Difference between translating Japanese LNs and Chinese webnovels?


  • Member

    In advance, so you don't misunderstand. I love J-Novel. You guys bring me quality translations at a likeable speed. Never really felt comfortable about asking this, since in a way, it can be seen as a me complaining about the translators, wich in a sense i guess is true? :/ But been holding back for a while, and never really seen anyone talk about this before, so gotta ask. I tend to stay away from Reddit, so no idea about there.
    Anyone know wich is more difficult? Been wondering for a while really, ever since i picked up the xianxia/wuxia webnovels. I always assumed they more or less had a similar difficulty, modified by a translators familiarity and experience with each language. I am of course talking about translating it into English :) wuxiaworld.com always deliver high quality translations(official btw, they partner with authors or whomever holds licensing rights, so no piracy) at a rather astonishing speed. At least it was when i first got introduced to it.
    I started out with Anime, then went to manga ending with LNs. Now i prefer LNs waay more then the other two. I rarely watch an anime or read manga if i can instead read the LN. Biggest exceptions would be if it has music and/or battles involved. Since those are always better when animated and with sound. However, after i started reading chinese translated web novels, i guess i got spoiled? Between 1 & 2 chapters a day. Every day. For years. That is roughly 2-4k english words a day, give or take a few hundred words. Depending on if they do 1 or 2 chapters daily. No idea how many Chinese characters that is though. These chapters are also fully edited, and they often also fixes anything they miss if pointed out by commenters. That can pretty much be anything between 50-100k+ words a month.
    So i've always wondered why it usually takes 1-2 months to translate a Japanese LN volume. I have no issue about the time it takes to start on a new one, since i believe that is related to licensing and other things outside of translation. But once a volume has started being translated, even J-Novel, the fastest as far as i am aware, takes 2+ months to finish.

    At the end, i want to say this frustration mainly, if not fully, comes because of those guys translation Mahouka Koukou no Rettousei:p Been waiting 4 years for the official translation to overtake the fan ones. And i still gonna have to wait at least another 2 years. As stated in the beginning, i actually like the speed here, especially the format that enables me to have several series to read each day. So even if it takes a while for the volume to finish, it doesn't really feel that way.


  • Premium Member

    For Japanese novels, you may see a lot of your answers on the recent r/LN AMA https://old.reddit.com/r/LightNovels/comments/j2zuwi/tl_im_the_translator_of_oregairu_rokka_magical/g7940a0/

    For Chinese, I suspect it is very difficult to translate officially if you're not owned by a Chinese corporation. Plus the fact that there is no publisher to talk to, no illustrations or even covers to use, no established way of splitting the CN wn up in sellable volume sizes, and possibly a million other things.


  • Premium Member

    @KopiCAT said in Difference between translating Japanese LNs and Chinese webnovels?:

    So i've always wondered why it usually takes 1-2 months to translate a Japanese LN volume. I have no issue about the time it takes to start on a new one, since i believe that is related to licensing and other things outside of translation. But once a volume has started being translated, even J-Novel, the fastest as far as i am aware, takes 2+ months to finish.

    I think you kind of have the right of it here. It's not that they couldn't in the strictest technical sense, but it's really not in their best interest to.

    It's pretty unrealistic to expect a company that has to pay for licenses and translators who need to directly sell the translations to be able to compete in speed with unpaid fan translations that aren't licensed, and don't need to be sold directly. That means I think it boils down to three things basically:

    • Cost - Obtaining licenses itself is an expensive and lengthy process let alone also hiring translators and editiors. Every volume they translate adds to the cost so translating quickly adds up quickly. Fan translations have no upfront or long term costs and when one person quits someone else can just take their place. Even the website you posted 'begs forgiveness' instead of asking permission, meaning they wait until a translation is popular and long running to try and form a relationship with the publisher/author.

    • Quality - When people pay for something they expect higher quality than if it's free. The unpaid fan translations can come out very rough and then be fixed up later on and no one (ok there's always someone) will complain cause it's free. Places that expect you to pay have to put out the material in either a completed (when you purchase the LN) or nearly completed (like JNC prepubs) state which takes a lot of time.

