ŷ������

Ah, interesting reviews about translation are interesting.

Now (I'm thinking aloud here) why is the problem so acute when you're translating literature? It seems to me that the answer lies in the unusually strong constraints associated with the demands of literary text. Even requiring a text string to be a sexy joke is a strong constraint.

I think you partly miss the point. Literature isn't nearly as hard as either poetry or cartoons. Cartoons are, in effect, poetry. It must be far easier to translate a novel. Though, of course, harder again than something factual, I imagine.

JEH Smith last month wrote the most interesting 3 Quarks Daily post on this, also reprinted on his blog. Let me quote at length:

Prose is the (more or less) formally unrestricted use of natural language for the telling of captivating things about the world. The formal restrictions of poetry, by contrast, bring it about that whatever poetry says about the world, it is always also saying something about language. This means, among other things, that translating poetry is at least something quite close to writing poetry (unless we take as an example Nabokov's hyper-literalist translation of Evgenii Onegin, which was meant precisely to illustrate that a true translation of one language's poetry into another can only come out as prose). Someone who has translated a novel, by contrast, certainly could not be said eo ipso to have written a novel.

What language is poetry about? Generally, it is about the language it is in. In translation, in turn, poetry is about the limitations of the fit of one language with another. These two facts together mean that, in writing poetry, in contrast with prose (more or less), it matters what language one is writing in. I have become convinced, in fact, that good poetry, the best poetry, is the poetry that seeks to lay bare the essence of the language that serves as its medium. Now I understand that from a historical-linguistic point of view languages do not have essences, but are ever-evolving accretions of borrowings, local adaptations, creolizations and mishearings, but that does not change the fact that, in terms of expressive power, 'life', 'earth', and 'kin' sound closer to the soul of English than, say, 'vitality', 'terrestrial', or 'family'. I have thus also come to appreciate the extent to which the essence of English is Anglo-Saxon and Germanic, and to think that no one understood his task as a poet better than Seamus Heaney, when he undertook to translate Beowulf into modern English, in part, as he explained, to come to better know not just the source language, but also the target language.

One thing I have noticed in my attempts to translate poetry from French, Latin, Russian, and German into English, is that it is only with the last of these that I feel like I've ever obtained a result that could be called an equivalence, rather than a rendering or an approximation. This is achieved in large part when the source poem relies heavily on monosyllables, and when suitable monosyllabic, Germanic-rooted words can be found in English for their translation. For reasons that are not so hard to understand, German and English tend to retain a basic, shared vocabulary for the very most basic things, for the things with which they were familiar before the Romans, and then the Normans, came to tell them about art, science, morality, and so on. Those things are, namely, the things of nature, and this is why poets such as Rilke and George are often such a delight to translate, in a way that, say, Brecht would not be.

There is lots more to what he has to say, you can find it here:

I understand so little, but I appreciate the fine style with which it is written.

"One thing I have noticed in my attempts to translate poetry from French, Latin, Russian, and German into English, is that it is only with the last of these that I feel like I've ever obtained a result that could be called an equivalence, rather than a rendering or an approximation."

I love this point. Reading French-English translations and vice versa, I always feel that there is something missing.

Manny, have you checked out ? The translation question takes on a whole other dimension when it comes to a poem like Jabberwocky. I'd love to hear what you make of the different translations since you can read more than a few languages. Hofstader has an intriguing little exercise in his book Le Ton Beau de Marot where he asks several people to translate the same poem from French to English, and he shows all the different results and comments on the choices made.

You may find interesting some of William Morris' later writing, such as The Water Of The Wondrous Isles, where he is very conscious of the origins of words as he assigns them to different speakers.

Not: on reconsideration, I agree that it's not so much about translating literature. It's about translating while preserving strong formal constraints. "Being a moving epic love poem" is one of the strongest formal constraints I can think of, though the mad robot inventor in The Cyberiad would no doubt also want every word to start with an S.

Now I think about it, I believe Hofstader discusses the translation of La Disparition as well in Le Ton Beau de Marot... more strong formal constraints. Really must read that book.

