proposition that Hesperus is a planet. But this proposition is the Russellian proposition that Phosphorus is a planet, so that if (25) is true, according to neo-Russellians, so is

• (26) John believes that Phosphorus is a planet.

But although (26) is true according to neo-Russellianism, John does not assent to and, although not reticent, is not disposed to assent to, assert or affirm (24). Therefore according to neo-Russellianism he believes a proposition expressed by a sentence even though he would not give assent to or affirm or assert that sentence, contrary to the strengthened version of the disquotational principles. Thus because of how they individuate propositions, neo-Russellians need to see a discrepancy between what we can call the positive and negative cases: although behaviour (assertion, assent, etc.) is a reliable sign of the presence of an attitude—if you assent you believe—behaviour is not a reliable sign of the lack of an attitude—it can happen that you believe even though you do not assent. This is somehow puzzling: why is it that a subject is a reliable measure of her own positive attitudes, but not a reliable measure of her lack of such attitudes? But neo-Russellians would say that this discrepancy is to be explained in terms of the fact that a subject can be mistaken: it is because John is mistaken about whether Hesperus is Phosphorus that he does not realize some of his own beliefs.

Still, there seems to be a problem. In order to see this, let us start from noting that the weak disquotational principles link assertion, sincerity and belief in such a way that from an assertion via which a subject represents herself as believing something, we can conclude that she does believe that. But then if, as the disquotational principle tells us, we are justified in moving from a subject’s representing herself as believing something to her believing that thing, then we should also take the following principle of credence to be true: If a normal English speaker, on reflection, sincerely assents to ‘I believe that p’, then she believes that p.¹² Now belief is only one of the attitudes we might have. Given that belief does not seem special and that we had better take this principle of credence to be true, then we had better take other principles for other attitudes to be true as well. Forget about knowledge: if a subject affirms that she knows something, it does not follow that she does know it. For example, Macbeth would affirm that he knows that there is a dagger in front of him, but he does not know that, because in fact there is no dagger. But take attitudes such as holding, or maintaining. If a subject affirms ‘I maintain that p’, we are arguably safe in concluding that she does maintain that p. Neo-Russellians should presumably accept these principles on belief and maintaining as well, because the rationale behind them is the same rationale behind Kripke’s weak disquotational principles, and, as we saw, those are self-evident truths. Nonetheless, neo-Russellians surely would hold that there is an asymmetry between the positive and the negative cases also as to what a subject holds or does not hold, maintains or does not maintain: even though a subject would affirm that she does not maintain that Hesperus is a planet, it might be that she maintains that Hesperus is a planet, for example because she would affirm that she maintains that Phosphorus is a planet. But now take doubt: is it a positive attitude or the lack of an attitude? Because they accept the weaker forms of the disquotational principles, which make us conclude that a subject believes something, for example, but not the stronger forms, which make us conclude that a subject does not believe something, it seems that according to neo-Russellianism an attitude is positive when expressed by a predicate that has a positive form, such as “to believe”, “to hold”, “to fear”, and negative when expressed by a predicate that has instead a negative form, such as “not to believe”, “not to hold”, “not to fear”. It then seems that we should take doubt to be a positive attitude, so that if a subject affirms that she doubts something, then we are safe in concluding that she does in fact have such a doubt. But this leads to problems. For suppose that Suzy, sincere, reflective, competent, etc. assents to:

• (27) Do you doubt that Hesperus is a planet?

Since Suzy is reliable in the positive cases, i.e. what she believes about her attitudes is what is really the case with her attitudes, then if positive cases are those with the predicates in the positive form, doubting is a positive case and then it really is the case that she doubts that Hesperus is a planet. We can then conclude that the following is true:

• (28) Suzy doubts that Hesperus is a planet.

Now let us imagine a language, English[to not pirrt/to doubt], which is exactly like English, but in which there is no “to doubt” or any synonymous predicate positive in form, and what “to doubt” means is in fact expressed by “not to pirrt”. This cannot do any harm to the expressivity of the language.¹³ Now suppose that Suzy is a perfect bilingual speaker of both languages. She assents to the English[to not pirrt/to doubt] question

• (29) Do you pirrt that Phosphorus is a planet?.