    • Market Saturation - People only have so much money to spend on LN and if you put too many out, they will just stick with the free stuff. For example with JNC even at the rate they put out new LN, I read 7-8 new volumes for free each month, and despite buying 10 credits a month don't really have enough to buy every volume. I'm probably pretty high up there in people that spend. If there were 20-25 free new volumes each month because they put them out in 20 days instead of 70 days, there would be even less volumes I could buy each month. That's probably doubly true for people that buy less.

    TL;DR - Companies that need to turn a profit can't compete in speed with places that have no labor costs, no liscensing fees, and don't have to worry about market saturation.


  • Member

    @db0ssman Great response. You do make a lot of sense. And thanks to that, i do have a clearer picture. Thank you.
    But you should be aware, the examples i gave about Wuxiaworld.com are official releases. As in, they have the rights to translate. They negotiate with authors and/or the chinese companies that owns the rights for the webnovels. They also pay their translators as well as editors. So i'm not quite sure why you are talking about fan translation:p That is why i became so curious. Both are companies that are involved in making profit through official & legal translation. But one side are waaay faster. Webnovel is a shit site and i won't mention it because quality there can be as low as machine translation. But wuxiaworld has always had great quality. They don't accept translators unless their quality are decent. Some of the translators are even in contact with the authors so they can ask questions about certain things in the novel that to them might not make sense. Not to mention the crazy word padding the authors use since they get payed by word count.
    Wich brings me to another question, are there big differences between LNs and Webnovels? Never really read much Japanese webnovels. Mostly stuck to LNs.


  • Member

    @LegitPancake Yeah. That post made good arguments. Never really took into account all the quirky accents and personalitites most japanese characters often have. In most xianxia the characters are rather similar in attitude and behaviour. So i can guess the xianxia ones are quite easier to translate actually, speech-wise at least.


  • Premium Member

    @KopiCAT said in Difference between translating Japanese LNs and Chinese webnovels?:

    Both are companies that are involved in making profit through official & legal translation.

    I've never heard of the site before so I just poked around for a few minutes yesterday and that means what I saw might not be representative of the website as a whole. With that said, the first thing I checked was a Korean web novel with over 600 chapters listed and its licensing status was in negotiation. That means it's been posted for 2 years with no licensing. It's possible there is some kind of unofficial agreement to host it or that it had a license that lapsed and they need to renegotiate. I just used the logic if such a long-term project hadn't even been licensed yet, then it is probably pretty common for new ones to not be.

    The translators and editors also weren't credited in any way which kind of suggest that either they don't keep set ones, or they aren't being attributed for pay/compensation reasons. I'm not sure what you would call it to translate in the hopes of getting paid when it is actually licensed vs a fan translation, but consider it that I guess.

    My statements are more about web hosting for translated WN in general than that Wuxia place in particular, but I'm not sure it's quite as upstanding as you seem to believe.

    @KopiCAT said in Difference between translating Japanese LNs and Chinese webnovels?:

    Wich brings me to another question, are there big differences between LNs and Webnovels? Never really read much Japanese web novels. Mostly stuck to LNs.

    I don't really read the WN either, but based on what a lot of the afterwords in the LN say, it sounds like they take the base of the story written in the WN and then restructure it to fit more typical book formats and conventions. They might also flesh out certain ideas, add new characters or arcs. The amount of changes, additions, and such varries a lot based on the series. For example, the author of Lazy Dungeon Master has said that he added in characters, significantly changed characters, taken what are basically a paragraph or two in the webnovel and made a book from it, and restructured the story to fit novel format.

    On the other end of the spectrum, Black Summoner very much shows its web novel roots; especially when it comes to setup and payoff. Most problems are just solved on the spot with a never before seen power or ability which tends to happen with the ad lib nature of web novels, and that stuff usually gets fleshed out and reworked when turned into a LN.


  • Translators

    To address your question about the speed of releases from a translation perspective, 8 weeks for a ~300 page LN volume is a pace that allows a translator to fit the work into a schedule on top of full time work and still have maybe something left over for everything else life entails. Speaking personally, it's still pretty grueling to do 40+ hours a week of a day job plus the weekly JNC release. It's perhaps a little easier if you work part time for your day job, but not everyone is in a position to do so, especially if it would entail giving up an already stable source of income.

