Is That a Fish in Your Ear?
Translation and the Meaning of Everything

by David Bellos

Farrar, Straut and Giroux, 2011, 2012

YUTAKA TADOKORO [Director, in Japanese, to the interpreter]:
"The translation is very important, O.K.? The translation."

AKIKO TAKESHITA [Ms. Kawasaki, Interpreter, in Japanese, to the director]:
"Yes, of course. I understand."

YUTAKA TADOKORO [in Japanese]:
"Mr. Bob. You are sitting quietly in your study. And then there is a bottle of Suntory whisky on top of the table. You understand, right? With wholehearted feeling, slowly, look at the camera, tenderly, and as if you are meeting old friends, say the words -- as if you are Bogie in Casablanca, saying, 'Here's looking at you, kid' -- 'Suntory time!'"

AKIKO TAKESHITA [in English, to Bob]:
"He wants you to turn, look in camera. O.K.?"

"That's all he said?... I mean, it seemed like he said quite a bit more than that."

Lost in Translation, 2003, Focus Features

David Bellos is a Professor of French and Comparative Literature at Princeton University and the Director of the "Program in Translation and Intercultural Communication" there. His book is then, not surprisingly, about translation, of which he thinks highly. That in itself is a welcome difference from the contemptuous attitude about translation found in much of academia today, which is very different from the prestige of translation, for instance, back in the 19th century, when translators like James Legge and Max Müller represented a very serious, respected, important, and desired department of scholarship. Indeed, one of the evils of "Theory" in the contemporary academy is that it renders even knowledge of various languages, let alone a vocation in translating them, unnecessary -- and this may very well be its attraction (as Bellos notes "the precipitous decline in the teaching of foreign langauges in the English-speaking world in the last fifty years" [p.270]). After all, Sanskrit is pretty daunting, as is even the Russian that Bellos himself quotes freely. And any conscientious scholar should be shocked to learn that "multiculturalism" in education usually explicitly exempts any knowledge of the history, language, or literature of, in fact, other cultures. If we think to wonder what is left, we might be happier not to ask -- if not for the damage it is doing to education and the future of the Nation.

The title of Is That a Fish in Your Ear? derives from something that Bellos references late in the book:

It makes people think that it's only a matter of time before we can all have a device to stick in our ear -- the "Babel fish" of The Hitchhiker's Guide to the Galaxy -- to provide us with instant communication with all the peoples on earth. [p.270]

This comes up in an informative chapter (#24) on simultaneous translation, such as is done at the U.N. and other international agencies. A title of the book like that, of humorous and popular reference, is characteristic of a book that is by turns delightful, enlightening, annoying, and appalling. Bellos is occasionally careless with his historical facts, sometimes even facts about language. And his book develops a curious feature, more than once, of affirming a thesis about language or translation that his own argument then succeeds in refuting. Philosophers generally try to avoid that sort of thing.

First let me look at some of the carelessness. Thus, Bellos says, "In Sumerian, the language of ancient Babylon..." [p.29]; but the Sumerian language had died out by the time of the I Dynasty of Babylon (1894-1595 BC). The last instance of a state primarily using the Sumerian language was the III Dynasty of Ur (2112-2004 BC), by whose end even the Kings had Akkadian names. The "language of ancient Babylon" was the aptly named "Babylonian," originally a dialect of Akkadian, which was unrelated to Sumerian. However, since the Babylonians were well aware that their civilization derived from the Sumerian, knowledge of the Sumerian language was preserved as long as Babylonian itself continued; and Assyrian kings would later boast of their facility with Sumerian. But this was like Calvin Coolidge reading Greek and Latin -- the last President to do so.

From other indications in the book, Bellos seems to have some awareness of this; but he clearly has not gotten it all straight. Thus, he later refers to "the Assyrian god Marduk" [p.326]. Not only was Marduk not an Assyrian god, he was the patron god of Babylon itself. The orginal patron god of Assyria was the god Asshur, , of the city of Ashur, where obviously the names are all of common origin. One wonders where Bellos has been getting his information about Mesopotamian history and linguistics. There is a serious muddle here between Sumer, Babylon, and Assyria.

There is a great deal of untranscribed Russian in Bellos's book, and Bellos does seem to know his Russian. However, there are some problems with his knowledge of Arabic, Hebrew, and Greek. Thus, he says:

...the existence in Arabic of ghanam, a word that means "sheep" and "goats" without distinction, does not prevent speakers of Arabic from sorting sheep from goats when they need to. [p.89]

I think that this is the only example in the book of citing an Arabic word in the Arabic alphabet, and it is a disaster. The letters are written in the wrong order, left to right instead of right to left, and the independent forms of the letters are used rather than the connecting forms used in most writing. Ghanam in Arabic is . Bellos seems to be unaware of things about Arabic writing that a student may begin to learn on the first day of class. Bellos also misses the circumstance that there are other words for sheep and goats in Arabic, while this word is related to words for "booty," where tossing sheep and goats together makes perfect sense. You may capture them together. On the other hand, Bellos is trying to make the point that people can be aware of differences in reality that are not necessarily reflected in their language. This is a sensible observation, and it is of a piece with the sound substantive theory of the book, although not always consistent with other things [note].

It is not only Arabic with which Bellos has this kind of trouble:

...the corresponding expression in the Torah, , pronounced "adonai ilehenu," can be worded out as "God, the Lord-Our." [p.173, with vowel points added]

I don't know what is supposed to be. The Ben Yehuda dictionary has , "lamentation," and , "lo! behold!" Neither one of them would occur in this case. For these letters to be "pronounced 'adonai'," the text must have , or, with the vowels, . This is the name of God (, Ha Shêm, "the Name"), its actual pronunciation is forgotten, and it is always read , adonâi (or adonoy). The vowels in are the vowels of adonai, although they have mistakenly been used by Christians to generate "Jehovah" as the supposed name of God. So we are left to wonder how Bellos managed to quote and transcribe it adonai. If is actually some traditional abbreviation for , I have missed that; and, using it, Bellos should at least explain what is going on. Since he doesn't, I think this is just all confusion here.

What Bellos apparently means to quote from the Torah is , adonai elohenu. But even if he had gotten the Hebrew right, he gets the transcription wrong. Thus, I have supplied the vowel points to show that the expression "our God" is read elohenu rather than ilehenu. I would like to know where he got that. See the discussion elsewhere for the words for God in Hebrew, Greek, and Arabic.

As far as I have noticed, Bellos only gives one Greek word in the Greek alphabet in the book, and it is not even a Greek word but the Greek transcription for "cherubim" [p.106]. Otherwise, he doesn't seem to know his Greek very well.

For the ancient Greeks, the sound of the foreign was the unarticulated, open-mouthed blabber of va-va-va, which is why they called all non-Greek-speakers varvaros, that is to say, barbarians, "blah-blah-ers." [p.48]

Here, of course, the Greek word in question is . If Bellos does not know that the ancient Greek letter beta was pronounced "b" and not "v" (as in Mediaeval and Modern Greek), then his Greek education has not gotten very far. And if he accepts the modern Greek ahistorical claim that Ancient Greek was pronounced exactly like Modern Greek, he should say so and explain why he thinks that the pronunciation of languages never changes and why all the people in the discipline of historical linguistics are wrong about Ancient Greek. Instead, he doesn't explain anything, and my guess is that he has picked up the word from Modern Greek and actually knows very little of anything about Ancient Greek, or the "ancient Greeks." The business about babbling barbarians can be found in multiple popular sources. Since in the same passage Bellos gives two Russian words in untranscribed Cyrillic, the impression is unavoidable that he is showing off his Russian but has not done his homework on Greek.

