In recent years, the technologies for processing natural language have been developing at a dizzying pace. Writing text and generating images from a description are examples that stand out and throw down the gauntlet to the skeptics.
But could another quiet revolution be on its way? In another important field, one whose history is as old as the story of the Tower of Babel? Is the technology finally here for truly high-quality translation, including to and from less widely spoken languages like Bulgarian?
It’s time for a most curious experiment. DeepL not only boasts of being “the world’s most accurate translator,” but plenty of people already use it in their daily lives. The platform, developed with the help of machine learning, claims it can translate far more fully than, say, the familiar Google Translate.
Well, now we’re going to throw it in at the deep end… which isn’t always the best idea for a robot. We’ll test what AI is capable of, not with everyday correspondence, but with high literature! Will we end up lost in translation… or will it become clear that translation is one more field where we can now rely on artificial intelligence? And the money spent on the kids’ private foreign-language lessons will turn out to have been thrown to the wind…
If machine translation interests you, come to a special event on the subject!
On December 11 (Sunday) at 7 p.m. in the marble foyer of the National Palace of Culture (NDK) in Sofia, we’ll talk about “The Translator: Human or Machine?” in a panel discussion that is part of the Sofia International Literary Festival.
More information here.
Let’s start with a few obligatory caveats. Bulgarian is definitely not among the most widely spoken languages on this planet (at least not yet), and we would certainly get better results between, say, English and Spanish.
One more thing: literary translation is undoubtedly an art. It would surely be fairer to give everyday tasks, such as translating letters. But… why make it easy for the one we’re testing? After all, that’s the only way we can be sure it really does a good job.
So, as a warm-up, we begin with a few classic English idioms. In English, if you say it’s “raining cats and dogs,” you mean it’s pouring buckets. Will our electronic hero catch on?
Original (idiom):
It’s raining cats and dogs!
DeepL:
A rain of cats and dogs!
Original (idiom):
I’m saving £50 a month for a rainy day.
DeepL:
I’m saving 50 pounds a month for black days.
Original (idiom):
Can’t you see you’re only chasing rainbows?
DeepL:
Don’t you see that you are chasing nothing but actual rainbows?
Oh no, no, my friend! A spectacular false start. It is definitely not raining dogs, and nobody is chasing rainbows (get it, get it, DeepL, it’s getting away!). In fact, out of the three English idioms, the electronic translator figured out only one—the rainy day, which it turned into the “black days” of the Bulgarian equivalent. The rest it translates mechanically, without giving us much meaning.
At this stage, it looks like we won’t have anything resembling the Babel fish any time soon… You surely remember it if you’ve read Douglas Adams’s cult-beyond-cult Guide: you pop it in your ear and it starts translating every language in the galaxy for you…
And speaking of Adams, let’s begin the real test. It would be rather inhumane (AI-unfriendly?) to make the algorithm translate Shakespeare; even the British themselves struggle to understand him (unlike Valeri Petrov, the poet whose Shakespeare is the Bulgarian canon).
On to the real test. Here and in all the quotations that follow, I include for comparison the most popular official Bulgarian translation, naming its translator.
Are you, like me, impatient to see how the student will do?
Original (Douglas Adams – The Hitchhiker’s Guide to the Galaxy):
For instance, on the planet Earth, man had always assumed that he was more intelligent than dolphins because he had achieved so much — the wheel, New York, wars and so on — whilst all the dolphins had ever done was muck about in the water having a good time. But conversely, the dolphins had always believed that they were far more intelligent than man — for precisely the same reasons.
DeepL:
For example, on the planet Earth, man has always believed that he is more intelligent than dolphins because he has achieved so much—the wheel, New York, wars and so on—while the dolphins have only wallowed in the water and had fun. And conversely, the dolphins have always believed that they are far more intelligent than man—for exactly the same reasons.
Professional translator (Sarkis Aslanyan):
For example, on the planet Earth, humans had always believed that they were more intelligent than dolphins, seeing as they had created so much—the wheel, New York, wars and so on—while the dolphins can’t do anything but splash about in the water and enjoy themselves. And conversely, the dolphins had always believed that they were more intelligent than man—and for precisely the same reasons.
Oho, pretty good! Of course, it failed to grasp that “man” here is collective. I’d also very much like to see dolphins “wallowing.” But the truth is that, in terms of meaning, a reader could understand “what the author is trying to say.” DeepL has at least handled the “grunt work.”
