I was thinking which is better way to allocate the testing work.
We have a software which need to be tested on four different languages. So I have to ways to allocate the work.
1. Tell Different person to test whole software in single allocated language.
2. Divide the software in different modules between people and tell them to test the allocated module in all the language.
Both has their advantages and disadvantages. I mean in first case a person will come to understand the whole system, so will be able to test it better. However it would require more time to execute the test cases as compare to second choice, as in second option person have to concentrate on single module only he will be able to grasp it more quickly and chance are the localization errors will be discovered through this option will be more then choice one.
Does everything about your software actually need to be tested equally in every language? If so, that's unfortunate.
Many shops would separate the local aspects (language, currency, dates, icons, etc) from the code itself, and divide up the testing like this:
1 Overall functionality
3 Localization for language 1
4 Localization for language 2
5 Localization for language 3
6 Localization for language 4
Only one language is needed for step 1. Usually, it would be the dominant (or "home") language.
The ability to correctly support multiple languages is the focus of the testing in step 2.
Then, testing each language in turn for steps 3-6 is much quicker, and need only be performed folks sufficiently fluent in the target language/culture.
That's the way I've done it in the past - it worked well.
I totally agree with Joe. Also, just as a side note, any time that you're doing localization testing, it can be a good idea to get someone who actually speaks the language to look over it.
We had a couple instances where the translation was actually quite insulting to the reader. Of course we'd have no idea if we didn't have people internally who could speak the language. It's just little quirks here and there I think. So someone who can speak the language fluently are assets.
9 out of 10 people I prove wrong agree that I'm right. The other person is my wife.
what kind of software?
i think if it's sdk and you can see the code, you may just compare 4 packs (for example, with Araxis Merge) to see that there are differences with texts only, than check semantic conformity and test just one version.