How to make the best possible translation of a book?

I own a book which has seven different translations from the ancient Sanskrit text "Ptanjali's Yoga Sutra" all in one book, and with the help of the partially different interpretations of the Sanskrit words into the English language I gained a much deeper understanding of the text as a whole.

That being said: There is no single best translation in my opinion. For me, the true meaning is inbetween all translations, in what is common and what is different. If you want to really know something, you need more than one angle to observe. In the sense of "The truth is in the middle" I enjoy to compare different perspectives from different translations. No single one can deliver that to me.

Om Shanti :)
For translating words by authors long gone, I agree there is no “best” translation, or the best is whatever the reader likes best. But while the author is alive, perhaps the “best” is the one the author likes best? — Derek

Consider organizing a weekend workshop where you would invite 7-10 bilingual fans to collaborate on a translation.
They could each translate a book ahead of time, and then go through one paragraph at a time, each person sharing their version, reading it out loud, and providing suggestions to one another. Your books are short so you could get through a whole book in a couple days. I think engaging deeply in your text as a group like that would be an amazing experience, particularly with you there to answer questions.

Claude 3.5 Sonnet is remarkable for translations, and there is a new service, ScribeShadow which uses it for translation and also proofreading. I did the translation with Claude 3.5 Sonnet and then used a proofreader from Upwork for a story last week.
Perhaps Opus when it comes will be even better, but right now, 3.5 Sonnet is the best for fiction or more creative narrative non-fiction. It really keeps the 'voice.'

Why don't you use ChatGPT to translate? ChatGPT is the best tool to translate.

How about you 'gamify' it, Derek?

Your book chapters are really concise. So here's my idea:

- Make each chapter a 'project, and invite as many translators who want to participate

- Collectively, each translates either a segment of it, or the whole chapter

- Then everyone in the group offers feedback, improvements and tweaks to make the translated version tight, crisp, even poetic

- When all members on the 'project' sign off, the final version goes into the pre-publication draft

- All team members who 'actively' contributed to the translation process get to share the kitty (a pre-determined amount of cash or other equivalent), in a similar manner to how Kindle distributes KENP royalties based on how many pages of an author's book is 'read'.

It becomes a fun collaborative project for all involved.

And a final editorial run through for consistency in tone and style at the end would polish it up.

Hope this sounds interesting. I look forward to seeing other ideas that are sure to crop up on this conversation.

All success,
Dr.Mani

Quite possibly concise and efficient prose works in English but not in other languages? Having a native/local translator may be of benefit - a literal translation may not be the best approach, a little local colour and embellishment may help the book be received and understood in other non-English speaking territories.

Maybe the best approach is to find a translator who is also an author, or at least has work they can point to by way of a reference. The "interview" could to be in the overall feel and flavour of the style of their other works - for example, find an Italian translator who has translated Italian to English and see how the English translation 'feels', if it sits well then they're likely to be able to go the opposite direction (in this case English to Italian).
Good point! Thanks Andrew. — Derek

Perhaps a private / secure / members only wiki-translate page: assuming proprietary rights (Copyright) needs to be protected from piracy?

A good translation needs community input from both the source language speakers and from the receiving language speakers, to ensure that the original intent is not lost and the local dialect and idioms are reflected in the translated text.

Many software apps have crowdsourced their localization, by translating individual words, action labels or phrases. Translations are accepted after multiple independent submissions match. The number of submissions may vary based on desired accuracy and community size. This approach depends on a large volunteer community. Ex: WhatsApp had a huge community who contributed across hundreds of languages (I did too for my mother tongue).

Those days, the only incentive was that one could have their most used apps in their native language - to facilitate same network usage of the app (ex: with non-english speaking parents). For books, I don't think that same incentive may be as strong. So authors may need to consider alternative incentives (like recognition, book credits, or compensation).

Some authors build volunteer teams from their readership, using pairs of translators and 1-2 proofreaders per language. I know that the publishers of "Ancient Secrets of a Master Healer" (original in English) did this, and have been able to release a few languages across the world in the past 2 years.

AI translation platforms, which I've tested in the last 3-4 months for English-Kannada projects, are evolving rapidly. But currently, I felt they perform better with global languages (e.g., English to Spanish) than with Indic languages. For me their translations were never fully accurate. So, if this approach is used, a human proofreader (per language) would be essential.

