May we get light novel translations with honorifics
-
I'm getting vibes of asking a scientist "Why not try [simple solution to complex problem]?" in this thread... the answer tends to be, "Because if [simple solution] worked, we would've already done it by now."
-
@kevin-s said in May we get light novel translations with honorifics:
@taapaye said in May we get light novel translations with honorifics:
I propose asking the translators to just encode both options in the source.
One thing I don't think anyone has brought up yet is that I just don't think this would be nearly as simple an undertaking as it initially sounds. I'm thinking of the dozens of times that names appear in any given part; making sure every one of them is accounted for and that the correct honorific is used (remember that each character won't necessarily be addressed with the same honorific all the time); the additional time and complexity not just for the translator, but for the editors and QA staff making sure everything is correct and accounted for.
You could use regular expression to check whether everything is accounted for or not. A small program can be used to scan the document and check for every names that have not been accounted for (given a list of names that might appear in the book).
Regarding replacing names automatically with their honorific counterparts, as long as the English naming is consistent, it shouldn't be a problem. But, if they differ, then as long as there are no omitted Japanese sentences in the English translation nor are there added sentences in the English translation (sentence count between JP and EN is the same), then this could also be done automatically with a program.
-
@otonokakuei said in May we get light novel translations with honorifics:
Regarding replacing names automatically with their honorific counterparts, as long as the English naming is consistent, it shouldn't be a problem. But, if they differ, then as long as there are no omitted Japanese sentences in the English translation nor are there added sentences in the English translation (sentence count between JP and EN is the same), then this could also be done automatically with a program.
This is not a safe assumption. One sentence can be translated as two sentences or two sentences can be translated as one sentence. Neither of those is an omission or an addition. Different languages have different ways of expressing the same idea.
-
@myskaros said in May we get light novel translations with honorifics:
@otonokakuei said in May we get light novel translations with honorifics:
I would say "why not?"
Because it is not as little work as you're making it out to be, and it would not sell enough additional copies of books to make up for the additional time and labor required to make this a reality.
As previously mentioned. LaTeX, for example, can do this already. In the case of using LaTeX, the work that would need to be done would be:
- translators have to adapt to using LaTeX instead of word or whatever word processor they were using
- JNC also need to make changes to their website to reflect this change (show different versions, settings, etc).
@stardf29 said in May we get light novel translations with honorifics:
@otonokakuei said in May we get light novel translations with honorifics:
I agree that there will be extra work for the translators, though I believe it won't be anything too excessive.
Unfortunately, the amount of extra work for translators this would be is not really something those of us strictly on the consumer side can judge reliably. Suggestions like these should be made with the attitude of "might be worth considering if it's not too much work, but if it is, then don't bother". We fans should avoid entering that dangerous territory of making demands of translators and editors under assumptions of how easy or hard we think it would be; the last thing we want is for "fans" to cause issues with working conditions for the localization team.
If that's the mentality we should have, then I suppose there's no real reason in having a discussion in suggestion threads. Staffs should first enter the thread and say whether they will consider the suggestion or not before we can engage into discussion.
@rsog412 said in May we get light novel translations with honorifics:
@otonokakuei said in May we get light novel translations with honorifics:
Regarding replacing names automatically with their honorific counterparts, as long as the English naming is consistent, it shouldn't be a problem. But, if they differ, then as long as there are no omitted Japanese sentences in the English translation nor are there added sentences in the English translation (sentence count between JP and EN is the same), then this could also be done automatically with a program.
This is not a safe assumption. One sentence can be translated as two sentences or two sentences can be translated as one sentence. Neither of those is an omission or an addition. Different languages have different ways of expressing the same idea.
Yes. In that case, it'd not work unless there's a dedicated software that could output a map of translated sentence(s) and the original sentence. Which of course wouldn't be as trivial to do compared to the previous solutions.
-
@otonokakuei said in May we get light novel translations with honorifics:
If that's the mentality we should have, then I suppose there's no real reason in having a discussion in suggestion threads. Staffs should first enter the thread and say whether they will consider the suggestion or not before we can engage into discussion.
There's a clear difference between "maybe this idea can work" and "this idea is totally easy to do". But if your idea of "discussion" on suggestions involves making assumptions on how easy it will be to implement it, then yeah, I'd say don't have that "discussion" because at that point it's more like making demands.
translators have to adapt to using LaTeX instead of word or whatever word processor they were using
Keep in mind, translation is a freelance job and forcing a bunch of freelancers to adapt to new software, whether for a specific company or for the entire localization industry, is way more complex than just getting the folks in a single office used to new software. And that's not even taking into account whether or not LaTeX might have disadvantages in other areas compared to existing systems that make it not worth the "advantages" of appealing to a niche group of readers who like their Japanese honorifics a lot.