    You might be thinking "why not just translate full time?"

    The reality is that much of JNC's (and the industry's at large) contractors do their work on the side. The reason for this isn't about any single company's rates, but rather the fact that the LN industry is not equipped to dole out fat sums of money to the people working in it. Margins are thin all around, and everyone's just trying to keep things afloat and turn enough of a profit to sustain some form of growth. For the contractors, this means going full time would require us to juggle an inherently stressful freelancer work schedule for pay that very well could be lower than a stable day job. Many choose not to take that leap of faith/passion. After all, faith/passion isn't worth much if your car breaks down, your bank account's empty, and your next invoice isn't due for another month and a half.

    It's true that full time work would allow most translators to crank out volumes much more quickly, maybe even twice as fast as JNC's current schedule, but I trust you see the problems that would arise should JNC fail to enforce a pace that is reasonable for all its contractors. There would be no end to complaints in the vein of "why my series take so long when quof is doing bookworm so fast," which would only sour relations between staff and readers.

    I have no doubt heavyweights like Steiner and quof could put out volumes at greater speed than they are now, but spacing the releases out also gives them more schedule flexibility and allows them to juggle more simultaneous projects, which is a necessity for having consistent income (as a freelancer, once a project finishes, there's no guarantee you can find another one to take its place right away, meaning your income immediately stops). This applies to editors as well, and possibly more so, since they have to wait on translations before beginning their work.

    So in summary, I have no idea how quof stays sane.

    Wait, no, that's not it. In summary, JNC's pace is probably nearing the limit of what can be feasibly maintained, given current industry pay standards and circumstances at large. You can go faster, you can go slower, but that depends heavily on the individual in question, and as a company, it's necessary to have a uniform pace that all workers can reasonably conform to.

    But who knows. Maybe one day machine learning will come out with DeepTranslate, and we'll all be out of a job while you people can just plug 100k moji into a program and enjoy immediate gratification. So long as humanity doesn't implode in the next century, anything is possible.

    Addendum
    I'm not familiar with the CN->EN scene, so I can't offer any useful insight. Perhaps their business model is different. If it's possible to hire translators on staff and have enough work to keep them stably employed for years and years, I can see how they'd be able to pump out translations at that kind of pace. Also, there's a lot of money floating around in China, so maybe it pays better in that scene as well (or worse, maybe? If they can hire Chinese people who can write English well, they can certainly get away with much lower pay). I'm uneducated in this topic though, so this is pure speculation on my part.


  • Member

    @db0ssman Fair enough. Again, thanks for the input. Learned a lot. Been using that site for years. So we have in fact a lot of contact with the translators through the comment sections. And also the forums. Everytime a new novel is introduced, RWX, the boss, also introduced the translator who will work on it. So in a sense, that's is where the credit is i guess? Never really thought of it in that vein.
    For anyone interested in learning more about wuxia/xianxia, deathblade has a great Youtube channel. He's one of wuxiaworlds biggest translator. And one of the biggest in the world as well i believe(within this genre). With over a million or two translated words under his belt. https://www.youtube.com/channel/UCgf8s70_dmcDIqXruv2MS2g
    He also has a great video about how long it takes to translate a chapter :)

    Edit: About the licensing. I don't know too much. Only what they've said. They are still somewhat new to the Korean novels. And they have removed many novels from the site because of licensing issues. In China, the authors don't hold the rights to their novels, but in Korea they do. So they have to negotiate with each indivdual more often i believe, but in China as far as i know, there is usually one or more companies that hold the rights to most of the novels. So negotiating is easier.


  • Member

    @DeiLight Yeah. I honestly don't know how they can do it as a side job. Credit where credit is due. When i'm back from work, just having to do house chores is hell :p And also thanks for your input. Again, i learned a lot. And as i have mentioned, i honestly like the setup J-Novel has. Since even though it takes time for volumes to complete, i follow enough series here that i have something to read every day. Keep up the great work :)
    Even if machine translation hits a point where it's good, i would still prefer human translators. I feel that is also better for the community. Considering that a lot of feelings are involved in how words are conveyed, i honestly don't see machines doing a better job.