Another idea, I haven't tried, but think could work is - partnering with language schools and offering translation projects as practical assignments, with multiple layers for better accuracy.
Great ideas. Thanks Sunil. — Derek

I studied French in college and my thesis was a translation. I think your plan might be the best idea out there, for scale. But do you need scale, or do you need four translations?

Nothing will beat a skilled person's attention, for capturing tone and style--if you can find the right person. You would need to find a professional writer
or translator and get to know them well, so you can trust them and they will understand you. You'll need this trusted editor even if they aren't the translator.

I think you want one editor you work with heavily, who helps you find a couple translators who work collaboratively--or who is the backstop for Inchword. This would be an expensive way to go, but you asked for the best!

You need first a person that read and speak English very well. Your native language . A person that likes your books and the way that you think . Second this person has to have the other language as native: Spanish, Chinese ... The way that her,his words match with the English words and more important the way that you think.

Definitely check out Eleven Labs for translation. I learned about them on a podcast (not paid / endorsed) and found out a number of podcast and larger multi-language YouTube creators are using it. It seems to be very good so it is worth a look. Here is a link: elevenlabs.io/blog/elevenlabs-launches-voice-translation-tool-to-break-down-language-barriers-for-content

Oh I love this!
I’ve struggled a bit with translating my books because they are so damn long, so too expensive to have a translator do. But, I’d love to do something like this after I use ai to translate the book to the local language because it is quite good, but not perfect.

A translated text is by definition an approximation and, therefore, the best translation is the one most faithful to the source. The clearer the source text is, the less difficult it is to approximate and so the easier to understand in translation.

Ultimately every reader must perform their own analysis and interpretation of a text; so, the fewer steps between author and reader, the better. A dutiful translator should respect that their intermediary role is meant to be as unobtrusive for the reader as it is accurate for the author.

From a practical standpoint I think inviting a multitude of interpretations is just second-guessing the translation process: if you're going to entrust a professional to do the job, make an informed choice and then let them do their job.

However it's translated into another language (AI, crowds, etc.), consider translating it back into your language. Then, read it carefully, to flag what might've been WAY off in the original translation.

Curation of the team translating your work will be crucial. I would recruit people who are already fans and very familiar with your work. I would actually physically work with them in a space that is 100% you. Look for already fans that have an excellent command of English and the desired second language and primary culture that speaks that language. A great example would be Tim Ferriss. He knows you, he loves your work,and he knows Japanese and has spent extended time in Japan. Practically speaking Tim might not have the time available for your project, but he is an excellent archetype of the who you need on your team.

Hi Derek, is there an author with similair values in the language. is there someone with the same values. Or the same area of work. Could you travel over there for a short while to meet with people to discuss meet?
vinnychoff

It's such a daring, cool idea!

My thought is that a good translation is really the work of an artist making sense of another artist's work. If you open your work to translation by committee, it may be like 1000 people all telling you what they saw in the square that day. You'll get a composite of what happened, but not a description the way one person experienced it that a listener can RELATE to. No story.

If you are looking to communicate your voice to someone, rather than just your content, I'd suggest going with a single translator, perhaps using the committee to check it.

I like this, a lot.

I'm bilingual (English/Dutch), and have helped out with translating a few times; currently, I'm helping a start-up translate their content for the Dutch market. I collaborate with another translator, and we're both doing it in our spare time.

Had a look at Inchword and love it - enhancements I think would make this even better:

If possible, there should be a style guide for the original version. It helps with tone of voice ("prefer the active voice over the passive voice"), consistency ("numbers up to ten are spelled out, 11 and above use numerals"), and idiosyncrasies in the style. This might also clarify the intended audience - a lot of translating is choosing between equally valid options. Having that style guide available in the UI would be great.

Secondly, I think translating at the sentence level is great, but I think the context of paragraph and chapter are important. One of your examples had a question about the word "hard" - which in English can mean "difficult to execute" ("running a marathon is hard"), but also "difficult to understand" ("hard problems in computer science"). Other languages have different words for those two meanings. Seeing the context of that phrase would help choose the right translation.

If there's a way to do global search (and replace), that might help. It's quite common that you change the translation of a specific word.

There may also be some value in versioning - again, not uncommon to go back and forth on a word, and seeing what happened before is useful.