MAL
Beneath the Tangles (the blog I write for) -
@otonokakuei
full disclosure: I do not speak Japanese, I am an English speaker with a smattering of several Latin based languages (Spanish/French/Italian)these "rules" you keep referring to - automatically substitute title "a" for honorific "b" or visa versa - may work when translating related languages with similar rules (like Castillian to Catalan) but fail when the languages are structurally (and culturally) very different- that is why MTL sucks - every time a substitution might be applied, human judgement needs to come into play. Interpretation isn't that simple. I assume that the translation team come up with something like a 'style guide' when starting a project, so that tone, character mannerisms etc. are consistent throughout the work, and if/how/when honorifics are used is part of that 'style' - it is integral to the interpretation of the work. If a word for word substitution were sufficient to translate a work - google would've put JNC out of a job already - we'd all just read MTLs
you can put ketchup on a steak, should you?
-
@otonokakuei said in May we get light novel translations with honorifics:
If that's the mentality we should have, then I suppose there's no real reason in having a discussion in suggestion threads. Staffs should first enter the thread and say whether they will consider the suggestion or not before we can engage into discussion.
I think we can engage in discussion just fine - but our discussion should be limited to our own experience. We can talk about how much we want/don't want/don't care about a thing, but we shouldn't make assertions about how easy it would be to implement.
Taking my own advice here, I'm going to stop arguing the technical points. While I know enough Japanese to read novels (very very slowly, with the help of a dictionary), I've never done translation professionally and wouldn't be able to. My experience in that only comes from looking at translated works and comparing with the source text. In the end, I only see things from the consumer side.
From the consumer side:
- If this feature existed in the form you describe, where the Japanese address and honorifics are re-inserted back into the final product after a translator has done whatever they can to render the meaning that the honorific conveyed into English, I would not use it. (Please don't construe my acknowledgement that this cheaper form is possible as me agreeing that creating it would be cheap or easy.)
- If this feature existed in its ideal form, where the net result is as if the translation team had done the translation twice in different styles (one with honorifics left in, and one with honorifics rendered into English), I would probably use it for at least some series.
-
@stardf29 said in May we get light novel translations with honorifics:
@otonokakuei said in May we get light novel translations with honorifics:
If that's the mentality we should have, then I suppose there's no real reason in having a discussion in suggestion threads. Staffs should first enter the thread and say whether they will consider the suggestion or not before we can engage into discussion.
There's a clear difference between "maybe this idea can work" and "this idea is totally easy to do". But if your idea of "discussion" on suggestions involves making assumptions on how easy it will be to implement it, then yeah, I'd say don't have that "discussion" because at that point it's more like making demands.
My argument was that "such and such idea could work because I believe there are pros to the idea as well the amount of effort that would be needed to do such work". It is not a demand in any way, not quite sure why you see it as that. Yes, I emphasize how easy it is to implement to sell the idea. It is not a baseless assumption, since it's based on the experience I've had working as a software developer.
Keep in mind, translation is a freelance job and forcing a bunch of freelancers to adapt to new software, whether for a specific company or for the entire localization industry, is way more complex than just getting the folks in a single office used to new software. And that's not even taking into account whether or not LaTeX might have disadvantages in other areas compared to existing systems that make it not worth the "advantages" of appealing to a niche group of readers who like their Japanese honorifics a lot.
Well, I can't really reply to this since you don't have an understanding of how LaTeX works. If I say it's not that hard to use since you can use templates and such, would you say that this is yet another demand? Sigh, let's just say I give up. The staff already said it's not worth the effort anyway.
@jon-mitchell said in May we get light novel translations with honorifics:
@otonokakuei
full disclosure: I do not speak Japanese, I am an English speaker with a smattering of several Latin based languages (Spanish/French/Italian)these "rules" you keep referring to - automatically substitute title "a" for honorific "b" or visa versa - may work when translating related languages with similar rules (like Castillian to Catalan) but fail when the languages are structurally (and culturally) very different- that is why MTL sucks - every time a substitution might be applied, human judgement needs to come into play. Interpretation isn't that simple. I assume that the translation team come up with something like a 'style guide' when starting a project, so that tone, character mannerisms etc. are consistent throughout the work, and if/how/when honorifics are used is part of that 'style' - it is integral to the interpretation of the work. If a word for word substitution were sufficient to translate a work - google would've put JNC out of a job already - we'd all just read MTLs
It's just a dictionary based substitution, nothing too complicated. It doesn't really matter whether the structure's similar or not as long as it's possible to find the equivalent sentence in the raw. If let's say we want to substitute "Mr. Bell" with "Bell-[honorific]", we'd have to find the equivalent sentence in the raw, find the word "ベル" in that sentence, extract the honorific trailing behind the word and do another lookup in dictionary to see how the honorific should be translated to English, then substitute [honorific] with the actual English honorific. Why the rules don't matter is because we're only looking for specific words (name) within the sentence.
