[Follow-up set to mozilla.dev.l10n]
[1]<http://support.mozilla.com/>
--
Chris Ilias <http://ilias.ca>
List-owner: support-firefox, support-thunderbird, test-multimedia
Well, aside from the in-product glossary, not especially. We've aimed for word choice consistency, but I don't think we've ever fully documented exactly which words we choose when a choice is possible. The closest thing to it is <http://www.mozilla.org/projects/help-viewer/documentation_language-style>, but that's hardly exhaustive, and while it's not outdated or obsolete, it's a bit aged. Also, back when we were most active on docs, it was essentially "community knowledge" that went unstated more often than not.
Jeff
Some localization teams have glossaries for themselves, see the comments
at http://blog.mozilla.com/axel/2007/10/30/firefox-2-glossary/#comments
I would expect to see synergies in the knowledge about which terms
should be translated consistently, not just in the product, but in help,
too. Sounds like an interesting tool.
For Help in particular, I could imagine that translation memories had
particular strenghts, though. I would expect them to find phrase like
"To change this, open the preferences dialog...", for example. There
might be other. I've never used a TM, but I would expect that technology
to be better suited for full text than for menu strings, for example.
Axel
My two pence
Yes, I think so, too. I'm having a hard time to actually tell the
difference between translation memory and glossaries. I guess that
glossaries are really more word based, whereas from what I read about
translation memories, those like to cover up to whole paragraphs. I
haven't read too many help articles yet, but I can certainly picture
that those are cool for the "about mozilla" PR blurbs at least. Help
might have similar things.
Axel
I'd say that's also more or less what I understand, but not just for
single words. I envision glossaries as a list of terms, sometimes
single words (ie., "Download", "Location"), sometimes a tighted couple
(ie., "Download Manager"), with a list of possible translations bound
to each term, requiring always exact matches.
OTOH, a translation memory should use a predefined set of sentences or
paragraphs with possible translations, and use it in a (hopefully)
fuzzy approach, so when you first find an untranslated sentence that
it is 90% close to another one already translated, you get the
translation for this latter one as a fuzzy match.
Ricardo
--
If it's true that we are here to help others,
then what exactly are the OTHERS here for?