Finally - we have used the various LLM translation options, and they are impressive, but get to about 80% in terms of quality. It's fine for disposable content, but tone of voice is a problem, and your tone of voice is a key part of the value of your books.
Nice! Thanks Nev! This is really helpful. — Derek

When I did a 4 country virtual tour of South America for the U.S. Department of State as an Arts Envoy, I had to get creative. Usually, prior to going on a visit, I’ll spend at least a month learning the language using an app called Mondly, and then learn the cultural aspects when we arrive in country. Romania was that way, and it was great. But the virtual visit planned for Bolivia, Venezuela, Columbia and Peru posed a challenge. I took Spanish for 4 years in high school, and felt I had a decent grasp on the language, but how to discern the cultural nuances when you’re not actually going there? I hired the mother of one of my kids’ friends who came from Peru to help me prepare. The information she imparted was priceless and the Envoy sessions would not have gone so smoothly without her expertise and knowledge. For example, she taught me about the people - their cultural differences and what makes them unique in each of the places. She also steered me away from making a big mistake, namely using the word “simple” as opposed to “corto/corta.” The goal of our Songlife program is to write a simple song lyric, and in translation the word “simple” had major negative meaning to native speakers. I would not have known that without Rossy. Can you imagine offending everyone right off the bat simply by using the wrong descriptive word? Therefore, my suggestion for you is to hire a native speaker from each language and work closely with them to make sure your words come across the way you intend. I can help if you want referrals. I have a lot of native Chinese friends, and others from around the world including Romania, Germany, France, Ukraine, all over South America and even Jamaica where they speak a hybrid of languages that no book can teach you. Great work, Derek! You’re awesome. Love & Respect, A

Honestly I would just use an LLM to translate the book.

I DID translate one entire of your books into swedish, I believe it was "Your Music and People". I did not recieve any compensation for it, possibly because I took liberties with the one chapter in which you, to my surprise, blatantly lie and claim that booking is easy. I could not let that pass. Had to edit. Aggressively.
This is where I learned the importance of having reviewers and editors. I trusted Melanie when she offered to translate my book into Swedish. Then I released it, and readers told me that the 'translation' included 'comments from the translator' saying (in Swedish) 'everything in here is a lie, this is bullshit, he's totally wrong here', etc. Melanie taught me not to trust. — Derek

Depends what you mean by “best”.
If best means going for highest impact, easy to understand the point, easy to recall, then choose simplest most used words criteria. I have the feeling you are writing for you always.
It is difficult to hear Derek’s ‘tone’ in other languages.
I’d say go for meaning and choose a translator that already read your books, resonates with you and has the drive for the work.
Good luck!

The start of this reminded me of the story of a letter by Mark Twain that said “I’m sorry this is so long, if I’d had more time it would have been shorter.”
I don’t have anything to add for translation advice.

Isn't it remarkable how it takes more time to write fewer words and yet more impactful and, yes, more beautiful!! The ultimate expression (pun intended) of quality versus quantity.

I would look for native speakers with experience in writing and translating poetry, not (just) books or articles.

Start with Why.

Select translators and take the time to explain the rhythm, intention and meaning behind either the entire book, or the sections that contain crucial meaning that must not be lost. Recording yourself giving this detail and sharing it with the translators might be most efficient.

Give the translators a window of time to submit questions or requests for clarification and then let them get writing.

Request that they prioritise translating the essence and be less precious about getting a verbatim translation of words and meaning.

If they understand your purpose and you give them the psychological safety to work freely and meet your expectations, I imagine they’ll be excellent translations. Would also recommend they read other material you’ve written and listen to podcasts to deepen their understanding of your sense of purpose.

It will certainly take longer and cost more, so you could adapt by prioritising which elements of the book deserve this approach and which might suffice with a standard translation.
Beautiful point. Thanks David. — Derek

After you dead, you won't care what you've left behind.

After we dead, everything is nothing.
Useful for others, useful for you... it's all nothing.
It feels something because you still alive now, then you're projecting it would be a good feeling too after you dead.
No... it is absolutely not.

Our feeling is not exist anymore after we die.
The world, the universe, every single thing... is nothing after you die.

In order to feel... you need to survive first... to be alived first.
I know nothing about brain cancer... but I hope you would find a way to live longer... at least 150 yrs... and being young again.

Not a book but I just translated my upcoming Hindi song into English. Last time I tried this, it took me forever and then I had to depend on another gentleman whose English is much better than mine. This time, I used Chatgpt to get the first version and then improved it myself. So the method could be a combination of using Chatgpt and then some sort of crowdsourcing that you are talking about. Eventually I think chatgpt would be so good you may not need the second step.