you can put ketchup on a steak, should you?
Funny you ask this. I usually do add ketchup (chili sauce actually) to my steak if I don't like how it originally tastes.@rsog412 said in May we get light novel translations with honorifics:
@otonokakuei said in May we get light novel translations with honorifics:
If that's the mentality we should have, then I suppose there's no real reason in having a discussion in suggestion threads. Staffs should first enter the thread and say whether they will consider the suggestion or not before we can engage into discussion.
I think we can engage in discussion just fine - but our discussion should be limited to our own experience. We can talk about how much we want/don't want/don't care about a thing, but we shouldn't make assertions about how easy it would be to implement.
The "assumption" that it's easy to implement is based on my experience as a software developer for 9 years.
From the consumer side:
- If this feature existed in the form you describe, where the Japanese address and honorifics are re-inserted back into the final product after a translator has done whatever they can to render the meaning that the honorific conveyed into English, I would not use it. (Please don't construe my acknowledgement that this cheaper form is possible as me agreeing that creating it would be cheap or easy.)
- If this feature existed in its ideal form, where the net result is as if the translation team had done the translation twice in different styles (one with honorifics left in, and one with honorifics rendered into English), I would probably use it for at least some series.
I don't think I've argued that I prefer having honorifics more or not in my posts. I just believe that having the option has its pros that could make it worth considering.
Personally though, as you said in your first point, I have no problems with not having honorifics if it's translated like that. I think I only abhor the lack of honorifics in Danmachi and Shield Hero.
-
@kevin-s
In regards the automatically replacing "Mr. Tanaka" with "Tanaka-san", I don't yet fill this is infeasible. The high level assumption was "Mr." would also come from "-san" and not "-san" to "Mr." I don't know if this assumption holds, but "-san" being translated to possible multiple things isn't a real issue. The only other example I no Fuji-san being translated to Mt. Fuji. So in this limited example we could map- Mt -> san
- Mr -> san
- Mrs -> san
Now if it just gets dropped, that is unfortunate. However, it would claim that if it gets dropped or if the many to 1-to-manny mapping isn't true then we've already lost the info and are no worse than the current version at the very least
I do agree we shouldn't lower translation quality to force this system to work. If anything we should just accept missing honorifics in a few places. Your explanation about the pronouns seems convincing enough to accept that it is too much work. For the narration, it also seems fine omitting the honorifics since they don't seem to add anything.
I'm not trying to be a purist. If it just covered the majority of cases, I'd be content.
I do really appreciate the concrete examples.
-
@otonokakuei The assumption that since it's easy for a software engineer to implement, it must be easy for a freelance translator to implement is, frankly, probably not as correct an assumption as one might think. I've been working in libraries for years and can learn how to use a new piece of cataloguing software quickly. I'm not sure I'd expect an office worker to be able to figure out how to work cataloguing software all that quickly.
My grasp on sanity remains absolute.
Oh, right, the poison, the poison for Kuzco, the poison chosen especially to kill Kuzco, Kuzco's poison. ...That poison?
-
@thegrimlich said in May we get light novel translations with honorifics:
@otonokakuei The assumption that since it's easy for a software engineer to implement, it must be easy for a freelance translator to implement is, frankly, probably not as correct an assumption as one might think. I've been working in libraries for years and can learn how to use a new piece of cataloguing software quickly. I'm not sure I'd expect an office worker to be able to figure out how to work cataloguing software all that quickly.
I've never said that the translator should be the one implementing it though? The format (if LaTeX is not used) and the parser/epub-converter should of course be implemented by someone who could. Translators only need to adapt to using said format. And by adapting I meant getting used to writing "\v1{Mr. Bell}\v2{Bell-san}" or whatever syntax the format requires instead of "Mr. Bell".
-
@thegrimlich
I believe the initial approach of asking the translators to explicitly encode honorific info has been shot down as too much work already.But still implementing the idea (ie writing the parser to go to the correct format isn't hard), the objection related to, at least in part, the effort involved in do so.