This is perfectly fine: Suzy, who is perfectly bilingual, would in fact assert, for example, the English

• (30) I do not doubt that Phosphorus is a planet.

Because Suzy assents to the English[to not pirrt/to doubt] question in (29), and since all the predicates are in positive form, if the form of the predicate establishes whether a case is positive or negative, we are in front of a good case, and Suzy is then reliable about it, so that we can conclude that the English[to not pirrt/to doubt)] sentence

• (31) Suzy pirrts that Phosphorus is a planet

is true. But then, if neo-Russellians are right, also the English[to not pirrt/to doubt] sentence

• (32) Suzy pirrts that Hesperus is a planet

is true. Then, we should conclude that the English[to not pirrt/to doubt] sentence

• (33) Suzy does not pirrt that Hesperus is a planet

is false. For while, at least according to neo-Russellians you can have the same attitude toward a proposition and its negation—for example you can believe both that Hesperus is a planet and that Hesperus is not a planet—, it cannot be the case, even for neo-Russellians, that you can both have and not have the same attitude toward a certain proposition—it cannot be the case, for example, that both you do and you do not believe that Hesperus is a planet. But (33) is an English[to not pirrt/to doubt] semantic translation of the English

• (28) Suzy doubts that Hesperus is a planet,

and while (33) is false, as we saw, (28) is true. But how can they have different truth-values if, in accordance with neo-Russellianism, they are semantic translations of each other, and then express the same proposition?

Sentences such as (28) and (33) do not show that neo-Russellianism is false. The example does show, though, that we cannot rely on the positive or negative form of the predicates in establishing what attitudes are positive and what negative, and thus which of her own attitudes a subject is reliable on. Neo-Russellians then have two options. The first option is to hold that whether an attitude is positive or negative does not depend on the linguistic form of the predicates denoting it, but on some other grounds, and that the disquotational principles and the other principles of credence, holding, etc. should only be applied to those attitudes that are genuinely positive, no matter what the form of the predicates is. But what are these other grounds? In other words, does this alleged difference between positive and negative attitudes really cut at any mental joints? The problem does not seem to concern only “to doubt”: think, for example, about ignoring or being unsure. If neo-Russellians choose this option, they owe us an explanation as to whether, no matter what the linguistic form of the predicates is, they should be taken as positive or negative, and what reasons they have for classifying those attitudes in that way. Alternatively, neo-Russellians can simply deny that all the principles we relied on are true, by holding that behaviour simply is no indication of an attitude. But, as we saw, the weak disquotational principle seems a self-evident truth, and the other principles we saw about credence and other attitudes seem to be justified on the same grounds that justified the allegedly self-evident weak disquotational principles. Whatever option neo-Russellians opt for, therefore, it does not seem that they have an easy task to perform.

5 Conclusion

Translation has often played a role in establishing what truth and meaning in general are, or what the meaning is of some particular bits of language. But it is dubious that we can reach such semantic conclusions by relying on considerations concerning translation. First of all, translation is very much a pragmatic matter, so that it does not seem that we can conclude anything about the semantics of a certain bit of language by merely considering how we usually translate such a bit. For example, it would be a plain mistake to translate from English into French “but” as “et”, instead of “mais”. But this does not show that the difference between “and” and “but” pertains to semantics. One might well hold, with Frege, that “[t]he word “but” differs from “and” in that with it one intimates that what follows is in contrast with what would be expected from what preceded it. Such suggestions in speech make no difference to the thought” (1956, 295–296), and add that the way in which we translate “and” and “but” should be explained in terms of preservation of what a word intimates. Moreover, it seems that we cannot use the notion of purely semantic translation in order to establish, for example, what proposition is expressed by sentences such as

• (1) John believes that Mont Blanc is 4000 m high.

For in order to understand what translation of (1) is semantic, i.e. proposition preserving, we would have to already know what proposition (1) expresses.

Still, although less powerful than what one might think, translation is indeed a useful tool in our semantic enterprises, such as that of establishing the meaning of proper names as they occur in sentences such as (1), since it can, for example, make it manifest that a certain account relies on some choices, distinctions, and theses that are somehow arbitrary.¹⁴