Your predicament reminded me of when I was searching for "the best" translation of Meditations by Marcus Aurelius. Most translators were approaching their work from the perspective of a scholar or academic - they wanted the perfect word for word translation - which leads to the book being quite dense and difficult for many readers. However, Gregory Hays approached his translation from the perspective of Marcus Aurelius himself. Meditations is Aurelius' journal and was never meant to be published, and so Hays wrote his translation as someone in modern times would write in their journal.

I think you put a lot of thought and effort into what and how you say what's on your mind. It would be a disservice to translate your words without also translating your meaning. In my opinion, gathering all of the translations as you are and then iterating could very well work, and I haven't seen any of that process so you might already be doing this, but be sure the people doing the translating are translating what you're meaning and not just the words.

As someone who has conducted localization projects, especially for e-learning content, here are a few tips I learned the hard way.

1. Have a translation design guideline. This guideline is based on a sample of content that has been perfectly translated, such as one chapter of a book or several blog posts. This sample will provide the tone, style, and cultural background to give insights for the translator and the audience, as some cultural references can be completely out of touch.

This is particularly important since languages have standards regarding tone and style. In French, when you are very close to people, we have a very informal way to talk that feels natural in this language.

This phase is the blueprint to build the foundation of a translation that will be consistent, especially if you plan to crowdsource it.

2. Make a "draft 0" using LLMs such as ChatGPT or Claude, using the translation design guideline as a reference point. LLMs are getting really good, but I suggest not populating the content into a website; instead, use a simple add-on that can be plugged into a word processor, the content sent via APIs, following a simple and effective format.

It is much easier to start from a draft than a blank sheet of paper.

3. Have a program that presents side-by-side translations, paragraph by paragraph (not sentence by sentence, as sometimes, depending on the language, things might be merged rather than split, and also because topics in some languages are introduced at the end, not at the beginning).

With LLMs, automation, and integration, it is now possible to have the flexibility and comfort to edit text within a proper word processor and send the content to a remote server for proper synchronization, similar to a lightweight GitHub for content.
Thank you M.A.C. Very helpful. — Derek

If I have a book and I want to translate it to other language considering I have a group of people who are proficient language speaker in English and Spanish and I trust them. I could do:

1 - Use a tool to translate the book (from English to Spanish) like www.deepl.com, translate.google.com, www.reverso.net, www.babylon-software.com.

2 - Send the book split in parts and organize to people who are bilingual in English and Spanish and ask them to translate sentences in the book from English to Spanish having the machine translation as reference, send to them both versions in English (original) and Spanish (machine translation) paragraph side by side.

3 - Ask your bilingual people (translators/editors) change those paragraphs who look weird in the translation like slangs, local references, wrong meaning, broke rhythm, plus your personal rule like avoid long words or try to emphasize some words in the text; with one rule, they must justify why they are changing the translation.

4 - Send the book translated using now original English plus machine translated version plus human text edited/translated to a final approval by a native Spanish reader.

Why this method:
I believe machine translation tools are great. They are able to translate fast in large quantity with low use of human cognitive resource. But those machines have two problems, they can be trick in particular use of slangs, local references and collocations and machines have no emotional connection for what is being translated so they cannot identify and translate for another language the style consistence in the writing, for example, end each sentence with a rhyme.

So I believe a productive method is to let the machine do most of the work and let humans make small changes and decide if that is good or not.
I agree. Thanks Raul. — Derek

Derek,

I share your frustration, as well. My wife is Thai, so she speaks and writes Common Thai, Isaan (a dialect common with Laos), and Vietnamese. Look at what we call "common Thai" and you will understand. Beyond the speech part, "translations between English and Thai loose allot". They do not have a separation between words. It is one giant string of weird characters (I think maybe 42). Worse is that the all the tenses we use in English are not used in Thai.

My advice is to hire at least 2 translation "services" employing two people, each with at least an IQ of 115, who have at least 15 years as a polyglot in the required language. One will review the work from another, and vice versa. (Students in Academia, may be able to "do this" very economically, particularly if you source these folks via the University teaching these languages, that you seek translation services for). Teachers can easily (and likely for a small fee) incorporate some of your "text" in translation work within the classroom. You win additional "subscribers" and your model may be copied by others.

