狂笑 Why India is struggling with localized language computing
Why India is struggling with localized language computing (NewsForge)
(Posted Aug 17, 2004 20:02 UTC (Tue) by khim) (Post reply)
“Continuous innovation and improvement!” is good idea but on practice… instead of changes to make programs smarter people just accept computer speak broken english. I can not speak for Indians but here are sample from Russian.
It’s in fact a long story about why not a single one non-native program have correct translation for simple two words phrase “N file(s)”.
Yes, i18n rules dictate separate translation for N == 1 and N > 1.
But does it help much with Russian?
Not at all! Why the hell not?
Well… Here is correct tranlations:
* N = 1: “1 fajl”
* N = 2: “2 fajla”
* N = 5: “5 fajlov”
* N = 10: “10 fajlov”
* N = 11: “11 fajlov”
* N = 12: “12 fajlov”
* N = 21: “21 fajl”
* N = 22: “22 fajla”
* N = 123: “123 fajla”
Now… Can you see the problem ? We have three variants of the same word and it’s not just one for N == 1, other for N == 2,3,4 and still other for N >= 5… I’ve yet to see i18n package where this problem has solution (few Russian programs where I’ve seen correct language – mostly from banking area – use hardcoded algorythms suitable only for Russian).
I’m pretty sure there other similar problems but this is thing I’ve biten the most. Translation of “New->Folder” as “Novy->Papka” is just curious (correct translation is “Novaya->Papka” while “Novy->Papka” is in fact correct russian phrase – it just means “New->Dad”). There are a lot of such problems with other languages, I’m sure…