Is this too easy for it, perhaps? We move on to the aerobatics: the famous opening sentences of a classic, beloved novel, and this time we’re translating from Spanish.
Original (Gabriel García Márquez – One Hundred Years of Solitude):
Muchos años después, frente al pelotón de fusilamiento, el coronel Aureliano Buendía había de recordar aquella tarde remota en que su padre lo llevó a conocer el hielo. Macondo era entonces una aldea de veinte casas de barro y cañabrava construidas a la orilla de un río de aguas diáfanas que se precipitaban por un lecho de piedras pulidas, blancas y enormes como huevos prehistóricos. El mundo era tan reciente, que muchas cosas carecían de nombre, y para mencionarlas había que señalarlas con el dedo.
DeepL:
Many years later, standing before the execution, Colonel Aureliano Buendía remembers that distant afternoon when his father takes him to see the ice. Macondo was then a village of twenty houses of mud and reed, built on the bank of a river of clear waters, which flowed into a bed of polished stones, white and huge like prehistoric eggs. The world was so so recent that many things had no name, and to mention them, you had to point with your finger.
Professional translator (Rumen Stoyanov):
Many years later, before the firing squad, Colonel Aureliano Buendía would remember that distant afternoon when his father took him to see the ice. At that time Macondo was a village of twenty houses of mud and reed, built by the bank of a river of clear waters that rushed down a bed of stones, white and huge like prehistoric eggs. The world was so recent that many things had no name, and to mention them, they had to point at them with a finger.
“The world was so so recent,” stuttered word and all, plus an ending in the wrong grammatical gender in the Bulgarian, is unquestionably a blunder. And yet the overall content seems to be there, no doubt, even against the backdrop of the mistakes. It would certainly struggle more with some of the novel’s later sentences, the ones that run on for pages… but natural intelligence often stumbles over those too, especially in times when we don’t read long texts.
A charming detail: DeepL translates “prehistoric” as праисторически (praistoricheski), while the classic Rumen Stoyanov calls the eggs предисторически (predistoricheski), both meaning “prehistoric.” Who is right, I couldn’t say. But there will always be those little whims and touches in a translation that come from great translators, and we will hardly ever teach them to their artificial copies. Then again, maybe there’s no need?
Original (Leo Tolstoy – Anna Karenina):
Все счастливые семьи похожи друг на друга, каждая несчастливая семья несчастлива по-своему. Все смешалось в доме Облонских. Жена узнала, что муж был в связи с бывшею в их доме француженкою-гувернанткой, и объявила мужу, что не может жить с ним в одном доме.
DeepL:
All happy families are alike, and every unhappy family is unhappy in its own way. In the home of Oblonsky, everything gets mixed up. The wife reportedly found out that her husband has an affair with a French governess who was in their home, and reportedly announced to her husband that she cannot live with him in one house.
Professional translator (Georgi Zhechev):
All happy families are alike; each unhappy family is unhappy in its own way. Everything was in confusion in the Oblonskys’ house. The wife had learned that her husband was having relations with the French governess who had formerly been with them, and had declared to her husband that she could not live in the same house with him.
Here DeepL visibly got tangled up; evidently it isn’t much good at gossip and intrigue about French governesses. The first sentence, however, turned out pretty well…
Original (J. D. Salinger – The Catcher in the Rye):
If you really want to hear about it, the first thing you’ll probably want to know is where I was born, and what my lousy childhood was like, and how my parents were occupied and all before they had me, and all that David Copperfield kind of crap, but I don’t feel like going into it, if you want to know the truth.
DeepL:
If you really want to hear about it, the first thing you’ll probably want to know is where I was born and what my nasty childhood was like, how my parents were busy before they gave birth to me, and all that nonsense of the David Copperfield kind, but I don’t feel like getting into them, if you want to know the truth.
Professional translator (Nadya Sotirova):
If you really feel like hearing this story, then you’ll probably want to find out where I was born and how I spent my stupid childhood, and what my parents did before they made me, and whatever else, in a word—all that chaff of the “David Copperfield” sort, but I don’t feel like rummaging around in it.
A little more English, since that is surely the language our student has “read” the most in. How did it do here? Acceptably again, in my view; the reader could grasp the meaning of the text, even if at times it sounds wooden… or perhaps it would be more accurate to call it “ironclad”?
And how would the electronic translator cope with something even more untranslatable, namely poetry?
Original (Salvatore Quasimodo – And Suddenly It’s Evening):
Ognuno sta solo sul cuor della terra
trafitto da un raggio di sole:
ed è subito sera.