More shocking is another statement about Greek, "Classical Greek has no proper word for word" [p.92]. However, the word for "word" in Greek is , which ends up being one of the most famous words in the language:

In the beginning was the Word [ ], and the Word [ ] was with God, and the Word [ ] was God. [John 1:1]

We thus might wonder whether Bellos knows his Bible any better than he knows his Greek. However, does indeed have an unfamiliar range of meanings, which at the very least means it can be translated verbum, "word," or ratio, "reason," in Latin. But the meaning is larger even than that, so that its ambiguity could be played upon all the way from Heraclitus to the Gospel of John itself. Unless Bellos wants to argue that a word with a range of meanings cannot be used for one of those included meanings, he has made a bad mistake. But his larger point in this passage is no better, that somehow the very idea of a "word" is a fiction. His examples that words have different kinds of boundaries in different languages mean no more than that words have different kinds of boundaries in different languages -- rather than that words don't exist or are only artifacts of writing.

He seems to think that Greek not having a word for "word" proves something, when elsewhere we have seen he otherwise argues that Arabs can distinguish sheep from goats even though he (falsely) says that they only have a word that means both. So even if the Greeks had no word for "word," and spaces were not left between words in ancient writing, there is no reason why they could not understand what such a thing was, as they must have done when Mediaeval scribes began, unambiguously, to introduce those spaces, as we still do. This is already an example of the inconsistency of Bellos's arguments, and he is reaching for a larger confused argument that "words and rules" somehow don't really exist, which I will examine in turn.

Other problems in Bellos's book concern some larger historical issues. He says,

Throughout the period of learned Latin in Western Europe, immersion was not an option. There was no environment in which everybody spoke Latin as a native tongue, and so the language had to be taught by teachers, in classrooms, through writing. [p.113]

This is quite false. While it is true that nobody "spoke Latin as a native tongue," there were indeed Mediaeval environments where "immersion" was "an option" and where Latin could be learned where "everybody spoke Latin" and it did not need to be "taught by teachers, in classrooms." This would have been at the universities and at many venues of the Church, where Latin was actually the living language of daily business and daily life, which is why there were even drinking songs (e.g. Gaudeamus Igitur) in Mediaeval Latin. Students from Poland or Spain or Ireland could all go to the University of Paris and converse in the common language, Latin. It is doubtful that anyone in Mediaeval Europe learned Latin "through writing" as the modern Latin student might -- in a period when books were rare and expensive and no humble student could have afforded the equivalent of Wheelock's Latin, such as there may have been.

Bellos has other problems with languages in the Middle Ages, particularly with Greek again:

Similarly, in the fourth to eighth centuries C.E., Syriac (a Semitic language closely related to Aramaic) is said to have flowered in the hands of Severus Sebokht, a bishop, scholar, and translator who imported quantities of Greek words and expressions together with the mathematical, medical, and astronomical knowledge of the ancient Greeks that the Latin West had ignored (and would not rediscover for centuries, until Arabic translations of those Syriac translations of Greek science were translated once again in the middle of the twelfth century C.E., in Toledo [Spain], by Gerard of Cremona, into Latin, for wider distribution throughout Europe). [p.188]

First of all, Severus Sebokht cannot have been doing his work "in the fourth to eighth centuries" -- that's four hundred years -- so obviously Bellos is thinking of other translators of Greek into Syriac -- and, without dates, I'm not even sure who "Severus Sebokht" is, unless Bellos means the important 6th century Severus who was Patriarch of Antioch (512-518). Second, Syriac is a daughter language of Aramaic, and is sometimes still called "Aramaic" itself, as I have discussed elsewhere. So while it is not untrue to say that Syriac is "closely related to Aramaic," a less vague characterization would have been easy and appropriate. Third, while some early translations of Greek philosophy were by way of Syriac translations, Arab translators soon learned Greek and did their translations directly. Fourth, the Latin West did not "ignore" Greek literature; it simply didn't have any. Fifth, at the very time, in the 12th century, when translations from Arabic to Latin began to be available in Western Europe, Greek texts were themselves already beginning to arrive.

While this passage is more just sloppy than it is incorrect, it may be a clue that Bellos has gotten the wrong idea about some things. Thus, Bellos says, "Arabic was the pivot language that allowed Greek philosophy to be translated into European tongues" [p.213]. This may have been true for a while, since St. Thomas seems to have mainly relied on translations of Aristotle from Arabic, but the statement is no less than bizarre when we realize that Greek was itself one of the "European tongues" in which Greek philosophy survived in its original language. Nobody in Constantinople needed Arabic translations to read Greek philosophy. This is not as bad as Christopher Hitchens thinking that Greek philosophy only survived in Arabic, because Justinian had burned it, or something; but it does not look like Bellos has gotten the matter quite straight. Of course, people forgetting, sometimes in published books, that an entire civilization existed at Constantinople, with Roman law, Greek literature, and monumental architecture and art, is not that unusual, even with someone like Bellos who seems to know a fair amount about the Ottomans.

Here and there Bellos gets a matter of pure history entirely wrong. So, he says, "On his return to Genoa in 1298 B.C., Marco Polo was flung into jail" [p.196]. Since Marco Polo was a Venetian, he did not ever "return" to Genoa. He was "flung into jail" in Genoa as a prisoner of war after he had been captured in a naval battle, the Battle of Curzola, between Genoa and Venice, which Venice lost. This did turn out to be fortunate, since Polo, no writer himself, narrated the stories of his adventures to someone who was, and who wrote it up.

Next we get:

When the French seized the throne of Sweden in the Napoleonic Wars, they didn't start translating from Swedish. In fact, the new Swedish royal family and its court carried on speaking French for more than a hundred years..." [p.208]

Since Charles XIV John Bernadotte of Sweden, who was French, did not come to the throne of Sweden until 1818, Napoleon would have needed to conquer Sweden from his exile on St. Helena. A nice trick. So the French did not seize the throne of Sweden. Instead, the old Royal family died out, and Bernadotte was chosen for the throne, in part because his military experience was seen as helpful to the Kingdom. One is tempted to ask Bellos what battles were fought in the French conquest of Sweden.

Another habit of Bellos is annoying more than incorrect. He is one of the people who seems to think it clever to refer to the Persian language as "Farsi." This is the name of the Persian language, , in Persian, not in English. With Bellos, I think this would be less annoying were he consistent about it. Thus, we see "Persian" on pages 122, 124, 169, 211, 255, and 326 (as "Old Persian"), but we get "Farsi" on pages 244, 246, and 255. Both "Persian" are "Farsi" are used on page 255. The untutored reader might take away from this that Farsi and Persian are different languages, which is not the right impression for Bellos to give. He should stick to one or the other. Indeed, I do wonder why he alternates. It looks a bit like he uses "Persian" when dealing with scholarly sources that say that, but "Farsi" when he is voicing nothing but his own thoughts.

After beating up on David Bellos on all these points, one might begin to wonder "Why bother?" But Bellos is not reluctant to wade in on issues of philosophy of language; and what he does with them is always going to be of interest. Much of what he prefers is radiantly correct, but in other areas he has made a mess of it.