Just wanted to highlight that aspect to support the viability of the 2nd idea since it wouldn't require extra work from the translators ideally.
But I guess talk is cheap. If the claim is in doubt, could draft the script as an example if someone wanted to provide test samples
-
part of the practicality assessment of making such a change, I would think, is how prevalent mis-use of honorifics/lack of honorifics is among JNC translations. In my reading of the many, many fora I just don't recall it coming up much. Imouto vs. little sister/ junior, yea a few times but the examples in this thread (other than the 1 from Realist Hero) seem to be from works translated by companies other than JNC (i.e. Danmachi, published in English by Yen Press) This would require investment in labor/training/tools - what is the return on the investment? This looks like a problem that largely doesn't need fixing (at least not by JNC)
-
I was going to avoid jumping into this discussion, but as someone who's been doing software development for years and has a basic knowledge of how Japanese -> English translation works because he's worked with translation projects in the past, it seems like some of the software developers here have a fundamental misunderstanding of how Japanese to English translation works, and are therefore making incorrect assumptions about how much work would be involved in their replacement mechanism.
There is no one-to-one reliable way to translate Japanese to English. This is why machine translation from Japanese to English sucks so bad. The structure of the language is fundamentally different, and many things that English requires you to spell out specifically are only inferred by context in Japanese, so you need the surrounding context taken into account.
Just the "simple" aspect of whether or not you have honorifics is not just a one-for-one swap out, and even if you're letting the translator create the alternatives, in a lot of cases it's the entire sentence or a whole paragraph that needs to be changed because of the presence or absence of honorifics, and not just the person's name. There are a lot of subtleties in the Japanese language which are expressed by which honorific is used in a name. When that subtlety is important to the meaning of the story for it to be preserved, doing so in a language like English that doesn't have those subtleties often requires massaging the entire surrounding paragraph to find some other way to express that subtlety in a way that still sounds natural in English. Taking that and adding the honorifics back into it afterwards would just sound really really strange, even if you do understand the honorifics. If you're going to have an honorific version and a non-honorific version, you'd want to exchange the entire paragraph in that case so that you don't need all the extra sentences explaining the subtlety that would inevitibly end up getting added in the non-honorific version.
The example I've always seen used to get this point across is when someone who has been using a specific honorific when speaking to a specific person changes which honorific they use because they've become closer friends with someone than they used to be. Usually this doesn't need to be explained in Japanese. Person A uses a more-familiar honorific, Person B is suddenly really happy. Everyone knows why Person B is happy because it's a natural bridge to cross when becoming more familiar with someone. In Japanese. In English we don't even have this concept. Person B suddenly being happy is important to the story. If we're doing it without honorifics, now we have to come up with a way to explain why. A lot of the time, the only way to pull this off is going to be to add an additional sentence spoken by Person A explicitly pointing out "I feel like we've become closer now" or similar. There's a whole sentence added that didn't even exist in the original Japanese, but to leave it out would remove meaning from the story.
Doing it both ways also requires the translator to actively think about two different ways to do it, which isn't really conducive to speedy translating. I suspect even if they tried this, it would slow down releases by a lot more than you think. It would work better to throw two different translators at it, one tasked with translating it each way. And that's more money spent on translators and twice the effort.
Just my opinions.
-
Have you considered the impact a dual system would have on series where we do use honorifics? I'm game to change Iroha's affectionate "Sumire-chan-sensei" to "Teachy-teach Sumire" if you're all up for it, but I feel like some consideration ought to also go to series where we do use honorifics.
-
@pcj said in May we get light novel translations with honorifics:
I was going to avoid jumping into this discussion, but as someone who's been doing software development for years and has a basic knowledge of how Japanese -> English translation works because he's worked with translation projects in the past, it seems like some of the software developers here have a fundamental misunderstanding of how Japanese to English translation works, and are therefore making incorrect assumptions about how much work would be involved in their replacement mechanism.
There is no one-to-one reliable way to translate Japanese to English. This is why machine translation from Japanese to English sucks so bad. The structure of the language is fundamentally different, and many things that English requires you to spell out specifically are only inferred by context in Japanese, so you need the surrounding context taken into account.The automatic replacement mechanism was not meant to be a cure-all solution. It should only be there as a way to automate some of the more trivial replacement processes. I also mentioned some pre-requisites in order for it to function correctly. In cases outside of those, it must be done manually.