    Edit: I also am a firm believer in that people should be allowed to do their work at a pace that suits them. I've always liked it better when translators use a pace they can handle. And then deliever what they say, rather then overpromise and later find out it's too much. Slow and steady wins the day.


  • Premium Member

    @KopiCAT said in Difference between translating Japanese LNs and Chinese webnovels?:

    And one of the biggest in the world as well i believe(within this genre). With over a million or two translated words under his belt.

    I know @DeiLight was half-joking about @Quof but between LDM and AOB I'm pretty sure he has over 2m translated already... Though to be honest the most prolific translator in the world is probably someone no one's ever heard of that's been quietly translating books for 50 years.


  • Member

    @db0ssman Yeah no doubt. And i did mean within the wuxia/xianxia genre. No doubt at all that there are just as big or bigger elsewhere. Might even be within the wuxia/xianxia as well for all i know. After all, what i know is what i read. But what i read is not all there is :D


  • Translators

    In general, fast translations have three sources: multiple translators working on one project, MTL (with possible human editing), and crazy people. In addition, the quality of the TL is like a modifier, where (in general) the better the TL, the longer it takes. I don't know enough about wuxiaworld to have a solid answer about which it is here. I read chapters from 5 different series and did a bit of googling, but of course that's not enough to know for sure.

    CN->EN more commonly has multiple translators on one project. I'm not sure if it's just due to China's massive population, or it being easier to learn the two languages, but it definitely seems like China has way more TLers available. Sometimes I have seen massive projects get tl'd in 2 weeks since the Chinese side dumped 10 translators on it. Does wuxiaworld do this? I have no idea. One thing to know is, though, that not everyone is honest about this, and even just two translators can double the speed of a translation (though often with significant quality reduction).

    MTL with human editing is something that can never be discounted without a direct comparison of the CN and the ENG, but as I don't know Chinese, I can't do that. In general when it comes to web novels I've become very paranoid about people MTLing. Since it's connected to making money a lot of them lie very extensively about it and do the best they can hide it. The amount of lying that goes on is actually insane, for the JP->EN scene at least. I'm actually not sure what the quality of MTL is for CN-EN, but the better it is the more plausible this is. I can't confirm this for wuxiaworld since I don't know CN, and without the source language to check a human editor can make even the worst MTL make sense, so again I don't have much concrete to say. (For the record, I've seen someone MTL a LN volume in 3 days and then be very adamant that it was a real TL. And people were adamantly defending it as real.)

    And finally, crazy people. The thing about translation is that it takes a lot of mental focus and energy. For JP->EN I generally see people have diminishing returns after about 4 hours of work (usually with breaks and "contemplation sessions" between lines to roughly pad it out to an 8 hour work day. These are necessary to keep your sanity). They get slower and worse as they lose focus and energy. But, let's say someone instead gets diminishing returns after 8 hours. This crazy person could translate twice as much as the other person, and that's that. The person with a 4 hour limit before diminishing returns kick in just can't translate as much as the guy with an hour 8 limit. Plus, JP->EN has so much demand right now that most translators are on multiple projects. There's just not enough TLers to go around. So the 4 hour limit is then stretched out across 2-3 projects. It's always possible for a crazy person with an 8 hour limit to end up with only 1 project, and then translated seemingly 4x as fast as a 4 hour limit person with 2 projects. There's really nothing that can be done about this - people have their limits, and even if some other company has a bunch of crazy people working long hours on a single project, that's not something that can be "fixed" on our end - we can't just go out and hire a bunch of crazy people and force them to only work on 1 project. That's just how the cookie crumbles. As a side note, if CN->EN is as easier to translate than JP->EN as I've been told by more informed people, then it might be the case that a ton of CN->EN translators have a way higher work limit before they get diminishing returns. I don't know too many of them myself so I can't say one way or the other.