Whitaker, I have a Quiz question about the French translation of Jabberwocky :)

I was recently discussing this while trying to explain why I had so many different translations of Rilke's Panther. Either the translation is literal and loses all the beauty of the language, or sounds lovely but doesn't really say the same thing at all. You, however, articulated it much better than I did.

Thank you! I do machine translation for a living...

You're a nerd mate...

... and proud of it!

Hahahaha....I'm envious.

Ted Hughes espoused a philosophy of literal translation of poetry. He used to "translate" poems in various languages by contemporary Europeans using the following method:

1) Get someone to do a word for word trans-literation for him.

2)Change a word here and there.

3)Publish it.

However, when he dealt with dead folks e.g. his versions of Classical Greek poetry/drama he took an extremely liberal approach. (I'm pretty sure nuclear explosions aren't mentioned in Ovid's Metamorphoses.)

Having read a fair amount of both types of his translations, I find that the former approach is almost uniformly worthless to me, where-as the latter contains some genius and is what got me interested in Classical literature...

On translating jokes and puns, the Asterix books come to mind. When faced with some witty reference to Moliere, Bell and Hockridge would not only not attempt to literally translate it but would do things along the lines of replacing it with a Shakespeare joke... Goscinny considered the English language versions funnier the than the original French in many cases...

I must look at some of his classical translations! They sound interesting.

I describe here a little experiment of my own in this department, contrasting the two styles of translation. I agree about the worthlessness of the first approach!

I recommend his Tales from Ovid: 24 Passages from the Metamorphoses as the place to start and following that his Oresteia which made me fall for Pallas Athena in an alarming fashion!

Alternatively, Selected Translations will give you an over-all feel for all his translation styles.

I might check out the Ovid! My Latin is insufficient to be able to appreciate the original, and I've never seen a translation I liked. Thank you!

You're every welcome! I will look out for your inevitable review!

As a lover of Poetry especially of Omar Khayyam, recently I have came across a book Called, Thus spoke Khayyam, and in my opinion that is a best translation so far I have came across, Thus spoke Khayyám is unique in the sense, which the Translators not only have tried to translate just Persian into English neither have interpreted what Khayyam have expressed in his poetry, but he has brought the language of Omar Khayyám and his unique expression. please try to check it for yourself. Thanks

That sounds very interesting! Thanks Ramy.

I thought some of the sentiments in the poem were Islamic! You see how much I know about it. I will change the shelf now, thank you.

I was struck by this in the Thousand and One Nights. They are all so fervently Muslim, but most of them also seem to be borderline alcoholics. I had no idea what to think! Wine has always been haram?

BTW, the lines from Khayyam most often quoted in English are "A loaf of bread, a jug of wine and thou" and "The moving finger writes/And having writ, moves on".

Nor all your tears wash out a word of it.

Elham wrote: "Manny wrote: "I thought some of the sentiments in the poem were Islamic! You see how much I know about it. I will change the shelf now, thank you."

He had a very multidimensional personality. I th..."

What does "haram" translate as?

'Tis all a checkerboard of nights and days/Where Destiny, with men for pieces plays/Moves hither, thither, moves and mates and slays/Then one by one, back in the closet lays.

Often quoted by chessplayers.

Manny, what a review. Just to keep the ball rolling, I know that it is an overkill to explain to readers the intricacies of Bayesian theorem, still it would be great to shed some more light and make it a bit more accessible for any average reader of this funny, insightful and thought-provoking review. I wouldn't get your point if I didn't know something about machine learning.
I don't know what's the best way to do it though. Just to start I'd explain what these arcane E and F and P characters stand for.
Something like, hey guys, we have the Noisy Channel Model used in conjunction with Bayes�� Theorem. We can use it for Farsi to English, and put it like that, P(E|F)P(F) = P(F|E)P(E), where F stands for Farsi, E - for English and P - for probability.
No wonder that P(F) is constant, since we already know in what language the text is (it is Farsi), so we can skip it. Now, what we have left is:
P(E|F) (probability that the input Farsi gives this particular sentence of English) is equal to
P(F|E) (probability that the speaker actually wanted to say it that way, given what I know of the way English sentences are and how they relate to Farsi) multiply by
P(E) (probability that evaluates the grammar-wise correctness of English words; by the way, we have lots of material for that so we shouldn't worry about this part).
After that you count all possible links between words in F and corresponding E.
Now, why I don't like it. I doubt that bag-of-words approach, which statistical MT is fond of, is what Hofstadter really meant in his Le Tombe.
Focus on individual word meanings may distort a translation, yet the problem of n-gram modeling is the sparsity that you will get for some translations. Besides, depending on the machine translation algorithm you use it may also be biased to some translation, and not the others.