@pcj said in May we get light novel translations with honorifics:
he example I've always seen used to get this point across is when someone who has been using a specific honorific when speaking to a specific person changes which honorific they use because they've become closer friends with someone than they used to be. Usually this doesn't need to be explained in Japanese. Person A uses a more-familiar honorific, Person B is suddenly really happy. Everyone knows why Person B is happy because it's a natural bridge to cross when becoming more familiar with someone. In Japanese. In English we don't even have this concept. Person B suddenly being happy is important to the story. If we're doing it without honorifics, now we have to come up with a way to explain why. A lot of the time, the only way to pull this off is going to be to add an additional sentence spoken by Person A explicitly pointing out "I feel like we've become closer now" or similar. There's a whole sentence added that didn't even exist in the original Japanese, but to leave it out would remove meaning from the story.
Fair point. Although I find it odd to do it like that, if the translator does decide to translate it that way, then it would make adding a version with honorifics a troublesome matter, especially if this occurs often.
@pcj said in May we get light novel translations with honorifics:
Doing it both ways also requires the translator to actively think about two different ways to do it, which isn't really conducive to speedy translating. I suspect even if they tried this, it would slow down releases by a lot more than you think. It would work better to throw two different translators at it, one tasked with translating it each way. And that's more money spent on translators and twice the effort.
I would certainly not suggest having 2 translators work on the same series.
Perhaps I've been too shallow in my thinking that a "name" can simply be replaced with another "name". I will add no further to the discussion, as I lack a real understanding of the intricacies that come with translating Japanese names with context to English.
-
@pcj
I think the misunderstanding lies more in how "perfect" the translations would be with honorifics forcibly added. I don't claim it would be as ideal as a person manually adding them, but I think it would be good enough. In your example with Person A/B, we'd had the extra line and also the honorific change. I don't think the redundancy detracts from anything.I think the idea would be immediately shot down if it had to involve extra translating effort. There have been some cases that the the simple find/replace mechanism can't handle, but for the ones it can, it seems to work fine without extra translator thought in most cases.
I'd love to hear edge cases/counter examples so the idea can be improved. -
So if this issue could be mollified by simple find/replace mechanisms...
I've heard of people editing the epub files that JNC's premium e-books come in, to do stuff like change the spelling of certain names to their preferred spelling. So it seems to me that such find/replace mechanisms are already possible on the consumer end. If that's the case, all we really need maybe would be a place on the forums where people can share whatever find/replace parameters they think might work to adjust the text to others' liking.
-
@taapaye said in May we get light novel translations with honorifics:
I'd love to hear edge cases/counter examples so the idea can be improved.
Maybe I'm misunderstanding - I think that the idea of a computer generating correct honorifics is basically science fiction. If that's what you're proposing, I can think of a couple difficult cases, but I wouldn't even consider these edge cases.
Example One: Suppose that the translation says, "Mr. Sugimoto," where Sugimoto is the character's family name.
Most of the time, I think the safest re-honorificification would be "Sugimoto-sama." If the translator went as far as to use any honorific at all (Mr, Lord, etc.) in the English, the honorific in the source text was probably on the more formal side.
However, that's not always the case. In a school setting, the most likely correct replacement would be "Sensei." (Maybe sometimes "Sugimoto-sensei," if the speaker needs to make clear which teacher they're talking about).
This would be possible for a skilled human to get right most of the time (which is better than most cases of trying to losslessly reverse a translation), but for an algorithm, I'd go as far as to say it's basically impossible. (Even a neural net would struggle because it would have to correctly identify the setting.)
Example Two: Suppose that the translation says, "Tanaka," where Tanaka is the character's given name. Suppose that your algorithm has a table where it can look up that the character's family name is "Sugimoto"
Unfortunately, this one is almost certainly irreversible. The algorithm has no way of knowing whether the family name or given name was used in the source material, and with what honorific (san? kun? chan? was there no honorific used at all?). Choosing the "correct" honorific would require the algorithm to understand the dynamics of the scene and the characters, and even then, it would probably get it wrong some of the time, because the source material might not have used the "correct" / most likely honorific for the situation.
-
@rsog412
Thanks for the examples. Don't know how I overlooked the teacher case. I was thinking how Professor/Master/Dr could work but overlooked how either '-san' or '-sensei' could have led to Mr. Thank you this breaks one of the assumptions I was making.For the 2nd example, I agree it is a lost cause if the honorific is dropped. I don't know under which cases nor how often that is true. One example that was brought up was was during narration, but I did kinda assume honorifics wouldn't be completely dropped in speech generally.