    Then, there's the quality modifier. I try not to criticize other translators, since I feel that I will be karmically punished with more people criticizing me. In general it doesn't feel good to criticize people and especially in translation I don't necessary know what was going on in their mind. However, to say this as neutrally as possible, I don't think that wuxiaworld translations are being slowed down by taking more time for quality. I read 1 chapter from 5 different series and they all had that very distinct bullet-point feel of a fast translation. They also felt poorly adjusted to English writing traditions at times, like this paragraph which for some reason refuses to use pronouns:

    "Yes, Grandfather." Huang Wei answered and stepped forward into the hexagonal patterned light beam. As the beam of light fell on Huang Wei, everyone in the hall focused on Huang Wei who was now standing within the beam of light.

    Whether this was inexperience with English writing or a side effect of MTL translating, I have no idea, and I don't want to presume.

    In conclusion. I am sadly incapable of providing a solid answer to you because I don't know Chinese, and I don't know the inner workings of wuxiaworld. I would need to know Chinese to check for MTL, I would need to know the inner workings of wuxiaworld to confirm whether they used multiple translators, and I would need more familiarity with their community to know if they just hired crazy people. It's one or all of these three, with the speed buffed by not exactly putting out extremely high-quality TLs.

    For J-Novel to translate faster than 2 months per novel, a variety of things would have to happen. Putting aside all the logistics of releasing ebooks on Amazon, since that's outside of my purview, the translators would first have to stop working on multiple projects at all (which j-novel has no control over since translators can get work from elsewhere as well). Then J-Novel would have to intentionally focus on hiring crazy people with an unnatural ability to translate quickly and for long periods of time. I don't know how many of these people actually exist, so I don't know how reliable it is, but I don't know many of them personally. Then J-novel would have to consider putting multiple people on one series to make 2 months 1 month - some sites do this without mentioning it, and it's pretty standard for video game translation. If we were really getting nutty, J-novel could assign 4 people to a series and have one volume per 2 weeks after a warmup period. This sounds crazy but I have seen a Chinese -> English project with 10+ translators on it. Quality would suffer, but the speed improves. And finally, the quality modifier. J-novel encourages translators to write in past-tense, the standard for English prose, and structure the text into paragraphs in accordance to common convention. If these were dropped, and translators encouraged to prioritize speed over good, we could probably go from 2 months to 1 month on that alone. It would be much faster to not have to wrangle LN prose into past tense.

    Okay I think that's as comprehensive as I can be. Hope this helps. I'm not actually sure paying TLers more and letting them go full time will increase speed, since I think the tendency is for people to take multiple projects when they go full time, and that cancels out the extra time they get. It's also worth noting that as far as I can tell J-novel is significantly above the curve when it comes to speedy J->E translation. So we're already pushing the non-crazy people to their natural limit I think.



  • tl;dr answer would be:

    quality
    speed
    cost

    pick two.


  • Premium Member

    @db0ssman said in Difference between translating Japanese LNs and Chinese webnovels?:

    I know @DeiLight was half-joking about @Quof but between LDM and AOB I'm pretty sure he has over 2m translated already... Though to be honest the most prolific translator in the world is probably someone no one's ever heard of that's been quietly translating books for 50 years.

    Wouldn't be surprised. Novels used to be approximately a million characters long. This was prior to some of the door stoppers we currently see. And average word is about five characters, so call it about two hundred thousand words per novel. Light novels tend to be a bit shorter, so call it a hundred fifty thousand words. Twelve for LDM and nine for AoB, gives a rough estimate close to three million words.


  • Translators

    @nofairytale Nonono. I don't think cost factors into the question at all. Paying translators more won't make them work faster. In fact there's this funny thing where really high-profile translators for games can make huge bank but they don't actually work that much faster. If you aren't being paid so little that you just stop caring about a project entirely, I think that money has 0 influence on TL speed or quality. (generally people do the best they can do with what they're given, and they don't like subconsciously do worse because they're being paid 4 cents a character instead of 5. etc).

    Quality and speed are the important factors, but even then the focus isn't on a normal translator prioritizing one over the other, so much as it's about whether you use a normal translator at all. It's about how close to a normal human you are. The less human you get, the faster the translation is. MTL is the fastest, then multiple humans at once, then individual crazy people, and then, finally, normal people, where the quality of their TL modifies the speed a bit. The less human you are, the faster the TL is, and in topics like this, the answer is "they are being inhuman in some way or another". It's kind of a coincidence that the more inhuman you are the more the quality drops. Once the AI singularity hits, then MTL will be divine texts produced by god-like machines, and at that moment, quality and speed will be unified.

