How do these tools work?
In an under the hood sense, how do they work? Related to my other post about translations, I'd like to be able to grab the text for a textfield that belongs to another app and then lookup a translation for the string. I'd like to write (or find) an app that will allow me to load in english and japanese resource files for my AUT and then translate on the fly as I hover over objects in my AUT. Any suggestions?
Re: How do these tools work?
In short, Asian languages are not Latin or Germanic based. What I recommend is that you hire someone who has good experience testing DBC. It's not as simple as testing FIGS.
Quick understanding of Japnese/Chinese/Korean testing:
If one thinks that all languages can be translated word to word, or even sentence to sentence, then they do not comprehend that all languages are not the same. Try to understand the following sentence in Japanese.
"Watashiwa anataga sukidesu".
Seems like a simple translation to "I love you". But it's not always "I love you". Sometimes, it's "I like you", or even "I feel passionate for you". "Love" as defined in common English can be translated to "Koi", "Ai", and "suki". On top, these are combined with several other 'characters' to redine it. In short, there are often no 'equivalent' language matches, and requires human 'relative association' based on context of how things are stated.
Now, having a computer trying to figure such things out, is virtually impossible. There are tools out there to do some rough stuff (I belive Basis claimed to have something as such, and perhaps even Lionbridge), but it requires human edits, and often, people say it saves 10-20%(max) of translation time.
Capturing Japanese text is more difficult then English. Remember that English can be diplayed in 8 point font just fine, with most complex characters containing 4 strokes. But in Japanese and Chinese they use characters consisting of 16 or more strokes. Most people will state that minimum is 12 points. If you are going to use all 'katakana', as many did, that's fine. But anything more then that would be just too much of pain. So, capturing text is way too difficult. Last time I tested DBC was 3-4 years ago, and I don't recall a tool that did bitmap->utf8/dbc conversion.
This is just the beginning. Go hire someone who actually did real work (automating Japanese), and they can help you. This type of resource is extremely rare though. Often, you find people who are strong in language, but are technically very weak.