DeepL:
Everyone stands alone in the heart of the earth
pierced by a ray of sun
and evening comes at once.
Professional translator (Dragomir Petrov):
Man is alone upon the heart of the earth,
pierced by a single ray of sun.
And unexpectedly the evening falls.
In English:
Everyone stands alone on the heart of the earth
transfixed by a ray of sun:
and suddenly it’s evening.
For comparison, I also give an official English translation. With the caveat that I don’t know Italian either, I really like how the software “caught” the strong verb “to pierce.” From there on it’s clear that poetic rhythm isn’t its strong suit; it would have been surprising, even slightly frightening, if it had done a decent job in that direction.
Now then, my friend, let’s see you with children’s books. Not that this one is for children; in my opinion, not at all, but that’s another story.
Original (Antoine de Saint-Exupéry – The Little Prince):
Les grandes personnes aiment les chiffres. Quand vous leur parlez d’un nouvel ami, elles ne vous questionnent jamais sur l’essentiel. Elles ne vous disent jamais : „Quel est le son de sa voix ? Quels sont les jeux qu’il préfère ? Est-ce qu’il collectionne les papillons ?“ Elles vous demandent : „Quel âge a-t-il ? Combien a-t-il de frères ? Combien pèse-t-il ? Combien gagne son père ?“ Alors seulement elles croient le connaître.
DeepL:
Big people love numbers. When you tell them about a new friend, they never ask you about the basic things. They never say: “What is the sound of his voice? Which are his favorite games? Does he collect butterflies?” They ask you: “How old is he? How many brothers does he have? How much does he weigh? How much does his father earn?” Only then do they think they know him.
Professional translator (Konstantin Konstantinov):
Grown-ups love figures. When you tell them about some new friend, they never ask you about the most essential things. They never say to you: “What does his voice sound like? What games does he prefer? Does he collect butterflies?” They ask you: “What age is he? How many brothers does he have? How many kilograms does he weigh? How much does his father earn?” Only then do they believe that they know him.
Exupéry writes about complex themes with simple words in a book that has no equivalent. Well, here at least I’m quite pleased with our friend. I might pedantically remind it that in Bulgarian quotation marks are written differently, but it’s hardly the only one who forgets. And it handles the simple sentences in a way that is downright comparable to the real translation.
And once more, let me draw your attention to a small detail: the translator settles on едва тогава (edva togava, “only then”) rather than, say, чак тогава (chak togava, “not until then”). To me, at least, that’s a pleasing “feel for language.”
The experiment wouldn’t be complete if we didn’t test the algorithm’s strength in the most melodious and poetic of Western languages.
Original (Erich Maria Remarque – All Quiet on the Western Front):
Er fiel im Oktober 1918, an einem Tage, der so ruhig und still war an der ganzen Front, dass der Heeresbericht sich nur auf den Satz beschränkte, im Westen sei nichts Neues zu melden.
Er war vornübergesunken und lag wie schlafend an der Erde. Als man ihn umdrehte, sah man, daß er sich nicht lange gequält haben konnte; – sein Gesicht hatte einen so gefaßten Ausdruck, als wäre er beinahe zufrieden damit, daß es so gekommen war.
DeepL:
He falls in October 1918, on a day that is so calm and quiet along the whole front that the army report confines itself to the sentence that there is nothing new in the West.
He had slumped down and was lying on the ground as if asleep. When a person turned him over, saw that he could not have suffered long; – his face had such a calm expression, as if he were almost pleased that everything had turned out this way.
Professional translator (Nikola Georgiev):
He fell, killed, in October 1918, when it was so still and quiet along the whole front that the army communiqué had contented itself with a single sentence: “On the Western Front—nothing new to report.”
He had fallen face down and lay on the ground as if asleep. When they turned him onto his back, they saw that he could not have suffered much—his face had such a calm expression, as if he were almost content that it had happened so.
DeepL doesn’t seem to have grasped what happened to Paul Bäumer. Why the first sentence is in the present tense remains a mystery. It also embarrassed itself a bit with the punctuation: that “; –” is rather exotic, and I’m no philologist. Had it come up with the word ничком (nichkom, “face down”), I would have been endlessly amazed. “When a person turned him over”… Mowgli, is that you, my friend?
And yet again, in terms of meaning, things look quite decent. But wait a minute, this is once again too easy for our hero, and we’re here to make it sweat.
