> It does a *terrible* > job handling Chinese, Japanese, and Korean.
Does just fine for Japanese in a lot of cases. The result can be a bit robotic if you expect native sounding English but it usually makes enough sense for you to get what is going on. The live OCR feature of the google translate app is pretty good even on written text now. I use it to decode kanji dense documents like those that come from the tax office that even a native has trouble with comprehending from time to time.
> (in Chinese your tone of voice for each > syllable changes the meaning).
Why would those issues matter for written material?
> Most Japanese schoolchildren learn a little bit of English in school,
Japanese children are now learning English from kindergarten (source: my kids are). Most high schoolers learn a lot of complex vocabulary. Many of them are absolutely awful at speaking and listening because they have no real world experience of doing it. Reading is a lot better. (source: my wife can't speak English for shit but is fairly capable at reading. She only has high school level education.)
> Plus, computers and emulation have a specialized vocabulary which isn't covered in > normal school language courses.
For Japanese a lot of that vocab is directly imported from English. Japanese also has this phenomenon where there is quite often a bastardised English word that can be used in place of actual Japanese. i.e. you can get away with saying ribuuto instead of saikidou.
> several passes on the text since the translator doesn't always know the technical > vocabulary.
My guess would be that the translators are on the level of anime fansubbers and not actual translators.