I think even a bad translation is better than no translation at all, but for some reason the majority of books don’t seem to get translated. Why is that?
To translate from one language to another, you need to know both languages very well, including puns, idioms, etc., and have a deep understanding of both cultures involved. A word for word translation will rarely do; you may even have to come up with new jokes / references to replace ones that don’t work in the target language.
If you have read translations of manga by, say, Rumiko Takahashi, you might have seen footnotes explaining all the puns and references. And even this is not ideal, since it breaks the reader’s flow.
I can’t begin to explain how difficult translating a text actually is. When I was at uni, language majors had two dedicated classes, one for translating from your first language and the other for translating to you first language. Both were the hardest classes to pass. I had bilingual friends there who had shit grades solely because their methodology was lacking. In the philosophy department, teachers were constantly discussing who translated what text and how it might impact our understanding. Teachers would compare different translations and ask us to always tell who the translator was in our research exercises.
Language is a fucking mess, translation is horrible, that’s why we only do it when it’s a potential best-seller
Because historically (and for the most part today as well), it costs money.
Sure, today stuff like ChatGPT and the somewhat older Google Translate exists, but that doesn’t solve the cost issue. (And I’m skirting on the huge elephant in the room called quality for a bit of brevity).
There’s a huge chance someone paid a good chunk of money for all the books you find dirt-cheap at a flea market, check out at a library or happen to find in your own house.
Printing physical books is expensive. Publishers also want a margin, and a lot of authors want royalties.
In the end even if the publisher and author are both good souls demanding nothing, someone needs to foot the cost of printing. But before that, you’d need to go through non-trivial talks with most authors’ publishers and/or authors themselves.
Then you need to arange for translation, typesetting and printing if you’re not doing it yourself. That takes both time and money.
And if you were to do all that yourself, it’d be a huge time investment, with a potential lawsuit if you don’t do those damn talks. So most just don’t bother.
Businesses are incredibly inefficient, even though some are “successful” and have a lot of cash to burn. They need to pay workers, bills, buy and fix equipment, and of course, a cut needs to go to the top people. Usually the “golden” 80-20 rule applies to almost everything: 20% of books make 80% of money, 20% of employees make 80% of money, and a different 20% of people do 80% of the work, etc. And of course, in this world, it’s all about the money.
A translation is usually initiated by a publisher that has a manager who wants to get his section’s metrics up to go cry to his own manager about how good he is to get a raise or not get fired. This is a daily grind. Sometimes (but quite rarely), that leads the manager to the decision of publishing a new book. Usually such actions are guided by things like bestseller lists, reviews and personal biases of the manager and the company as a whole. Sometimes the publisher hires an agency to try to approximate the demand for such a book (even more money spent). Then they do the talks. This also costs money, and the result is also a cost of money (the royalties to be paid). Then comes translation, then printing, then distribution to bookstores, and finally advertising.
These are just the steps that come to mind. All cost money, and all the books you see for sale in a bookstore went through all of these steps. For a library, not as much (but still the vast majority) did.
Sure, not every situation is the same, so there are companies that specialize in providing translations of well-known works or companies whose manager at one point said they need to publish 25 translations yearly (instead of one individual one), so they kind of “flood” the market.
But sometimes it’s just the whim of a newspaper whose management thought printing classic works of shorter length and bundling them with their newspaper would drive up newspaper sales.
It’s incredible how each document (edition of a book or otherwise) has multiple stories (of the author, publisher, translator, seller, advertiser, buyer, worker in logistics/delivery driver,…) that shaped the life of it. Some lasted a few hours, and some took hundereds of man-hours. All of this somehow translates to money.
That’s the long answer.
The short one is: 80% the economy and 20% human laziness.
I think even a bad translation is better than no translation at all
I disagree with that. Example.
Brunhårig sjuksköterska
Vs
Brun hårig sjuk sköterska
They look the same. But I can assure you. They mean two very, very, different things.
Interesting how Google Translate returns brown-haired for both Brunhårig and Brun hårig while DeepL doesn’t but they both return brown-haired when provided with the whole phrase. Shows how trustworthy auto-translation is.
Interesting, The top one is a “brown haired nurse.”
The bottom is a “brown hairy sick caretaker”
One would be far more pleasant to deal with than the other.
Literary translator here. An average-sized book takes 4 months to translate, and I bill around 12k€.
90% of books don’t sell enough copies for the publisher to recoup the costs. And that’s books that have been picked amond hundreds because they’re good and/or might sell.Additionally, you’d need armies of competent translators. It’s a complex skill that takes years to hone. Who would do it?