Original (Haruki Murakami – Kafka on the Shore):
そしてその砂嵐が終わったとき、どうやってそいつをくぐり抜けて生きのびることができたのか、君にはよく理解できないはずだ。いやほんとうにそいつが去ってしまったのかどうかもたしかじゃないはずだ。でもひとつだけはっきりしていることがある。その嵐から出てきた君は、そこに足を踏みいれたときの君じゃないっていうことだ。そう、それが砂嵐というものの意味なんだ。
DeepL:
And when the sandstorm is over, you don’t understand how you managed to cope and survive. No, you are not even sure whether he has really disappeared. But one thing is clear. The person who comes out of the storm is not the same one who went into it. Yes, that is the meaning of the sandstorm.
Professional translator (Lyudmil Lyutskanov) :
And when the storm is over, you won’t remember how you got out of it, how you managed to survive. You won’t even be sure whether the storm is really over. But one thing is certain! When you come out of the storm, you won’t be the same person you were when you walked into it. To change you—that is all your storm comes for…
Yes, I know, there’s a missing comma in the original; I left it there on purpose to see how carefully you’re reading. Our electronic translator, however, doesn’t get jokes. Something went wrong with who exactly disappeared. The meaning is more or less there, the melody is not, especially in the last sentence. An important clarification: in Bulgaria, Murakami is translated via English, which may add extra nuances.
I think we’re close to the final grade, but before DeepL hands in its report card, we flip the paradigm a little. This time we’ll check how it does translating from Bulgarian into English, with the Bulgarian book that has had the most success in the West in recent years.
Original (Georgi Gospodinov – The Physics of Sorrow):
Детството и младостта са пълни с глаголи. Не те свърта на едно място. Всичко в теб расте, блика, развива се. После глаголите постепенно се сменят със съществителните на средната възраст. Деца, коли, работа, семейство – съществените неща на съществителните. Остаряването е прилагателно. Навлизаме в прилагателните на старостта – бавни, безбрежни, мъгливи, хладни или прозрачни като стъкло.
DeepL:
Childhood and youth are full of verbs. You don’t end up in one place. Everything in you grows, gushes, develops. Then the verbs gradually change to the nouns of middle age. Children, cars, work, family – the essential things of nouns. Getting older is an adjective. We enter the adjectives of old age – slow, cloudless, misty, cool or transparent as glass.
Professional translator (Angela Rodel):
Childhood and youth are full of verbs. You can’t sit still. Everything in you is growing, gushing forth, developing. Later the verbs are gradually replaced by the nouns of middle age. Kids, cars, work, family – the substantial things of the substantives. Growing old is an adjective. We enter into the adjectives of old age – slow, boundless, hazy, cold, or transparent like glass.
Wow… The first sentence is exactly the same! Even though the passage is quite metaphorical and the iron man had every reason and opportunity to be led astray. Here, though, I think the algorithm does its best work. And that’s perfectly understandable.
And so it’s time for the results; I’d be very glad if each of you chimed in with your own verdict. In my view, the electronic translator acquitted itself with dignity. Bearing in mind that what we most often need to translate isn’t world classics but everyday communication, machine translation really does show impressive things. It can definitely help with quick translations, with conveying the gist. Leaving the fine details and the editing to people.
Now it’s up to us to start taking a bit more care of our linguistic culture, spelling, literacy and the intelligence we put into our communication and into our verbal footprint.
And if we add other technologies to the paradigm, such as algorithms turning speech into written text, I think that very soon it will be quite easy for us to communicate fully without an interpreter.
But we will always need translators to add that particle of magic that brings true literature to life… Capisce, DeepL?
If machine translation interests you, come to a special event on the subject!
On December 11 (Sunday) at 7 p.m. in the marble foyer of the National Palace of Culture (NDK) in Sofia, we’ll talk about “The Translator: Human or Machine?” in a panel discussion that is part of the Sofia International Literary Festival.
After plenty of criticism and jokes about the quality of machine translation, the moment has come when we have begun to realize that artificial intelligence has come a long way in improving its translation results. Does the machine threaten to replace human professionalism? Can we artificially reproduce a translator’s creative imagination? Which areas of literary fiction remain open to interpretation by humans alone? Information technology and artificial intelligence specialist Yordan Darakchiev and programmer and journalist Georgi Karamanev will discuss the topic with Slovene translator Prof. Lyudmil Dimitrov and Italian studies scholar Assoc. Prof. Daria Karapetkova.
More information here.