Best wishes Derek, I hope that helps.
Regards, Patrick
Interesting! Thanks Patrick. — Derek

Our company has used Deepl for years as a more accurate translator than google translate since it understands the nuance in meaning over literal translations. They all have a new AI translator and feedback from our clients in Europe and APAC has been very positive. Our marketing team will process our collateral with this as a first pass and then where necessary we will pay a translator to sense check the document which works out much quicker and cost effective.

A good translation flows from one hand, one editor if you will. Otherwise there are too many different styles mixed up and it cannot convey the author's voice. Because the author is one person also. I agree that the best translation is that which conveys not only the words and meaning, but that can transmit the tone of the original language, which will be a different tone in the other language bc of cultural reasons.

Is it more of a technical question, a linguistic question, a cultural question?

I think that the problem domain is unclear to me, and if we try to address it maybe breaking it up into a set of sub-problems is a good idea.

The best translation is the one you write yourself, maybe tutored by a native speaker. But is that "possible"?
It depends. How much do you care? Would you learn a new language in order to deliver the most authentic translation to millions of non-English readers?
I am a Spanish teacher, and happy to help, should you choose this path :)

Would you write your books in a team effort? Probably not. Same with translation
- if it is done properly, it is an art. A solitary game. It is not a team effort - not more than writing a symphony. Pick a Beethoven or Mozart and award him or her the assignment. I suggest you pick one translator and one proof-reader for each language. Ideally, they should be top professionals who like your work and feel passionate about communicating your intended message to readers - to make an impact.
Translation is an approximation - that's true, owing to the inherent differences of languages. It also means, however, that the translated text may be better than the original, or at least as good. I would focus on finding the right translator and the right proof-reader. That should be your team. Not more.

Hey Derek,
I think your idea of using the Wisdom of Crowds to translate your books is awesome. I'd be willing to help out with the French translation, either as translator, reviewer, or editor, whichever role you have the most need for.
Cheers!

Here is a contrarian take: It's borderline impossible for you or anyone to get to know all the little idiosyncracies of a language enough to understand what an actual "best" translation is. Every language is a conception of the world.

I would suggest, therefore, to put every translation in the hands of a single, talented translator that resonates with your worldview and that you have a good back-and-forth with. If you go for something like this, happy to help with Spanish!

Hi Derek, the process I've moved to is ChatGPT (latest version) for the text translation and Suno.ai for the audio version. The paid versions of both work quite a bit better than the free versions.

I was just watching this video about realtime local translation (Babelfish.AI) with audio/web APIs and transformers. www.youtube.com/watch?v=uT945Rh5sl8

🎙️Realtime in-browser speech-to-text with OpenAI Whisper!
📡 Broadcast to subscribed clients with ‪@Supabase‬ Realtime.
🌏 Translate to 200 languages with ‪@meta‬ NLLB-200!

Find the source code: supabase.link/babelfish-ai-gh

I see there are many ideas to achieve a good translation. Unless you learn the target language you may need pairs of translators, one to translate and one to translate back independently to see if the result is acceptable.
I would not rely solely on computer translation. Proof reading by a real person is essential.
Making the right impact is a challenge. For example you could describe yourself as a bass player, or a more dynamic description of the music you create.
I went to a rock gig and read a review describing the stage effects as ‘enough pyro to take out a small town’. While that provided a good image of the stage effects, I dread to think how that would come across in other languages.
My experience as an English speaker was being asked to tidy up a rough translation of a short story originally in Japanese for a sci-fi concept music CD. It took many hours and many questions to arrive at the final product.

I've been a Japanese-to-English translator for 30+ years. Translators of the caliber you're seeking are vanishingly rare. All translators have their own ideas of what constitutes a good translation and they all consider themselves to be good. In reality, 99.99999% of them suck IMO. If you want an exceptional translation, you need an exceptional translator. Good luck finding him or her. All these ideas about crowdsourcing or getting input from lots of people won't work cuz the crowd sucks.

* Use beta readers from the target language community to provide feedback on the translation before final publication.

* Have the translated work read by a focus group of native speakers in the target language to ensure it resonates with the intended audience.

* Work closely with the translator to ensure they understand your vision, tone, and intent.

Hi. I was talking to my professional translator/interpreter friend yesterday and she was saying that these days, even if you send your translation out to a professional, it is likely to be partially machine translated. The translation will be proofread and edited by someone who previously would have done the translation.