If for nothing else, I should review this book because Bellos favorably quotes Jerrold Katz, whom I knew and admired:

Any thought a person can have, philosopher Jerrold Katz argued, can be expressed by some sentence in any natural language; and anything that can be expressed in one language can also be expressed in another. What cannot be expressed in any human language (opinions vary as to whether such things are delusional or foundational) lies outside the boundaries of translation and, for Katz, outside the field of language, too. This is his axiom of effability. One of the truths of translation -- of the truths that translation teaches -- is that everything is effable. [p.153]

This is an important and correct point, and Bellos is on solid ground with Katz. However, Bellos does not cite any of Katz's books, and it is not clear that he is familiar with Katz's arguments. Indeed, talking about an "axiom of effability," as I am not sure Katz does in those terms, may generate some confusion, since an "axiom," strictly speaking, does not need to be, or at least is not, proven. Instead, in the Metaphysics of Meaning, Katz's argues against naturalistic and Nominalistic theories of language, such as had been promoted by Wittgenstein, W.V.O. Quine, and others. In a footnote, Bellos mentions Quine, but not what he thought [p.346, 13:4]. In fact, Quine's Nominalism led to the doctrine of the "indeterminacy of translation," which is not what Bellos wants to endorse. Bellos, does, however, endorse some kind of naturalism or Nominalism. This involves rejecting, as it happens, Katz's own "metaphysics of meaning," and Bellos may not be aware of that.

The thesis of Bellos is that, "Translation is meaning" [p.87], which captures the whole Nominalist thesis also, that language does not exist apart from its material expression. Bellos even goes further than Nominalism, whose name derives from nomina, "names," since he apparently rejects the independence and reality of "words" and "meanings" as things that constitute, or underlie, language. Thus, he says:

"Scriptism," as Roy Harris called the illusion that a language consists of things called words...

The status of any utterance in a mental world without script derives mainly from the identity of the speaker, much less from the "meanings" of the "words" that are spoken. The concepts in scare quotes are probably not even thinkable without writing. The indeterminacy of the flow of speech and the dependence of meaning on the human context in an oral culture are pinpointed with affect and insight by Tolstoy in his portrait of the illiterate peasant-philosopher Platon Karatayeve, in War and Peace:

Platon could never recall what he had said a moment before... [p.119]

The occurrence here of "indeterminacy" is awkward when it reminds us of the theory of Quine that translation is indeterminate, which is not where Bellos wants to go. And this picture is of course all falsified by the existence of the Iliad and the Odyssey, which were created and lived in an oral culture, and by the professional and official "rememberers" of the Incas, any one of who would have been out of a job if he "could never recall what we had said a moment before..." We have lost the "human context" of the oral culture in which the Iliad and Odyssey were composed, but no one has had trouble understanding them since then, or of reconstructing much about the culture from the texts. And since many in linguistics study languages that only exist orally, it is not clear from Bellos how they can describe "meanings" and "words" in those languages, as they do. We also get swipes from Bellos at the rules of language as well as words:

If you remain attached to the idea that a language really does consist of words and rules and that meaning has a computable relationship to them (a fantasy to which many philosophers still cling), then GT ["Google Translate"] is not a translation device. It's just a trick performed by an electronic bulldozer allowed to steal other people's work. But if you have a more open mind, GT suggests something else...

GT is also a splendidly cheeky response to one of the great myths of modern language studies. It was claimed, and for decades it was barely disputed, that what was so special about a natural language was that its underlying structure allowed an infinite number of different sentences to be generated by a finite set of words and rules. A few wits pointed out that this was no different from a British motorcar plant capable of producing an infinite number of vehicles each one of which had something different wrong with it -- but the objection didn't make much impact outside Oxford. GT deals with translation on the basis not that every sentence is different but that anything submitted to it has probably been said before. [pp.256-257]

For "words and rules," Bellos could read an entire book by linguist Steven Pinker simply called Words and Rules, The Ingredients of Language [Basic Books, 1999]. However, we have reached a point here where Bellos more than implicitly sabotages his own thesis; for he begins his book by observing that every translation of the same source is different and that, "The variability of translations is incontrovertible evidence of the limitless flexibility of human minds" [p.9]. That is exactly what this "great myth" of modern language studies means. While it may be true that "anything submitted to" Google Translate "has probably been said before," Bellos cannot be saying that nothing new can be said. And, if new things can be said, what allows for this? It is, indeed, that with language the "underlying structure allowed an infinite number of different sentences to be generated by a finite set of words and rules." That is where the "limitless flexibility of human minds," at least as expressed in language, comes from.

So this is a case where, as I have said, Bellos's argument refutes his own thesis. Thus, what Google Translate does, which is to search a vast database for examples of where a submitted text has already been translated, is indeed a "trick" and not a real translation, stealing the translation work that someone else has already done. If this works, that's just fine, but it doesn't help if Bellos's argument is that it is a "fantasy to which many philosophers [like Jerrold Katz] still cling" that "language really does consist of words and rules and that meaning has a computable relationship to them" (as Katz and Pinker say).

Indeed, one wants to ask Bellos how new thoughts or new literature are generated by human beings. If someone, like, say Albert Einstein, develops a new idea, he then has the challenge of putting it into words. This is not always easily done, or done in what may be regarded as a satisfactory way. Having expressed our new thought one way, we may decide it needs to be restated. The new statement is not a "translation" of the old one, since the old one was defective in conveying the desired meaning. We want a new and better expression of the pre-existing, pre-linguistic thought. Bellos does not allow for the possibility of this, and he rejects the philosophical view, about the real existence of meaning, that allows for it. If "translation is meaning," then Einstein had nothing from which to have translated his original thought.

We also must note that what "a few wits pointed out" is wholly irrelevant to the case. The British motorcars, "each one of which had something different wrong with it," does not correspond to the case of new sentences generated that are grammatically correct, since they obey, ex hypothese, the rules of the language. So the "wits" did not understand what was being claimed; and it does Bellos no credit to cite this as demonstrating anything.

Bellos also says: know a language is to know how to say the same thing in different words. That is precisely what translators seek to do. Roget's wonderful Thesaurus reminds them that in one language as well as between any two, all words are translations of others. [p.101]

Thus, if we produced a translation of a text that was identical to earlier translations, as all the translators of the Septuagint were said to have done, everyone would reasonably suspect us of plagarism. And if we are being tested on our understanding of something in a class at school, and we merely repeated what the textbook or teacher or some other source had already said, we would reasonably be suspected of not understanding the matter. If we understand something, what we have "in mind" is not any particular linguistic expression, but something that can generate novel and multiple linguistic expressions. We put the matter "in our own words." The thesis of Bellos that "translation is meaning" cannot account for this, since the novelty that is sought is not waiting there in the previous expressions. The novelty lies in the protean underlying nature of the meaning, which is grasped by understanding, just as we are prohibited in these cases from saying something that has familiarly "been said before." It may be that "all words are translations of others," but the words themselves (which don't exist according to Bellos) bear nothing to show that other words have something to do with them. Even the Nominalists realized that there must be a "third thing," the objects of reference, that relate the words to each other. They thought that was sufficient to account for meaning, despite the particularity, rather than the required generality, of tangible objects.

Thus, if there is some actual philosopher or linguist who has gotten it all right, according to Bellos, we never do quite hear who that is. Wittgenstein and Quine see natural languages as involving what may be called incommensurable paradigms, which means faithful translation is not really possible. The philosopher he does favorably cite, Jerrold Katz, for whom "effability" and translatibility are a reality, bases his doctrine on a metaphysics that apparently is a "fantasy" and a "myth." But Bellos is not through with promoting thinkers who reject his regard for translation:

Throughout the twentieth century, the Saussurean doctrine of the sign provided a reason for disregarding translation and ignoring the resources it gives for understanding how languages are used... [p.220]

As we have already seen in these pages, Ferdinand de Saussure is one of the people whose idea of language, like Wittgenstein's, is of something, a system, self-contained, self-referential, and autistic. This provides no sustenance to the translator, as Bellos recognizes, but he fails to see the deep contradiction that is really involved between Saussure's kind of theory and what is needed for the meaningfulness of translation.

Ferdinand de Saussure...after his early death in 1913, students... produced a Course in General Linguistics, which has served as a breviary for much of the thinking about language that has gone on since then...