Elham, most Persian poets, contemporaries of Khayyam, could easily express their thoughts using prosody. They used a very elaborated symbolic system for that, which relied heavily on Islamic ethos. In order to understand Omar some knowledge of Sufism and its tenets is necessary.
Sufism exchanged the corporeal features of all things for the spiritual ones. Lots of tropes add spiritual meaning for a usual word or an object. In case you are interested you could google more about the dichotomy of baten-zaher. The bāṭen, as understood by the early Basran mystics, is man’s inner self, the complex of emotions which stir his soul. ʿelm al-bāṭen consists of knowledge of ways to train the soul and is a psychic discipline attainable by anybody through his own mental effort. Shortly, a charming islam mysticism. I highly recommend to delve into it. Now, few words on wine:
"For instance, when, like Omar, they mention wine, they mean a knowledge of God, which, extensively considered, is the love of God. Wine, viewed extensively, is also love: love and affection are here the same thing. The wine-shop with them means the murshid i kiamil (spiritual director), for his heart is said to be the depository of the love of God; the wine-cup is the telkin (the pronunciation of the name of God in a declaration of faith as: There is no God but Allah), or it signifies the words which flow from the murshid's mouth respecting divine knowledge, and which, heard by the salik (the Dervish, or one who pursues the true path), intoxicates his soul, and divests his mind (of passions) giving him pure, spiritual delight." (c) and so on.
By the way, five or six years ago I translated a bunch of Ottoman and Old Azerbaijani sufi ghazels. I humbly believe I know something of what I was talking above.

... when he never ever drunk wine in his life
exactly )) rephrase it "how can someone talk about allah if he never directly experienced him?"
I believe that this analogy somewhat helps in understanding.
How could understand that how it felt drinking wine to use it as a source of inspiration?!!
Here's another one - the cups of dead clay, talkin, should be filled with wine to inebriate one's soul. Thus mourid becomes united with the wine, oh, pardon me, with the Allah.

Mystical Dimensions of Islam just the best researched book and cornucopia of facts on Islamic mysticism.

Oh, I see. You were talking about profane, earthly wine, meanwhile I abode on a higher plane, sorry the pun :)
You don't have to persuade me, since Annemarie Schimmel mentioned a couple of times, that actual practice of drinking wine was not that rare event after all, esp. during sema
Besides, as you recall, it's haram to drink but a drop of wine. It means that a drop of wine is allowed. Sometimes a couple of drops don't cause much harm.

Oh, you mentioned boys. Its inappropriateness notwithstanding, there is a mystical excuse, to my utmost surprise. Palace homosexuality was a usual thing. Besides, sometimes, during recitations, young boys were asked in to join a meeting. Their adolescent beauty served as a reminder of the Beauty. I understand it may sound out there for any true believers though, it was ok then.

Ehnaton, please forgive my geeky speculations. I am of course well aware that n-gram modeling is a poor approximation to the truth here... but alas, for the moment, it's pretty much all we've got. Someday, we will hopefully know more and be able to get closer to what poetry is actually about. I'm impressed by your understanding of this area, where I am completely ignorant :)

Elham, just for clarification. I'm agnostic rather than atheist.

By the way, you can use the formula Manny has provided in this review and calculate the probability of existence of God )) There is a trick though.

Ehnaton wrote: "Elham, just for clarification. I'm rather agnostic than atheist. By the way, you can use the formula Manny provided in this review and calculate the probability of existence of God )) There is a trick though"

There is indeed! I have seen this calculation done several times, and it's amazing how everyone gets exactly the answer they want :)

)))
That's right. Oh, this bloody "prior" thing.

That was interesting. Thanks for sharing.

I attended a talk at a conference a couple of years ago where someone actually had tried to use statistical machine translation to translate literature. He said it wasn't quite as bad as you'd imagine, a fair number of sentences were usable. Though I had my doubts.

