This is not a request for translation, but a request for comment from experienced localizers.
Is there anybody here who is experienced with testing software localization or writing test suites and test plans for testing software localization?
For Wikimedia projects we mostly translate things in Meta or in translatewiki.net. In Meta these are usually simple and short tasks with pretty clear context. In translatewiki.net, however, the tasks tend to be larger and more complicated and it often happens that we see only a short string out of context and we aren't sure how to translate it. Sometimes we try to read the documentation or the source code or ask a developer and sometimes we just guess.
It also happens that the translated strings include complicated GRAMMAR, GENDER or PLURAL clauses, the results of which cannot be easily seen in translatewiki.net (but correct me if i'm wrong). FlaggedRevs is an example of a particularly hard-to-translate extension, but it happens in many other features.
Because of these reasons the localization should be properly tested before being deployed to the site. But to test it, the translator needs to be familiar with all of the features and the use cases of the software feature that he's translating.
So, is anybody here familiar with proper strategies for testing the localization of software?
Thanks in advance for any ideas and for sharing your knowledge and experience.
-- Amir Elisha Aharoni · אָמִיר אֱלִישָׁע אַהֲרוֹנִי http://aharoni.wordpress.com "We're living in pieces, I want to live in peace." - T. Moore
Yeah, I totally agree your idea. We should have an all-included wiki to test interfaces (both administrative and editorial parts). Of course, it is a bit complicated.
On Sun, Jan 9, 2011 at 12:19 AM, Amir E. Aharoni amir.aharoni@mail.huji.ac.il wrote:
This is not a request for translation, but a request for comment from experienced localizers.
Is there anybody here who is experienced with testing software localization or writing test suites and test plans for testing software localization?
For Wikimedia projects we mostly translate things in Meta or in translatewiki.net. In Meta these are usually simple and short tasks with pretty clear context. In translatewiki.net, however, the tasks tend to be larger and more complicated and it often happens that we see only a short string out of context and we aren't sure how to translate it. Sometimes we try to read the documentation or the source code or ask a developer and sometimes we just guess.
It also happens that the translated strings include complicated GRAMMAR, GENDER or PLURAL clauses, the results of which cannot be easily seen in translatewiki.net (but correct me if i'm wrong). FlaggedRevs is an example of a particularly hard-to-translate extension, but it happens in many other features.
Because of these reasons the localization should be properly tested before being deployed to the site. But to test it, the translator needs to be familiar with all of the features and the use cases of the software feature that he's translating.
So, is anybody here familiar with proper strategies for testing the localization of software?
Thanks in advance for any ideas and for sharing your knowledge and experience.
-- Amir Elisha Aharoni · אָמִיר אֱלִישָׁע אַהֲרוֹנִי http://aharoni.wordpress.com "We're living in pieces, I want to live in peace." - T. Moore
Translators-l mailing list Translators-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/translators-l
2011/1/9 Paul Selitskas p.selitskas@gmail.com:
Yeah, I totally agree your idea. We should have an all-included wiki to test interfaces (both administrative and editorial parts). Of course, it is a bit complicated.
I am talking not so much about the need for testing - it is quite clear that everything should be tested.
It's more about How to test: What are the testing strategies, how to plan the testing process, how to write test suites, how to update the test plan along with the software, how to check that the test suite covers all the aspects of the software etc.
A lot of it just seems like obvious common sense, but i guess that in the software industry there are also methodologies, which are written down in books and taught in courses. These methodologies aren't necessarily good - if there's anyone here who is experienced with a *bad* localization testing methodology, it will be very useful to hear about it, too! :)
translators-l@lists.wikimedia.org