His account was grounded in what was then a revolutionary new definition of the linguistic sign. A sign possesses both a material existence as a string of sounds or written marks, which he called the "signifier" (in French, signifiant); but it necessarily also has a power to mean -- a "significandum," or signifié. The sign is neither a signifier nor a significandum, but their combination, in a pairing so tight that the one can no more be separated from the other then the two sides of a single sheet... Finally, the inner relationship between the signifier and the signified that makes the two together coalesce as a sign has to be arbitrary...

A language is then nothing other than a system of differences, because a sign in any language is exhaustively defined by all the things that it is not... [pp.218-219]

If a "sign" is the combination of "signifier" and the "signified," then, if part of the function of a sign is to refer to material objects in the world, which are then part of the "signified," this means that material objects in the world are all parts of "signs." This is inconsistent with anything like a commonsense or reasonable definition of a "sign." And de Saussure's linguistics, as it happens, involves a systematic confusion of meaning, sense, reference, and reality -- or perhaps reality is just ignored. This is agreeable to the Nihilists who are drawn to de Saussure, who want reality to be "constructed" according to "power relationships," which means that truth and reality do not exist independently, and Joe Stalin, Fidel Castro, or Barack Obama can make reality be whatever they want. Indeed, the thesis that language is nothing but "a system of differences," eliminates, deliberately, any positive reference that would ground objective truth. Since the "system of differences" is self-contained and self-referential, this is why languages cannot really be translated. It is also why the theory is falsified by something correctly noted by Bellos:

Every speaker of a native language knows that she can (and frequently does) break the "rules" of grammar... A classic conundrum that computers could not solve is to attribute the correct meanings to the words in the following two sentences: "The pen is in the box" and "The box is in the pen." Understanding them calls on knowledge of the relative sizes of things in the real world (of a pen-sized box and the sheep pen, respectively) that can't be resolved by dictionary meanings and syntactic rules. [p.215]

The ability to break the rules of a language, and to commonly do so, refutes theories of language like those of Wittgenstein and de Saussure. Languages such as Pidgin languages exist without operative rules of grammar. The rest of the quote here also involves an error. Knowing that a "pen" can be something small to write with, or something large to hold farm animals, is a truth that you indeed find among the "dictionary meanings." Which is which in a particular sentence depends on the context. The context can be in the real world, as when you are standing in the barnyard, or it can be textual, which means it depends on the other sentences in the text that the translator is looking at. The "relative sizes of things in the real world" is a feature that is actually going to be tangential and secondary to the meaning that is gathered from the context.

As it happens, there is another point where de Saussure doesn't fit in with what Bellos wants:

...Ferdinand de Saussure's Course in General Linguistics, which, despite its profound innovations, firmly maintained the long tradition of treating language as the dress of thought...

I don't know whether language is possible without thought -- on the face of it, it must be, since so many people speak without thinking -- and I wouldn't dare contribute to the unending argument about whether thought is possible without words. The sole contribution I feel confident of making is to say that assimilating all uses of language to translation on the grounds that all speech is a mental translation of thought seriously diminishes our capacity to understand what the practice of translation between languages is about. [pp.313-314]

So de Saussure is off the reservation when it comes to Bellos's Nominalism. Why de Saussure is right here and Bellos wrong is displayed in the remarks of Bellos himself that follow along. Language is not possible without thought, if thought is that through which we comprehend meaning. But Bellos doesn't like either thought or extra-linguistic meaning; and of course when we say that people are "thoughtless," this does not mean they are automata with no minds, but just that they speak without sufficient thought or understanding. And of course thought is possible without words, which is what happens when anyone has an idea that they have difficulty putting into words, as I have previously noted. Sometimes people carry around new ideas that it takes them years to express satisfactorily. It is Bellos, in turn, who wants to assimilate "all uses of language to translation," which is dubious, while the thesis that "all speech is a mental translation of thought" does not "seriously diminish" anything, and in fact provides the only explanation of the "practice of translation" that is going to make any sense. But de Saussure had already scuttled the possibility of translation, whatever he believed about thought; and Bellos continues to embrace philosophy that makes his translation project impossible, while rejecting the linguistics of people like Katz and Pinker, for whom the reality of translation is obvious and natural.

Playing around with relativistic and nihilist theories of language, Bellos is in turn attracted to even more venerable self-referential theories of language, those of Whorf and Sapir, which I might mention have recently been ably debunked by linguist John H. McWhorter in The Language Hoax, Why the World Looks the Same in Any Language [Oxford, 2014]. This also turns out to be a case, again, where Bellos's argument refutes his own thesis:

One example of the radical difference of human languages was made famous by the American linguist Benjamin Lee Whorf, who had learned and studied many Native American languages. In the language of the Hopi (but also in quite a few others, distributed with no obvious pattern around the globe), there is a grammatical category called evidentials. For each noun-phrase, the grammar of Hopi marks not so much the categories of definiteness or indefiniteness ("a farmer," "the farmer") but whether the thing or person referred to is within the field or vision of the speaker. "The farmer I see" has a different form from "the farmer I saw yesterday," which is different again from the form of "the farmer you told me about." As a result, the English sentence "The farmer killed the duck" is quite untranslatable into Hopi without a heap of information that English sentence doesn't give you...

The use of Hopi-type grammars as evidence of the untranslatability of tongues is really a red herring. [p.161]

So our first impression is that Bellos does not endorse Whorf's theory. The self-contained nature of Hopi grammar does not make translation impossible. Indeed, the information from the evidentials is easily rendered into English versions of Hopi sentences; and as we know from the history of other languages, evidentials can begin as optional particles or words that gradually become "grammaticalized" and regarded as essential to grammatical statements. Translating in the other direction is a little more challenging, since an English sentence or its context may not contain the information needed for the evidentials in Hopi. But the Hopi would occasionally also need to deal with situations where they are ignorant of what is needed for the evidentials. If the third possibility above is hearsay, this can cover a lot of ambiguous ground. When evidentials become too cumbersome, of course, the language changes by breaking the rules -- a direction of development that we have already seen Bellos acknowledge.

Different languages, because they are structured in different ways, make their speakers pay attention to different aspects of the world. Having to mark presence or absence in language that have evidentials... or being obliged to mark time in the languages of Western European type, lays down what he [Edward Sapir] called mind grooves -- habitual patterns of thought. The question for translation (and for anthropology) is this: Can we jump the grooves and move more or less satisfactorily from one "habitual pattern" to another?

The view that you can't ever really do this has become known as the Sapir-Whorf hypothesis, despite the fact that Edward Sapir never subscribed to the idea. The trouble with the simple form of this misnamed prejudice -- that translation is impossible between any two languages because each language constructs a radically different mental world -- is that if it were true, you would not be able to know it... There would then be an area of "thinking in French" that was "ineffable" in any other tongue. That contradicts the axiom of effability, which, as we argued in chapter 13, is the sine qua non for translation to exist. [p.165]

So we still have Bellos denying the thesis that a grammar "constructs a radically different mental world," a theory he is willing to attribute to Whorf but not Sapir. But his discussion of Sapir already has something not quite right. The "mind grooves" Sapir and Bellos are talking about, can only be supposed to be the habits induced by grammatical rules (where Bellos doesn't really believe in grammatical rules, or words, especially for languages like Hopi, which had no writing in their pre-Columbian state). If we can "jump the grooves," this only means that thought (about which Bellos is also skeptical) is not really bound by the requirements of grammar. Jumping the grooves means that we can break the rules, or at least think about it. Bellos has no problem with this. But, for all his sensible observations so far, Bellos skids off the rails in what follows:

If Plato had had Hopi to think with, he would not have come up with Platonic philosophy, that's for sure -- and that's probably not a merely retrospective illusion based on the observable fact that there's no Hopi speaker who thinks he is Plato. Hopi thinkers think something else. That does not make Hopi a primitive language unsuited to true thought. It means that speakers of what Sapir called "Average West European" are poorly equipped to engage in Hopi thought. To expand our minds and to become more fully civilized members of the human race, we should learn as many different languages as we can. The diversity of tongues is a treasure and a resource for thinking new thoughts. [p.166]

Here suddenly we have Bellos saying that the grammar of particular languages is going to restrict or expand thought, and that for "thinking new thoughts" we may need the "resource" of other languages. This is suddenly the Vengeance of Whorf and Sapir, whose strong thesis we thought Bellos had rejected. And in the reference to Hopi, Bellos has actually gotten the grammatical situation backwards. The "Average West European" would have no difficulty adding citations of evidence in his statements, which would render him perfectly well "equipped to engage in Hopi thought." It is the Hopi who might be at a disadvantage, in dealing with forms of discourse, like Platonism, where the evidentials are not well suited to the concepts. What Hopi thinkers have thought about are things from Hopi culture and geography. The anthropologist, in any language, is as well equipped to think about those things as anyone else, as they do. If we actually need to "learn as many different languages as we can," because this is the only way we can have new thoughts, this contradicts the "axiom of effability," which is that any thought can be expressed in any natural language. Bellos seems to have forgotten his own foundational claim about translation. And, if he is expressing what Edward Sapir thought about this, then he cannot dismiss Sapir from adherence to the Whorfian Hypothesis. But there is something suspicious here. Since Bellos rejects the metaphysics that can account for novelty and creativity, perhaps here he is retreating to a claim that novelty and creativity come from learning other languages and thinking in their grammar. This is not going to explain much, however, since we find a lot of creativity in people who only know one language, especially the Greeks who thought that other languages only belonged to .

Bellos becomes rather heated over the idea that "poetry is what is lost in translation":

... toward the end of 2007 there were 666 Web pages in English that quoted the adage "poetry is what is lost in translation"... Like so many other received ideas about translation, this one turns out to have no foundation in fact. [p.149]

However, Bellos again manages, more by example than argument, to refute his own thesis. First of all, we might consider what it is about translation that most people would take to be what results in poetry being lost.

I dare say that most people would think of poetry as involving rhyme. Limericks about Nantucket all rhyme. The more sophisticated reader might see meter as more fundamental than rhyme. Homer doesn't have much in the way of rhyme, but the meter of the Iliad and Odyssey is why the Greeks called him a poet. Meter is probably what helped make it possible for long poems, like the Iliad, to be memorized in an oral culture. The most sophisticated modern, of course, knows that both rhyme and meter are unnecessary and that poetry can be "free verse" or just words (or something else) scattered across a page. A large part of the public however, probably thinks of that kind of poetry as part of the insanity and absurdity of modern art. Any idiot can babble a bunch of words about anything, like modern artists doing no more than fingerpainting; but rhyme and meter require real skill, and a real poet thus has remarkable abilities.

So let's start with rhyme and meter in place. In translation, as the words of a poem are replaced by words from another language, where the new words are mostly going to be very different from the originals, in size, sounds, and stress, the rhyme and meter straightaway are going to be lost. This is what Bellos fulminates about as the fictitious "literal" translation. But it is the easiest and most natural translation to do; and in fact there are many prose translations of the Iliad and Odyssey that make no attempt to restore the meter.

Aye, there's the rub. What is lost in the translation must then be restored. And by citing multiple poetic translations, Bellos doesn't seem to realize that he is refuting his thesis rather than confirming it. The translation is something that must be put in poetic form, and the translator is required to have very different skills from that of a mere translator -- who may translate prose beautifully, but be hopeless as a poet. To translate a poem and have it be a poem, with the same linguistic characteristics -- i.e. rhyme, meter, or whatever, hopefully involving some beauty of language -- the translator himself must be a poet. If the translator is not a good poet, the result cannot be a poem of the same quality, even remotely, as the original. Thus, Edward FitzGerald's translation [1859, 1889] of the Rubáiyát of Omar Khayyam is itself a great poem, but most experts are dubious about its faithfulness to the meaning of the original poem [note]. It may not make much difference. A more faithful translation, by a hack writer, can easily be insufferable. And this is where Bellos really goes wrong, in a way that turns up elsewhere in his book. Being a good translator and being a good poet are different things. Each may entirely lack the skills of the other, or combine them with varying degrees of success. If poetry is lost in translation, this just means that the translator, to produce a good poetic translation, must in fact be a good poet, or be working in collaboration with the good poet -- a good poet who must take liberties in obedience to his own aesthetic sense. Once the words and language change, because of the most minimal meaning of translation, a new task arises for a poetic translation.

A problem similar to that with poetry comes in for consideration later in the book. However, we should note first one of the earliest statements that Bellos makes:

It's a well-known fact that a translation is no substitute for the original.

It's also perfectly obvious that this is wrong. Translations are substitutes for original texts. [p.37]

This is not true, much less wrong in a "perfectly obvious" way; but with a bit of patience we find that Bellos will tell us why it is wrong himself -- as we may have come to expect.

...the babysitter who coos at an infant in a crib is performing a language act of the same general kind as the ambitiously student who greets me with a rising tone on the last syllable of "Good morning, sir." If these acts are communicative, then we must redefine communication not as the transmission of mental state from A to B (and even less a transmission of "information") but as the establishment, reinforcement, and modification of immediate interpersonal relations. But it would be better to say: that's not communication, that's a language. Language is a human way of relating to other humans...

...language is ethnicity. [pp.333-334]

This is strange enough in its own right. Cooing perhaps can be accepted as linguistic; but if Bellos is defining language as anything involved in humans "relating to other humans," including "ethnicity," this is absurdly broad. It makes a punch in the face in a bar fight into "language." And it is particularly odd for Bellos to like such an idea, since most of the universe of "language" would then not make the cut for the kind of language, the subject of Bellos's book, that can be translated. Much of his book seems to involve sabotaging his own project. Bellos even recognizes what he has done:

Translation deals with most of those other things. It does not and cannot attempt to perform or mimic or replicate the interpersonal functions of human speech...

If you're looking for the ineffable, stop there. It's blindingly obvious. It's not poetry but community that is lost in translation. The community-building role of actual language use is simply not part of what translation does.

But translation does almost everything else. It is translation, more than speech itself, that provides incontrovertible evidence of the human capacity to think and to communicate thought. [pp.337-338]

It sounds like "almost everything else" really does not amount to much in the universe of human "language." At least he does admit that something rather big is "lost in translation." Perhaps this why he is so vehement that things like poetry and style are not lost in translation, even when he demonstrates convincingly that the translator in fact must reinsert them into his translation.

On style, we get statements like this:

...The circulation of novels among all the vehicular languages of the world and their incontestable conversations with one another demonstrate without a shadow of doubt that style does survive translation. The means that translators use to ensure this are not more than the common skills used in all translation tasks.

In sum, the widespread notion that style is untranslatable is just a variant of the folkish nostrum that a translation is no substitute for the original. There is no more truth to it than there is in the idea that humor can't be preserved by rephrasing in the same or another tongue. [p.290]

Of course, the circulation of the finished products of translation proves nothing about what is preserved in translation, since it glosses over how the product is achieved. But a lot is subsumed under the "common skills" used by translators. As with poetry, a translator who does not write well, and is not any kind of stylist himself, is not going to be able to produce a stylish translation, whether or not the resulting style has anything to do with that of the original. A translation as a stylistic achievement in its own right goes way beyond the "common skills" of any writer. The "the widespread notion that style is untranslatable" is no more than a rhetorical straw man when we see the magician's slight of hand, which is the assumption of Bellos that it is a small thing, "common skill," for a translator to reproduce either poetry or style in a translation.