I would say that one way to do it would be to have an interlinear online version of the book available which splits the the text into paragraphs and has English on the left and the target language on the right.

Initially, the target language would be machine translated by hitting the API of one or more translation services and/or AI LLM translation prompts.

Once the paragraph was machine translated it would be open for trusted volunteer editors to edit the translation and for the not yet trusted reader to add a comment about the translation. A reader could also just flag the paragraph as being problematic and alert an editor to take a look at it.

One unintended advantage of having an interlinear text might be for those readers who do partially understand English to also be able to reference the original version. These "parallel texts" are also quite popular in the language learning community.

Another interesting thing to do would be to have a translators notes available too. e.g. some paragraphs might get flagged a lot but actually be correct so the translators might want add some explanation about why the text was translated that way or explain something interesting about the text.

Notes like this are actually quite common in biblical interlinear texts where decisions to translate a word might be very significant and/or differ from tradition but mistranslations that occurred in other versions.
Wow! Thanks Stuart. That was really helpful. I appreciate it. — Derek

I really like the ideas behind Inchword. I'm working on something like that myself - for my book-writing project - as a way to invite others to review, make comments, provide (draft) translations, and edit the original texts and translations.

One of the additional features I'm including is on-page statistics on the text, like word count, (average) word/sentence/paragraph length, and most occurring and unique words (by their stem or root words). Although it's very superfluous, it's also a way for the writer to see whether the edit or translation stats are within similar bounds of the original text. (Yes, excluding the major influence of implicit language characteristics.) The stats at least provide a hint (and little more) that the editor/translator remained relatively close to the characteristics of the original text... especially if you can compare the stats of translated parts to other translated parts.

With the advances in AI, I'm working to add functionality for AI-driven (or AI-inspired) edits and translations. Obviously, before approval those will require a human review, but for proofreading, spelling and style flags, and punctuation checks an AI tool provides an efficient headstart of human-powered rounds of editing.

Nudging a community of capable editors might be a challenge, especially if they're not compensated in some form (and mentioned in the final publication). How to keep editors engaged and committed? How to calculate and reward their contribution?

Finally, there's the subtle but important point of handling the copyrights of the edited and translated texts. The main author wants to keep copyright, but I'm not sure how edits and translations are treated here. I guess you want a big disclaimer and explicit confirmation that the copyright is transferred-in-some-form to the original author.
Good points. Thank you Jochem. — Derek

Place the book into a YouTube friendly format and set the embedded language to any language option you would like for the translated closed captioning. A visual book of sorts.
Can be embedded on site behind a pay wall.

Love ya brother!

Your Intention / Issue
1. Get the best translation
2. Definition of best - is what you think is best.
3. In your opinion, any one individual can't do it the best

Solution
Give to 5 - 7 different people. Let them translate the full work individually.
You get to pick whose translation, or whose specific part you would like to keep.

Think to avoid
Organising a workshop / discussion. You want to go by your opinion. That is best done going through the translated works one at a time. The more the discussion, the more it can get confusing.

AI to do the bulk translation. Should match your tone fairly well.
This does the bulk "grunt work"

Then get the highest paid, best translator or bilingual writer you trust to edit it.
This gives you your refined edit to mimic your own process.

Derek:

I'd love to help, but I don't feel qualified. All I can think to do is to question how I'd become comfortable in making these kinds of decisions.

Can you ask other authors? I first became aware of you through Seth Godin when he pointed me to CD Baby. Seth has lots of experience with translations.

Part of my career has included name creation. I know that successful brand names are never more than 2 words and never more a combined set of 3 syllables. Once a key word had been identified, my methodology now includes researching synonyms and antonyms. Picking an anntonym, then looking for antonyms (for the antonym) leads me to other options. In similar manner, can you ask for a translation of one of your writings, then ask for another translator to translate it back to English? I'd suspect that this might be disappointing. Few write like you. I believe that tl;dr is real. We all love the concise nature of your writing. I also believe in never taking a pole of one. Seeking the top 1 is one thing; identifying the top 3 is another and I love options.

I've played a bit with AI, mostly on ChatGPT. I am aware that you can ask for an answer in the style of a particular author. I am also aware that other authors are in litigation over incorporation of their work without authorization.

Best wishes for continuing successes.