I’d disagree with the premise that any translation is better than nothing. If you want to buy a book in one language and use machine translation to another that has been technically feasible for years, but the quality of the translation won’t be great. A good translation is far more time intensive.
Honestly even when it’s translated by a human you can usually tell it’s a translation. I try not to read translations, it’s just not the same.
There is no inherent value in it. A translation only makes sense if enough people will buy and read it to make it worth the effort and cost.
But I do think (actually fear), we’ll soon get AI-translated e-books where you can select the language at checkout.For literary fiction, machine translation is near worthless. For tech stuff and maybe some nonfiction, it can be usable with some human cleanup. But ideally you can do the machine translation on your own computer if that’s what you want.
YouTube voice auto-translation is completely worthless, but they still shove it down your throat.
and the translation will get progressively shitter the further into the story you go cos the compute resources start getting rate limited lol
But I do think (actually fear), we’ll soon get AI-translated e-books where you can select the language at checkout.
Hey, that might be one of the few things LLMs are properly suited for.
Not at all. Anyone who is bilingual and who have experienced LLM generated subtitles on videos knows they are really really bad at that. As soon as a word has multiple meanings they basically always choose the wrong one.
For some reason, subtitles are terrible, but if you plug a couple of pages into deepl, it’s pretty understandable. You need to clean it up a little, but that’s pretty easy. I am a language teacher, and actively have to design assignments so that my students can’t just plug them into deepl, because it renders them pointless (for about a third of the class, two thirds don’t clean anything up and it’s obvious).
Good translation work is difficult and can take a long time. In the past, it was either done as part of academic research or if a publishing company was willing to invest that much. At least as far as works of fiction, history, plays, and anything else with nuance and cultural context go.
Now, machine translation is OK for straightforward factual works and can convey the gist of a story, but still do not measure up to a studied, thoughtful translation. I disagree that a bad translation is better than no translation, since a bad one will give the wrong impression of a text or kill its magic or imply things about the culture it came from that aren’t accurate.
AI slop translations threaten to make what’s left of the cultural differences in this world bland, bias them based on the mostly English language works that AI is t trained on, and who knows what else. AI doesn’t understand meaning, it just provides a plausible answer to the input prompt. What would it do with Ulysses or The Bible if translations of them didn’t exist?
Sometimes translation changes the original book to something different. Translating a book is not a manual. Word by word process. You have to take cultural differences in account also.
It is not cheap and definitely an art form itself. And with modern works licensing and the og copyright owner sometimes overreaches making it even more harder.Translation is expensive, it needs to not only to translate the story but the author’s style and pun. A niche book won’t be translated, especially in smaller languages (you have a bigger market when you translate in Spanish/French/German than when you translate in Luxemburgish or Nepalese). With that regard, english-language writer have the advantage of using a language that many can read, giving publishing company a glimpse of the translation potential.
With the power of “LLM” and translation app like deepl, it may work better in the future. However, at the moment, deepl is far from giving a professional grade results, especially when you have ambiguous word or a “specific but uncommon context”. It work to share a blog-post/news article, but won’t be sufficient to get fun to read book or an usable technical manual (However, corporate legal department may have a different opinion, it will be an interesting trial the day an accident occur due to an ambiguity in an automated translation)
Money. There’s no money in it.
It would overwhelm the market but more choice would mean more purchases, but I guess not enough to bother.
It would not necessarily mean more purchases. For the majority of those translations which you suggest, there would literally be zero purchases, because the demand is non-existant.
Dumb argument. That’s just your opinion. You can’t know what people would or wouldn’t buy.
For the books I would personally most like to translate, I think the problem is marketability. Nordic children’s/youth literature often contains nudity/sexuality and/or darker emotional themes which are often viewed as inappropriate in English-speaking cultures.
In “Vi skulle vært løver” by Line Baugstø a young girl discovers her classmate is transgender, and for much of the book participates in transphobia before learning better and supporting her new friend. It’s a very well-told and realistic emotional experience, but would likely be seen as grooming by many English-speaking audiences. Not only does it support trans people, but it also spends quite a lot of time in the girls’ locker room. I think if you tried to give this to kids in the US or UK there’d just be a ton of controversy about it and it’d get banned.
I wonder this all the time. I can’t help but fantasize how I would translate things while reading, but there’s nothing to be done about it if the publisher isn’t interested. They could at least make it legal to distribute fan translations.