As for the "the folkish nostrum that a translation is no substitute for the original," Bellos ends up, in the end, admitting this himself. When it comes to "literary translation," perhaps Bellos realizes that what most translators produce probably cannot match the qualities found in great literature. Of course, this is part of the more general problem, which Bellos has tried to discredit. Certainly, by looking at Plato in Greek, we occasionally find that the translators don't always quite understand his point, or the grammar(!) of the text, and botch the translation.

The truth of literary translation is that translated works are incommensurable with their source, just as literary works are incommensurable with one another, just as individuals readings of novels and poems and plays can be "measured" only in discussion with other readers...

What counts as a satisfactory match is a judgment call, and is never fixed. The only certainty is that a match cannot be the same as the thing that it matches.

If you want the same thing, that's quite all right. You can read the original. [pp.308-309, boldface added]

So now we learn that, after all, "It's a well-known fact that a translation is no substitute for the original" [p.37] is not just as "folkish nostrum" but a hard truth about translation. But some care must be taken with this truth, lest we fall into the same incoherence that overtook Whorf, Sapir, and Bellos himself. That a translation is no substitute for the original does not mean that there is something "ineffable" in the language of the original text. It only means that, as Bellos admits to "the variability of translations" [p.9], and that "to know a language is to know how to say the same thing in different words" [p.101], and as translators certainly read any text in their own way, any particular translation will have difficulty conveying the whole or the ambiguities of the original meaning, certainly not as everyone is going to understand the original meaning. Thus, if no text is "ineffable," and if no translation is a substitute for the original, this just means that new translations are always warranted. And this should be welcome news to David Bellos, just as Gödel's Incompleteness theorems were a kind of good news to mathematicians, since it means that there will be no end of work for translators (or mathematicians). New translations, of course, can actually be worse than the old, which is bound to happen.

There are examples in the history of translation when translators really did believe that a translation was a substitute for the original. In translating Buddhist texts brought from India to China, Chinese translators sometimes figured they could then just throw away the original text. After all, Chinese is the best language; and, once rendered into Chinese, the original just became redundant and unnecessary. Fortunately, this was not always done, and different Chinese translations of some key works were later made [note]. We see a similar attitude, however, at work in the great Library of Alexandria. The Library was intended to contain all the books of the world. In Greek, of course. The Library retained no non-Greek originals. However, although books could legally be seized for copying (if in Greek) or translation (if not), the policy was to return the originals. Which was nice [note]. The most famous example of this kind of thing was the translation of the Hebrew Bible into the Septuagint Greek version. Since this translation was done by religious Jews and reflected their contemporary understanding of what the text of the Hebrew Bible meant, the translation is a key bit of evidence in understanding the original text. In fact, Bellos touches on a relevant case.

In the Book of Exodus, God has addressed Moses and told him what he must do to bring the Israelites out of Egypt. Moses figures that the Israelites are going to want to know where he has gotten these instructions, and he asks what he should say to the Israelites about it. In other words, who is this God he was talking to?

God famously answers in Hebrew, , 'Ehyeh asher 'ehyeh [Exodus 3:14]. Bellos quotes the King James Bible rendering it, "I am that I am." However, The Interlinear NIV Hebrew-English Old Testament [John R. Kohenberger III, Zondervan Publishing House, 1979, 1987] translates it, "I am who I am" [p.151]. As it happens, this disagreement exposes an important ambiguity. can mean "which, that" or "who." This problem is not addressed by Bellos.

What interests Bellos is the translation done by Martin Buber and Franz Rosenzweig:  "I will be-there howsoever I will be-there" [p.179]. Bellos gives us the original German of the translation in a footnote:  ich werde dasein, als der ich dasein werde [p.348 15:9]. They were attracted to the idea that the original Hebrew meaning should now seem rather strange, as this does. However, it seems stranger in English than in German, which raises a question about the faithfulness of the English translation Bellos gives in the text. The simple imperfect form of the verb "to be" in Hebrew here becomes a future (with the auxiliary German verb werden, "become, come to be") with a locative reference, "being there," from German dasein. However, werde doesn't always mean a future tense, and it is often used just about in the sense of "is." So the future tense translation may not be warranted. Also, using dasein for "exist" is much more natural in German, and the equivalent as "being there" isn't really used in English and does not occur in the Hebrew original. So the phrase, "I will be-there," displays translation artifacts of German, not of Hebrew, that really should not be carried over into an English translation. If this is not the right idea, we can't blame it on Buber and Rosenzweig, unless they did the English translation also. There is a similar problem with the English translation of as "howsoever." This does not seem much more informative than "who"; but it isn't what the German says. Als can just mean "as," and we notice that der, "the," has been dropped from the translation. So we could actually translate the German text of Buber and Rosenzweig as, "I am being [to-be? existence?], as I am the being [to-be? existence?]." This is still challenging enough, but in a different way from the English that Bellos gives, and it sounds more like the issues of the older translations, whether the Greek or the King James.

Regardless of what Buber and Rosenzweig intended, "I am that I am" and "I am who I am" are very different. The latter would lead to a follow-up question, "Then who are you?" (or "Howsoever are you going to be[-there]?"). The former would seem to say that God is his very existence (dasein, as it happens). This is still paradoxical, but the follow-up question is more, "What does that mean?" than "Then who are you?"

This is where the Septuagint may help. The Greek text is , egó eimi ho ôn, "I am the being [one]." Here God says he is the (active participle) being, for which we may or may not supply a noun, like "one." The interpretation that God is his very existence is reinforced by this Greek rendering. Nothing "who," "how," or "as" about it. This is reinforced when God then says, "This is what you are to say to the Israelites: 'I AM has sent me to you'" [Kohenberger, op. cit.]. Here "I AM" in Hebrew is our familiar , "I am," and the Greek is , "The being [one]." Since the Greek article may be dispensed with in English, an English translation could just be, "I am Being," which is something quite meaty enough for the theologians or philosophers.

There is a source for an alternative reading of . The Vulgate Bible translation of Exodus 3:14 is ego sum qui sum, which quite clearly is, "I am who I am". And God then says to tell the Israelies that qui est, "he who is" or "he who he is," sent Moses to them. "Who," of course, is a faithful (alternative) translation of . On the other hand, this translation is by St. Jerome, a Christian in the 5th Century AD. He was familiar with the Hebrew Bible and the Septuagint, and was living in Jerusalem, probably in contact with Jews. So we cannot dismiss his translation out of hand, but it is remote in time (700 years) and sense from the translators of the Septuagint.

Either way, of course, the translations are indeed no substitute for the original, which the devout Jew or Christian might do well to read in the original, as observant Jews are almost certain to actually do.

Another "lost in translation issue" that comes up in Bellos involves translating jokes, which often involve puns or other coincidences of pronunciation or meaning that can disappear completely from one language to another. Bellos of course resists the idea that anything can be lost in translation, so we have only grudging acknowledgement that this can happen, together with his characteristic response that if humor can somehow be restored to a joke, this is therefore a "translation." In the course of this, however, Bellos stumbles into some philosophically interesting issues about reference. By "metalinguistic" here, Bellos simply means the arbitrary physical characteristics of words and sentences that are the grist of things like puns but that are tangential or irrelevant to the function of language in expressing meaning (if Bellos believed in meaning which is not just the physical characteristics of the utterances, as where "translation [the physical sentences] is meaning").

