Professional, automatic book translations seem to be possible now, new tools are developed all the time and this blog post is about a promising one
Deepl versus Google Translate
Virtually everyone knows Google Translate, and Deepl works in a similar way. They are supposed to produce better quality translations than Google Translate and after having tried it, I do agree.
Deepl’s paid version comes with an additional advantage because it deletes translated documents, once the job is done. Google Translate does not guarantee anything like that and we can’t be sure how they are using our texts. With Deepl’s free version, however, you have to be as careful as with Google Translate.
Some books are easier to translate automatically than others
The general rule is that non-fiction is easier to translate automatically than fiction, but there are exceptions.
When you are dealing with special language, for example a scientific text, it would even require an expert to translate manually. The more specialized the book is, the more sense does it make to let an expert with language skills translate the book rather than a linguist, and in these cases, automatic translation is even more difficult for an app.
And then there are extra difficult texts.
I tell you a story to explain, what I mean. Many years ago, I attended a literary translators’ conference and there was a workshop with a famous British author plus two of his translators to give us an idea how they work.
The author mentioned one of his books and said he wrote it within two weeks. It turned out that both translators had spent months to translate it. They had obviously worked hard to figure out “What does the artist actually mean?”
I suspect there is not always as much meaning in the original as everyone thinks, but that’s another story and I won’t tell you the name of this particular author 😉
Childrens’ books, on the other hand, no matter whether non-fiction or fiction, are usually easy to read, but I am sure authors often work hard to make complexities understandable for children, which can then make it easier for the manual translator and for automatic translation as well.
I would always be cautious with automatic translations, because the trickiest mistakes that can occur are the ones that SEEM right but are not. Errors may even be easier overlooked than in manual translations, if we are not careful.
Change is happening fast
(updated in 2022)
When I first wrote this article, things were still different and I was not willing to pay for automatic translation, but this has changed. I am using DeepL most of the time. The more data is fed into a system, the better the result and there is now obviously a lot more data in various AI systems than when I first wrote this article.
In general, non-fiction will still work better with automatic translation than fiction, which tends to be more complex, but improvement can be seen everywhere.
When I first asked professional translators, whether they were willing to use artificial intelligence, no one was prepared to do so, but now I know I am not the only one, but many of my colleagues are also using automatic translation as a basis all the time.
Maybe there will even gradually be more authors, who translate their books via Deepl or similar tools – if they speak the target language – and ask linguists to do the final edit, which will then be faster and cheaper, although I am sure, a thorough final edit will always be necessary.
Something else to bear in mind
For those who want to publish internationally and do not want to use translations – I think it can make sense, if the original language is English.
In many countries, people are willing and able to read books in English, and Amazon offers affordable ads in Germany now, the most successful market for self-publishers after English-speaking countries.
One self-published American author even told me she had SOLD MORE ENGLISH-LANGUAGE BOOKS IN GERMANY than anywhere else…