Dan

In my experience reading translated texts, I've found I prefer literal translations paired with annotations providing context on the author's original intent. A foreword written specifically for translated versions of the book to underline your original intentions with tone/style might be helpful for readers.

Also, this reminds me a bit of BaniDB - www.youtube.com/watch?v=BJKALklMKao - crowdsourced translations of the Sikh holy book. Of course, for a modern book, other strategies may be preferable.

It’s better to have your friends like us to Translate your books

I can translate into Tamil as I did last time

Yours
VN

1. In case of Indian Languages, the tech tools are not effective. The reason is a word has a literal meaning and also a context based meaning. Identifying those subtle nuanced difference and using the right word is Human Intelligence and the AI has not evolved to that level.

2. Further those who build the tools are not well versed with Language Skills and hence no tool is sufficient for a proper Translation

3. That is the reason I said, let your friends like us to do the Translation Job. It does two things — our relationship gets strengthened and the thought process and ideas are getting properly understood by all of us for greater good.

I read both Chinese and English. I once saw a Wong Kar-wai movie called The Grandmaster with English subtitles. It was in a theater, so it was not the youtube AI type of translation. The translator took so many liberties that sometimes the plot was changed. For English speakers, the movie wasn't too good. With the original Chinese dialog, it was a wonderful and deep film.

I recommend you ask your multi-lingual friends to check samples of the translators before hiring them. If you are talking about music in your book, then I recommend you find a translator who is familiar with that specific genre of music.

My 2 cents.

Good luck!

Just checked Inchword. It’s a great idea from the surface, but not sure if it would finish the given task of a whole book on time. Furthermore, each person’s translation, shall be based on him/her understanding of the sentence, thus the whole book might be incoherent as each part is taken by a sole translator. A better idea would be that incentives are introduced, where the translator and reviewer get compensated just as sole freelancers. That solves the time and quality aspect of the finished product.
Thanks for the thoughts. To be clear, inside inchword.com, the translator sees the whole book, and translates a whole chapter. It's just that every sentence is saved and edited individually. — Derek

But what about the tragey of the commons

I love the quote a camel is simply a horse designed by a committe. I would probably hate a translation of Rilke or Rumi done by a crowd.

Writing is inherently biased. But if you try to crowd source tone, style, and energy you end up with mush (or I fear you might)

Hire one translater and trust they'll do a good job. You want someone who can express your ideas with the singularity of a voice just like your work is the singularity of your voice.

You could think of this person as French Derek, Spanish Derek, etc.

They would be biased (you are I'm sure) but at least it would have the cohesion of a voice. You can certainly ask for feedback from others, but avoid the committee problem and just find a good translator.

Hi Derek,

Just curious, have you translated any of your books into indic languages? For instance, I speak a language called Malayalam!
Yes. ☺ See the comment from V Narayanan, just a few above yours, who translated my first book into Tamil. — Derek

Hi Derek,

Your subscriber Cory P. sent your post to me and I am very interested in learning more about your book and maybe some possible translation contract work. I have two Masters in Linguistics and Japanese Literature, and a Bachelors in Mathematics and Philosophy, and am a native Mandarin speaker. Please let me know if you are interested and I am happy to share my resume and background with you.

Best,
Jack

I believe that everything has its own unique approach, and to truly excel, you need the best methods. I can provide a framework for achieving the best translation.

I think the reason you emphasize "best" is that many of the people you talk to may not be able to connect with you at all. For example, people often say, "Wow, Derek's words are so concise that I absolutely love them," but many of them don’t really understand what "concise" means, at least not in the way Derek uses it. They only know the definition from their own dictionaries or a popular dictionary.

Clearly, given the above, one person's translation is unlikely to capture Derek's style. Furthermore, the translated words might not be remarkable; they could just be common phrases.

"Concise" is popular, while "outstanding" is special. "Concise" is good, but "outstanding" is the best. The challenge is how to keep the translated words outstanding across different countries, with their diverse languages and cultures—this is the tip I mentioned at the beginning.

Derek, I can help keep your thoughts, methods, and words alive in China. Thank you.

Add non-text features such as pictures, paintings, drawings, graphics, diagrams, symbols, emojis, memes, charts, graphs, & color. Underline, highlight, bold, italics, font, etc. Give ‘em all a shot.

At the risk of cluttering your writing unnecessary sentences and words:

— Add callouts in the margins

— Use footnotes

— Include alternate translations

Consider the “cluttery” features above bonus material — what we call “lagniappe” in Louisiana.