Just as only some jokes exploit the metalinguistic function of language, so not all self-referring expressions are funny. Especially not those used as example sentences by philosophers of language, such as:

1. There are seven words in this sentence.
It is no trouble to find a matching sentence in German:

2. Es gibt sieben Wörter in diesem Satz.

However, that particular translinguistic match is regarded as a happenstance -- an arbitrary and irrational coincidence in a particular case. [p.277]

"Arbitrary and irrational" are kind of strong words for something that is merely accidental. No one should be surprised that translating a sentence into a different language produces something with a different number of words in it. Indeed, since Latin doesn't have articles, an English sentence with one or more definite or indefinite articles in it almost always corresponds to a shorter sentence in Latin. Similarly, a language that does need to use pronouns, as Latin does not (because person and number are shown by the form of the verb), but English does, will characteristically result in shorter sentences. Because of its case system, Latin also doesn't need to use prepositions as often. So, we say ars gratia artis in Latin, but "art for the sake of art" in English, three words into six.

But nothing beats Chinese. One four character saying in the Tao Te Ching, , "Heaven Earth not kind" [5:14], in one translation comes out as "The heavens and the earth are not partial to institutionalized morality." The succinctness of Classical Chinese and the obscurity of Taoism combine here to turn four words into eleven, although, to be sure, the translators have gotten rather carried away with it, producing something both prolix and, as it happens, inaccurate.

The strong words that Bellos uses, "arbitrary and irrational," makes it look like he's getting ready to be indignant over something here. Indeed, we know already that he doesn't have much regard for "philosophers of language," even for Jerrold Katz, who affirmed the fundamental axiom for Bellos's whole view of translation. He continues:

What's usually seen as problematic about sentences such as (1) is that they cannot reliably be translated into other tongues, and they thus appear to contradict the axiom of effability -- that any thought a person can have can be expressed by some sentence in any natural language, and that anything that can be expressed in one language can also be expressed in another (see chapter 13).

The real problem with a sentence such as (1) is that it can't be translated into English, either. "This sentence consists of seven words" rephrases ("translates") (1), but by doing so it becomes counterfactual, which (1) is not. Likewise, rephrasing it in French produces an untruth if you think that translation means matching signifiers one by one with equivalents provided by pocket dictionaries:

(3) Il ya sept mots dans cette phrase.

The main cause of problems is solutions, an American wit once declared, and the conundrums created by rephrasing self-referencing sentences taken out of any context seem to be good examples of that. [p.277-278]

Actually, the "context" here is not at all the problem. A sentence that refers to itself but is translated into a different language is no longer referring to the same sentence. "Context" is what is around what we are looking at, not the thing itself. Statements that refer to sentences pose no problems in translation, unless the reference is something that itself thereby changes, as happens with self-referential sentences. Bellos continues:

That because (3) is not the only way you can express (1) in French. Indeed, it's just about the least plausible version you could come up with. A better match would be:

4. Cette phrase est constitutée par sept mots. [p.278]

What does Bellos think this proves? He has simply produced an alternative sentence in French about which the assertion of the original sentence still holds true. So what? When he says that the other sentence was "the least plausible version" in French, is that supposed to mean that a proper translation is one that it written so as to preserve the truth of the sentence? In these cases, however, what is philosophically interesting are just the cases that don't preserve the truth of the original sentence. There are many curious things about self-reference, and these cases simply bring the issue to our attention. Bellos continues:

But because philosophy is written by philosophers and not translators, the clash between (1) and (3) is taken to be a demonstration of a wider, general truth:

Translation between languages cannot preserve reference (what a sentence is about), self-reference (what a sentence says about itself) and truth-value (whether the sentence is right or wrong) at the same time. [p.278]

Well, to trade barb for barb, perhaps it is because Bellos is a translator rather than a philosopher that he misses the point. But it would be a poor philosopher to make the assertion here that Bellos attributes to them. The word "cannot" does not belong. In these sentences of English, German and French, the reference changes precisely because the reference is self-reference to the sentences themselves. Since "sense and reference" are independent features in the theory of meaning of Gottlob Frege, the meaning of the sentences changes if the reference changes. In translation between languages, self-referential sentences will indeed lose their original reference. Whether the new sentence retains its truth-value is entirely a matter of the chance of the translation. There is nothing particularly mysterious about this, except for the dislike of Bellos for things being "lost" in translation. But he cannot avoid the problem with puns and wordplay.

Thus, when Sir Richard Burton went disguised as a Muslim on the Pilgrimage to Mecca, one detail he mentioned was when praying at the tomb of the Caliph Omar ('Umar) in Medina. Early in his trip, Burton had made the mistake of representing himself as an Iranian. This meant that people expected him to be a Shi'ite. While he soon realized his mistake, the rumor of him being a Shi'ite followed him for a while. This may have been fortunate, since suspicions about him being a Shi'ite may have deflected suspicions that he wasn't a Muslim at all. Rumors were also circulating than an infidel Englishman was on the Pilgrimage, as indeed one was. Now, Shi'ites detested the Caliph 'Umar; and when praying at the tomb of 'Umar in Medina, as all pilgrims were expected to do, Shi'tes often addressed their prayer, not with , yâ 'Umar, "Oh Omar," as proper, but with , yâ h.umur, "Oh Ass." "Umar" and "ass" in Arabic are phonetically close enough, that the Shi'ites could often get away with this trick; but Burton noticed that his companions listened carefully while he prayed, lest he try it himself. It would be pointless for a translation to try for an equivalent, since it is easier and more appropriate just to explain what was going on in Arabic -- as Burton does himself.

This would explain in a nutshell why puns and plays on words and all those kinds of jokes that exploit specific features of the language in which they are expressed cannot be translated. Because this is presented as a general assertion, it can be disproved by a single persuasive counterexample. But the reason it is wrong is not contained in any counterexample. The flaw in the axiom lies in its failure to say what it means by "translate." So here's my idea of a better approximation to the truth about translation:

Arduously head-scratching, intellectually agile wordsmiths may simultaneously preserve the reference, self-reference, and truth value of an utterance when fate smiles on them and allows them to come up with a multidimensional matching expression in their own language. [pp.278-279]

The "persuasive counterexample" may refute "a general assertion," but the dispute here is really over, as Bellos says, the meaning of "translate." The need for "intellectually agile wordsmiths" who benefit "when fate smiles on them" is itself the proof case that producing poetic, stylish, or funny translations depends on the translators also being poets, stylists, and jokesters. The dull soul who is none of these things may be able to produce a grammatically fluent translation, but it is not going to consist of anything like the poetry, style, or humor of the original. It will be a literal translation, despite Bellos rising to a fury and saying, "The expression 'literal meaning,' taken literally, is a contradiction in terms, an oxymoron, and a nonsense" [p.109]. Bellos must have it that way if he wants to collapse and confuse the work of the translator with that of the poet, the stylist, and the jokester, as he does. The "persuasive counterexample" is not where the translator can "match" the poetry, style, or humor of the original, but where he doesn't.

