Monday, August 18, 2008

Language localization

Gerard Meijssen is right. And it doesn't take much time contributing in a language other than English to see how right he is. Normally I've been shy about editing in my second language. Occasionally I'll translate material into English, but I don't get much chance to practice German in California. Lately I've gone over to the German Wikipedia a bit more, reading and a little contributing, and it's time to pound on that drum Gerard has been beating.

Have a look at the hosting page for the green plus sign, whose intended purpose is to make voting less language-specific. Its purpose is intuitive to most people, and had darn well better be intuitive because both the description for the symbol and the explanation of why it's edit protected are available only in English.

It doesn't take long to find a lot of little templates in use over at German that aren't German. And if you want to create a new account, even the security text to prove you're a human being is English only. That's great for an Amerikanerin like myself, but suppose I came from Dresden and hadn't been educated in English as a second language.

What's especially disturbing is that German is the second largest language edition of Wikipedia. It has nearly 800,000 articles, yet these language barriers are still so easy to encounter. That tells me Gerard is probably right on target when he says it's a significant barrier to growth at smaller startup Wikipedia editions.

It's hard enough to explain to a native English non-Wikipedian what a wiki is, and that it's really all right to click that "edit" button. Now imagine the challenge of persuading that friend to edit for the first time if that button said "edit" in Swahili, and if half the commands and all of the templates were in Swahili. Well the actual Swahili Wikipedia has that problem in reverse. As of this writing it's got 7383 articles, 9 administrators, and 1110 registered accounts. And it's the lingua franca for most of East Africa. I'm not certain how localized that Wikipedia really is, but it's a darn safe bet Swahili is less localized than German. If we want viable Wikipedia projects to really serve the world, we need to start by providing menus that the world can understand.

1 comment:

Joshua said...

Wow, I had no idea the problem was this severe that it even shows up on .de The only language I could reasonably help out with in this regard is the Hebrew wiki but I've never really seen any serious issues there. It looks like they've handled localization pretty well. Browsing there I've never noticed any serious issues but then again I'm not looking for them. If I see something in English I probably just read it without thinking.

Has anyone considered ever trying to get outside experts to voluntee their time for this? For some of these examples like Swahili it shouldn't be that hard to find people to help out.

Note also that for many of the smaller wikis there are other reasons that they have stayed small. In the case of Swahili many of the native speakers do not have great internet access. This is less true for Swahili than for other African languages but it is certainly another issue.

There's another reason we should consider localization to be an important thing: Wikipedia and its sister projects have a lot of infleunce. There are serious problems with languages dying out especially as English and a handful of other languages become more common. Failing to properly localize reinforces the trends driving those smaller languages to extinction.