Translation of Lagniappe: “A Little Something Extra”

Tangents. A healthy dose of tangents.

Hope all is well. Good luck!

Hello,

Since you mentioned that your books are the primary work of your life and you intended to have them available in other languages, why not translate them yourself?

I believe you have learned or expressed interest in learning other languages in the past. This would be a perfect opportunity to combine that interest with your translation project. Translating something is a classic technique in language learning. In addition, learning a new language brings new perspectives. Translating or rewriting your work in another language can improve the overall quality of the work by feeding the experience back to new editions of the original work.

As a result, you would be the best translator of your books if you can stomach the work. Otherwise, there are no good reasons to care so much about the quality of translations. If you don't know the language, I don't believe you'll be able to have more nuanced judgments about the quality of the translation beyond what you can glean from how people are reacting to the translation and your name. If so, choose a translation method that optimizes what you want to see from those two perspectives.

My suggestion is to test it out.

Have a professional translator do at least a few chapters, so you can compare the quality and feel for the reader of a single main translator against a wisdom of crowd scenario.

My guess is the same as yours, that the crowd edited given enough people participate would be the better one, but testing and trying would be extremely valuable. Give both versions to a sample of people and see what they say.

One point I considered is if it might feel like a patchwork quilt the crowd version, versus the consistency of one person going through it all. But I would test it if possible.

Cheers!

This made me think that, as questions, suggestions, clarifications and revisions take place, the errors become evident of those who accept the literal meaning of every word of an English-language translation of an ancient religious text. Judgments, assumptions and tradeoffs always have to be made.

My father has a sculptor friend.
One day his friend decided to make a bust of him and give it to him for his birthday.
My father told him "I'm very grateful, but... it doesn't really look like me."
His friend told him "Don't worry... when you die, THAT will be you."

Derek, your concern is valid. The solution is not clear.
:-) — Derek

Hi Derek-

I would use the current scheme of work from inchword.com but introducing some changes based on Marc-Alexandre Cartiant (#30) and Raul Borges (#31) ideas:

1- Author uploads text, with clarifications.
2- LLMs translates one sentence at a time, with paragraph and chapter context.
* LLMs must be previously trained with all your public and private writings and thoughts, it will be a global translation guideline and it will provide the tone, style, and cultural background necessary.
* Each clarification will work as a prompt, it will be a focused translation guideline.
3- Author doesn't need to answer Translator’s questions since the LLM was previously trained by him (LLMs can reinforce their previous training with questions to the Author).
4- Reviewers offer improvements as a secondary prompt associated with the firtsone. LLMs improves until done.
5- Editor makes final improvements. Ideally the Editor has to be a bilingual proofreader.
6- Manager oversees entire process.

Hello Derek. I have been following you for years from Türkiye. I even went to Singapore last week and you were in my mind :) I uploaded your articles on my YouTube channel translating them to Turkish language. I also read all of your articles three times. If I can help you in any aspect I will be proud Derek. Greetings.

According to me, to make a great translation, really grasp the original text, pay attention to cultural details, keep the author's style, be clear, and work with native speakers for accuracy.

Possible incentive structure building on the concept Derek discussed on Tim Feriss's podcast:

Whenever a new translation fails to beat the old translation, the author of the old translation receives an amount of money.

This fixes the problem that good translators don't have an incentive to go first.

It is more expensive for the author of more people make bad translations.
Maybe the new translation has to come close (30/70 or 40/60 percent of the votes compared to the better and older translation fx) for the older translator to get paid.


Potential problem with the whole approach:

Certain words makes more sense if they are the same throughout the book. This could be a problem when a word has multiple translations.

For example "true" in "useful not true" can be translated into Danish as either "sandt" or "rigtigt". I might prefer "sandt" and choose to translate the whole book with that, but another person could use "rigtigt", and when reviewers review individual sentences or chapters, one might sound better in a certain sentence, but it would make more sense to have it be the same throughout the book.

This is a somewhat trivial example that could be fixed with a bit of work and couldn't be automated, but there are probably harder to detect variants of the same problem like turns of phrase or consistentmy structuring sentences the same way in languages where there are multiple valid options.

Great insights, Derek. I love how you stress the human touch in translation—machines can’t capture the nuance alone. For anyone looking for quality book translations, this service might helpful