There are some other substantive issues I could address in Is That a Fish in Your Ear?, but I will wrap things up with what looks like some of Bellos's better research. Early in the book, as Bellos was confusing the Babylonians with the Sumerians, he reproduces the Sumerian cuneiform for "translator," which was . In Sumerian this was read eme-bal or eme-bala; and John Alan Halloran's Sumerian Lexicon gives its meaning as "interpreter, translator," broken down into two signs, "tongue" and "to transfer" [Logogram Publishing, Los Angeles, 2006, p.60]. Bellos analyzes it as "language turner," which may (obscurely) amount to the same thing. Later in the book, as Bellos discusses the translation services of the Ottoman Empire, we get something even more interesting:

...the Ottoman translators, called tercüman. This Turkish term has come into English as dragoman, but in only sleightly altered form it can be found in dozens of other languages that had contact with the Turks. Azerbaijani trcümçi, Amharic ästärgwami, Dari tarjomân, Persian motarjem, Uzbek tarzhimon (), Arabic mutarjim, Moroccan Arabic terzman, and Hebrew metargem () are all sound translations of tercüman. But whether written as dragoman or as tercüman, the Ottoman word for "translator" is not a Turkish word at all. It is first found in a language spoken in Mesopotamia in the third millennium B.C.E., as a translation of the even more ancient Sumerian word eme-bal. Akkadian targumannu thus has a descendant by way of Turkish tercüman in an admittedly obsolete but still extant word of English... [p.124, vowels added to Hebrew and erroneous samekh, , corrected to final mem, ]

This quite fascinating passage nevertheless would leave one with the impression that Persian, Arabic, and Hebrew must, as languages "that had contact with the Turks," have all gotten their words for "translator" from Turkish tercüman (pronounced terjüman). But the Turks came rather late into Middle Eastern history, and the Ottomans later still. Turkish certainly got its word from Arabic or Persian. This is easier to see if we look at another word in Arabic. Bellos gives us mutarjim in Arabic and motarjem in Persian, both of which are written and derive from Arabic (which we know because of the morphology of the word, an active participle). But the root in Arabic is unusual, with four consonants instead of the typical three of Semitic languages, and it exists in another form that also means "translator," namely , turjumân. Ottoman Turkish would have wirtten tercüman as , just as in Arabic. This now looks all but identical to the Akkadian targumannu(m), "interpreter, dragoman" [A Concise Dictionary of Akkadian, edited by Jeremy Black, Andrew George, Nicholas Postgate, Harrassowitz Verlag, Wiesbaden, 2000, 2012, p.400]. Thus, while Bellos has some good research here, for a word that must be one of his favorites in all languages, he does not go quite deep enough -- while he confused Babylonians with Sumerians, makes a mistake in Hebrew, and short-changes Arabic vis à vis Turkish. Perhaps that sort of uneven quality is the story of the whole book.

Philosophy of Science, Lingusitics


Home Page

Copyright (c) 2016 Kelley L. Ross, Ph.D. All Rights Reserved

Is That a Fish in Your Ear? Note 1;
Claire North's Fractured Arabic

Bellos is not alone in fractured use of Arabic. I find, for instance, other examples in The Sudden Appearance of Hope, a novel by Claire North [Redhook Books, Orbit, Hachette, 2016]. Such a book is not a work of scholarship, so its faults may be excused. But this makes the error in Bellos look all the worse. Proof readers and editors for novels may not care about the accuracy of the use of Arabic, but the proof readers and editors for a scholarly book should.

On page 68 Claire North gives terms in several languages that mean "welcome," including Arabic. In Arabic, this should be , ahlan wa-sahlan, which literally means "a people and a plain," i.e. welcoming conditions for travelers. These words feature the interesting grammatical device that the ending is the -an ending of the indefinite accusative case, which is used to make adverbs. So "a people and a plain" is used adverbally, as , "behold!" is used in Greek.

However, what we see in North's book is , 'hl' wshl'. This is what happened in Bellos. The consonants are written left to right rather than right to left, and the independent, uncombined forms are used instead of the combined forms that are properly used in words in Arabic. Similar confusion occurs on page 125, where words for "word" are given. This would be of interest to Bellos. In Arabic we see , klmh. This is supposed to be , kalimah, "word." Now, I wonder if even people who do not speak or read Arabic, but at least are familiar with what written Arabic looks like, could themselves recognizes the expressions in North's book (or Bellos's) as peculiar. They actually don't look like real Arabic.

This might remind us that "multi-cultural" education usually pays no attention to the language, literature, or history of other cultures. Its goal is actually to each Marxism, usually without the slightest concern for the values or ideas intrinsic to world cultures and civilizations. Traditional scholarship of, say Islam, gets called by the multi-culturalists themselves "Orientalism," which is supposedly projecting Western values and misconceptions onto others. But the multi-culturalists do almost nothing else. For all the knowledge of Arabic and the Middle East that Claire North's book seems to display, it cannot get right a few of the simplest words and expressions in Arabic.

Claire North's Los Angeles

Return to Text

Is That a Fish in Your Ear? Note 2

Bellos has a rather absurd argument that talk about "faithful" translations is derived from sexism and slavery [pp.128-130]. This is perhaps of a piece with his idea that there is no such thing as a "literal" translation. However, a faithful translation, literal or otherwise, preserves the meaning of the original and neither leaves things out nor introduces extraneous material. In these pages, a good example of a translation that is not very faithful involves Friedrich Nietzsche. Thus, Francis Golffing translated a statement from "The Genealogy of Morals" as:

The Romans were the strongest and most noble people who ever lived. [The Birth of Tragedy and The Genealogy of Morals, Doubleday Anchor Books, 1956, p.186]

In German, this looks like:

Die Römer waren ja die Starken und Vornehmen, wie sie stärker und vornehmer bisher auf Erden nie dagewesen, selbst niemals geträumt worden sind. [Zur Genealogie der Moral, 1887, Philipp Reclam Verlag, Stuttgart, 1988, p.42]

As in the marvelous movie, Lost in Translation [2003, see epigraph], there is a whole lot more in the German (as in the Japanese of the movie) than we get in the translation. What is that other stuff? Although we can't really say that Golffing has altered the message of Niezsche's statement, the translation is neither literal nor faithful because of all that it leaves out, and obviously because of the very different way that it expresses Nietzsche's thought. A more literal and more faithful rendering would go:

The Romans were indeed the strong and noble, just as those stronger and nobler hitherto on earth never existed, never even would have been dreamt.

Like most literal, or reasonably literal, translations, I allow that this is a little awkward. But it follows the German sentence pretty much sense for sense -- sensum exprimere de sensu, "to express the sense for sense," in the words of St. Jerome [p.104].

Return to Text

Is That a Fish in Your Ear? Note 3

Or not always. Athens kept official copies of the plays by Aeschylus, Sophocles, and Euripides. King Ptolemy III wanted to borrow them for copying. Athens insisted that Ptolemy post a bond against the return of the manuscripts. The King kept the originals and foreited the bond. But at least he send copies back to Athens [cf. James Turner, Philology, Princeton, 2014, p.9].

Return to Text

Is That a Fish in Your Ear? Note 4

Bellos overlooks the tradition of Chinese translation. Thus he says;

Its rich history of word lists is essentially linked to the tradition of writing commentaries on ancient texts, not at all with the business of translating foreign languages, in which traditional Chinese civilization seems to have had as little interest as did the Greeks. [p.96]

This indeed ignores the long history of translating Buddhism into Chinese. Until Europeans showed up, it is not clear what else really existed, as substantial literary or religious documents, for the Chinese to translate apart from texts in Sanskrit from India. It is true that Confucians never had much regard for this material, but then they were not the ones interested in the translation. Chinese Buddhists, especially pilgrims like Fa-Hsien and Hsüan-tsang, were quite motivated enough.

I might also note at this point what seems to be an error in Bellos's treatment of Chinese:

It was not until the seventeenth century that a device for classifying Chinese characters in a way that made them easily retrievable was devised by the scholar Mei Ying-tso, a few years before the Jesuit missionaries produced the first Western-style bilingual dictionaries of Chinese... [p.96]

Bellos does not say what this "device" was, but Chinese dictionaries are organized by the system of "Radicals," based on the structure of most Chinese characters as "Radical and Phonetic." If this is what Bellos means, he seems to be wrong. Thus, James Turner notes:

In Shuo Wen Jie Zi (121 CE), the Han dynasty scholar Xu Shen invented the strategy of indexing Chinese characters by the root elements they shared, still basic to Chinese dictionaries. [Philology, The Forgotten Origins of the Modern Humanities, Princeton, 2014, p.1]

Since more than 1500 years separate the Later Han Dynasty from the 17th century (and the Ch'ing Dynasty), Bellos is rather far off the mark.

Return to Text