Originally Posted by
fumei
KT was from Korean, but honestly, KT is bad, like REALLY bad, so I don't think it's an appropriate comparison.
Honestly, I'm not really even sure why the distinction is so important to Leftovers considering the workflow we're dealing with here. Because basically, the pitfalls of double translations should be obvious: you lose out on some (or a lot of, worst case) nuance, things like metaphors, wordplays, tone and metre - and all that's just when translating from JP into the first language. Now you go through that entire process again to get it into English, and there's obviously a lot lost in translation. I don't think anybody is really denying that this makes double translations kinda bad.
The way we've chosen to handle this project is specifically to counteract these issues with double translations.
1) Translate from FR > EN
2) Rigorously TLC between JP and the resulting EN
3) Proofread and edit for consistency
The French translation into English gives a basis for the first pass of TLC, and unlike what most people seem to think we're doing, we're not just using that and changing some slight errors or mistranslations, but simply using it as a basis. Effectively, we're invoking Cunningham's law in a translation context, because aside from having to match tone and such from JP, it removes the need for the JP translators to spend as much time and energy on choosing the right phrasing and sentence structure, or even just thinking of the right word in English, because the FR>EN already did that for us. JP proofreading is still likely the longest and most "intense" phase of this project, as we're going over every single line, and for the most part we're changing something in every single line from that French basis to be more accurate - effectively removing the issue of things being lost in translation (twice), insofar as that's possible and we're not losing things in our own translation, which just seems inevitable to some degree.
So then why not just translate directly from JP if we're still going over every single line and basically translating it from JP already? That's exactly because of what I mentioned about having a basis that allows for less energy spent per line. It's just much easier to point out what is wrong with a line than it is to give the correct line from the start, and I'd say this allows for the project to move ahead at a smoother pace.
Since we're multiple FR & JP translators, there's additionally an issue of consistency there, and we're basically just alleviating this by a consistency pass in editing by a single editor after initial editing (for everything else) is done, same with retention of metre and tone.
I didn't really mean for this to get so in-depth, but I think it's good for transparency's sake, and hopefully explains why I think it's kinda weird for people to be so hung-up on the fact that it's a double translation.