    (I don't really like pithy sayings like "quality/speed/cost pick two" since it implies the workers base the quality/speed of their work on their pay, which is not always true, and it also oversimplifies how the quality/speed is being obtained - all the nuance of MTL, multiple translators, etc is gone.)


  • Premium Member

    @Quof said in Difference between translating Japanese LNs and Chinese webnovels?:

    (I don't really like pithy sayings like "quality/speed/cost pick two" since it implies the workers base the quality/speed of their work on their pay, which is not always true, and it also oversimplifies how the quality/speed is being obtained - it implies that quality/speed can both be obtained by paying more, which is not true.)

    Then I suspect you don't realize the tradeoff. Someone could get a high quality at high speed by spending a lot of money. They would do so by hiring a large team, not by paying more to the same small team. And yes, throwing money at a problem is frequently useful in solving the problem. Although many times a large team runs into the problems Fred Brooks mentions in The Mythical Man Month in that communication overhead between the team members dwarfs the additional man hours available by a large team. And additionally, people are not fungible, which all too many managers seem to forget.


  • Staff

    Just to echo jcochran, the adage isn't about the worker per se, it's more about the customer, either whoever's buying the product or whoever's paying the employee. As the person with money, it's best to expect only 2 of the categories out of the 3: if you want something cheap, you can expect to have it quickly or done well, but not both; if you want both speed and high quality, then it'll cost you.


  • Premium Member

    I've mentioned before that the quality of the translation depends on the translator knowing more context about a particular work than simply applying a dictionary to the input. The original post mentions Mahouka Koukou no Rettousei, which has had a number of different translators, both legitimate and non.

    The legitimate translators missed some context in the beginning, such as the importance of Miyuki calling Tatsuya "Onii-sama", and substituted the use of his given name. The fan translators decided to use the un-translated "Onii-sama", instead of trying to come up with a better English equivalent that wasn't clumbsy.

    The MTL translation... has no consistent context. As I mentioned before, the gender of characters is all mixed up, because the MTL doesn't know it from past encounters with the character's name.

    Giving 2, 3, or more people responsibility for sections of the book can run into the same thing, if they're not communicating among the group. A textbook or technical manual is different from a novel.

    Note: I do not translate between human languages for a living, but I do work in translating specifications from "human ideas" into "what a computer understands".


  • Premium Member

    @jcochran said in Difference between translating Japanese LNs and Chinese webnovels?:

    And average word is about five characters, so call it about two hundred thousand words per novel. Light novels tend to be a bit shorter, so call it a hundred fifty thousand words.

    Most LN are between 50 to 80k words with some exceptions like Slayers being around 40k and AoB being around 100k. If you want to know the word count of a book, you can go to Kobo and look up a book. It will tell you the word count. I don't think an 'average' novel at any time was around 200k words. If you look at some infamously long books like War and Peace (500k) you find massive ones, but those are really more of the exceptions.


  • Member

    @Quof Wow. Thank you for all that. I as well hope i didn't come across as criticizing you and the others for your speed here. And if so, i'm sorry. As i mentioned in the OP, my frustration comes mainly from a novel you guys aren't even involved in:p
    And i am also very very far from an expert, simply a fan. Hence, i also would certainly not know if it is truly MTL with some godlike editing. But i do consider myself rather critical about quality. I can be a bit of an a-hole about it at times. I really notice grammar mistakes and inconsistent sentences. It just completely takes me out of the immersion. Something i rarely encounter both here and on Wuxiaworld. Obviously it happens, both because i myself might not be aware that it's acually correct, and wrongly thinking it's actually a mistake. And also because translators are only human. I'm not a native english speaker, so i'm sure i miss things and make mistakes as well. So i will say i really like the quality here and on wuxia. Here i get my fix for Japanese LNs, there i get my fix for xianxia webnovels. And i love you both <3
    And i've been given lots of useful information from all of you and have gotten a pretty good picture. Most of this i had no idea about, since as mentioned, i've stayed quiet as i don't really like coming across as complaining when i read this almost for free.


Log in to reply