I always thought (but could not persuade anyone else) it would be an interesting project to semi-translate some literature. You know, take something in a language one knew somewhat but not thoroughly and try to read it, and write out what you thought it said. If one couldn't figure out a reasonable-sounding translation, one must fill in the text to make a story that works.

The problem, I suppose, is that the result would only be interesting to a reader who knew the original language and the language being translated into.

I think that might very well work!

I'm kind of tempted to try it myself, though I should probably first finish the very dull piece of software engineering I'm in the middle of...

The surprising thing is how well machines often do when they can't understand the content; they just produce stuff that looks like other stuff they've seen. Remember too (as Plato pointed out) that poets generally can't explain where the poems come from.

Why would it be a disorder?

Anyway, I think this is at best a very partial explanation! I'm sure Plato would have torn it to pieces in twenty minutes :)

there is a book called "is that a fish in your ear". it's about translation. i suggest you read that, as you will enjoy it greatly.

aha! more info acquired!

Wow guys who reads these poems? I am kind of a beginner in Arabic ! And I thought I was good... Though seriously who uses these translations.. And where can I find them?

Safiya wrote: "Wow guys who reads these poems? I am kind of a beginner in Arabic ! And I thought I was good... Though seriously who uses these translations.. And where can I find them?"

Just search on Google and you will find many versions of the Rubaiyat. The first one I looked at was , but I think there are better ones.

Egad, Manny! Reducing the almost superhuman task of translating poetry to an algorithm? I suppose it can work OK when translating something that is deliberately literal (e.g. a news item a cake recipe or a scientific paper) but a piece of prose becomes another matter entirely. At the outset, one would need to grasp and internalize the overarching objective and mood of the piece. Then there remains the challenge of re-creating the flavor of the language (literal or not). It's unlikely that any one translation of a given work would satisfy all or even a majority of readers. I have for example yet to discover a completely satisfactory translation of the poetry of Anna Akhmatova, a particular favorite of mine. Regrettably, I haven't the patience to learn enough Russian to be able to enjoy the original. The Rubiyat appear to be deceptively straightforward, yet I fear that some of the atmosphere surrounding the original Persian becomes muted in translation. I doubt if any technical device can fix that.

Well Jim, obviously I don't think that today's crude statistical translation techniques are close to being able to do a good job of handling immortal poems. But even though it's an extremely rough and approximate model, it seems to me that there may be something in the analysis I suggest here. The more constrained the form, the less exact the translation does seem to be the general rule...

Interesting review Manny.
Omar Khayyam is a great person for Iranian; not only because he was a great poet but more because he was a mathematician. Even though now he is famous because for his poems (Robaieyat in Pesrian) mostly of Fitzgerald' translation. And yes; his translation is amazing! When I read Robaeyiat both in Persian (Farsi) and English I found he was combined some Robaeyi together. In fact he translated what he realized from Omar Khayyam. In other hand, Omar Khayyam was not Sufi (like Rumi). He was not the mystic at least in classic way rather he was a wise man; he always emphasized on every moment of life; it means he focused on living in Now...not the past...nor the future and it opposed with Sufism, Islam or maybe other religious beliefs. In fact he brought mysticism from heaven to earth; earthy mysticism. I think this is one of the most wonderful aspects of his works and maybe it' attractiveness for west; especially Edward Fitzgerald!

Thank you ZaRi! I have wanted to be able to read the Robaieyat in the original for a long time, and in fact it doesn't seem impossible that I will finally be able to do this - as you maybe saw, we have been developing online software to help people appreciate literature in languages they don't know well. Our initial experiment, with the Old Norse poem ��ö���ܲ���á, was very encouraging. Would you by any chance be interested in helping us do something similar for the Robaieyat? We could start by doing a sample of four or five verses so that you can see what's involved. You don't need software skills, just knowledge of Persian and English for adding the translations and ability to read poetry aloud for recording the audio.

Manny wrote: "Thank you ZaRi! I have wanted to be able to read the Robaieyat in the original for a long time, and in fact it doesn't seem impossible that I will finally be able to do this - as you maybe saw, we ..."

Sure Manny... it sounds interesting. It will be my pleasure to participate in this program.