I'm not sure if this belongs here or into the mother meta as a feature request, if the latter is the case, please migrate.

There are a few proposed sites that are StackOverflow in a language other than English.

I don't think that makes sense, because:

  • SO is multilingual already, to a certain extent. There are many programming languages and not every user knows every language. That's totally fine. We add tags to questions to differentiate. Creating a Java.stackexchange.com or c.stackexchange.com would not be the right thing to do.
  • It only helps big languages. A language that is only spoken by a few people will not get its own SO. Just like some rarely used programming languages would never get their own site. But with tags, even questions about esoteric programming languages can be asked and answered. Seriously, look at that list. I can barely constrain myself from adding Quenya or Klingon linking to Rick Astley.
  • It creates duplicates that are harder to manage. Duplicates are a reason to close questions, but apparently also a reason to create entire new sites. True, the duplicate questions are asked in a different language, but content wise, they are still duplicates. Instead of adding more sites, add more multilingual features to the original SO. Flagging a question as duplicate in different language will add it to the translation review queue where members that know both languages can translate the accepted answer. Reputation is shared between translator and original answerer.

Instead of proposing new sites, let's put our energy into making SO (or any SE site for that matter) multilingual on a meta level (that is, not just with tags, but the ability for users to set their languages, translation review queues, tag descriptions, site text, help text...)

To address some things from answers:

Comparing programming languages with human languages is... mind bending. Those who speak wildly different languages than English like Japanese or Russian will tell better than me for sure.

I compared programming languages with human languages in terms of whether it is possible to distinguish the language(s) used in a question with a tag. This is not mind bending in any way and has nothing to do with the differences between languages at all. Just like wildly different programming languages can be distinguished via tags, wildly different human languages can be distinguished just the same way.

The point about small languages is a perfect solution fallacy.

The barrier for creating a new language specific site is way higher than creating a tag. That's a fact, not a fallacy. A human language with only a few speakers could easily form a subcommunity, but never get enough questions/views per day/users to justify their own language specific SO.

Two equal questions in different languages cannot be duplicates, at least in 2015, when machine translation is still... fragile.

Two equal questions are equal and thus duplicates. You can give the same erroneous code to different people and will receive swearwords in different languages, but the underlying programming problem.

Suggesting that machine translation is fragile totally ignores the suggestion I made above. I didn't suggest any machine translation at all.

This should be discussed on Meta Stack Exchange. – JohnB May 8 '15 at 14:21
That would be very complicated. – mmking May 8 '15 at 14:30
Sorry about the delay, I'll have a response to this shortly. – Tim Post May 11 '15 at 15:11
@TimPost sounds good. Should I re-ask this in mother meta, should it be migrated as is or should it just stay here? – p4ssword May 11 '15 at 17:55
Here is fine, it's just as on-topic for folks that see these proposals on Area 51 as it is anywhere else. Sorry for getting a bit winded, it's a topic with a bit of depth :) – Tim Post May 12 '15 at 6:39
Portuguese StackOverflow just graduated out of beta and became a fully grown site. We also have the Russian and the Japanese in beta. What do you think that we should do about that? – Victor Stafusa May 21 '15 at 2:30
an interesting question +1 – Jad May 28 '15 at 18:32
I don't agree with what you are proposing, but I think it is a really valid, interesting and "to think" question. Even if it is already answered. So I will upvote it ;) – Rafael Jun 13 '15 at 2:24
Downvoted, see my answer. – astrojuanlu Jul 8 '15 at 9:27
I think the 2 alternatives (language-specific sites and multilingual SO) are not necessarily substitutive, i.e. we should not have to choose one. Given how much there is to accomplish before we have a fully effective automated translation system, I think language-specific sites are the best we can do. In the meanwhile I would like to see that day when automatic translation fully works and language-specific sites are no longer necessary. edit : I realized that I'm not sure what are the multilingual features of SO (if you mean something other than automated translation) – jeff Jul 30 '15 at 2:31
Sorry for this, but we are waiting for Turkish.SO for 3+ months, they did not release that (multilang update is what we're waiting for :/), they apparently use Valve Time, so releasing a this big update would be... "soon" in Valve Time. (TF2, 1998 -> Soon, 2007 -> Release). – Avery Aug 22 '15 at 22:43
@ardaozkal lol, bookmarked – p4ssword Aug 22 '15 at 22:48
This SO in spanish just puts me sad :(, y soy de habla hispana. – Jaime Hablutzel Aug 23 '15 at 23:56
up vote 24 down vote accepted

Traditional localization typically entails offering one site, where the 'chrome' (or user interface elements, help text, prompts, etc) is available in multiple languages. As I looked into how other sites were offering support in multiple languages, this was the predominate way to do it. Many still have the 'countrycode.domain.com' available for search engines to differentiate results, but it all directs to a single place.

That strategy won't work for us, and I don't think we're done realizing all of the reasons why it wouldn't. The first issue is one of community - these aren't just sites in different languages, each localized Stack Overflow has its own culture.

Our Japanese site is extremely welcoming, they don't like to turn away questions that they could conceivably answer. In fact, you could go ask questions in English about a Japanese product that developers use and you'd get an answer there. The quite rigid culture of the English Stack Overflow site would drive quite a few folks away. Don't get me wrong, they don't allow mindless fun - but they want to at least try the stuff that Stack Overflow normally discourages before writing it off completely.

The other blockers are duplicates, as you mentioned - search engines return results in the language that you used to search. The easiest way to make that possible is to do exactly what Wikipedia does, a different sub domain per language. As we work to more effectively surface more information in our sites, this would entail efforts to hide quite a bit of it depending on the language the visitor uses. Any diagram of how that might work quickly starts to look like spaghetti.

There are also features we have to consider per site to deal with semantics of the language. Japanese has no spaces, for instance. Russian has interesting plurals.

The way we're doing it now creates much less clutter and work for everyone involved, standing up a new site doesn't take a lot of work. Making them all work on a single site would be .. next to impossible. These [language] users keep casting close votes on questions in [other language]! Making privileges tied to tags like that is a nightmare in and of itself.

That's not to say that the way we're currently doing it is ideal, in fact, we're far from ideal. There's so much manual lever-pulling and button-pushing, and we still haven't found a way around needing to hire a CM for each language supported. We don't know if we'll ever be able to make the process itself as autonomous as we'd like.

But, going forward - having each one of these contained in their own 'sandbox' so-to-say is our best bet for maintaining the progress we've made.

Not a bad question, I think folks down-voting here didn't realize that it is rather interesting to talk about.

Thanks. From your answer I take away that technically this would be a "nice to have", but implementing it would be too much work. In addition to that, some sites prove to be their own independent communities rather than subcommunities of one big website. – p4ssword May 12 '15 at 11:42
@null Yes, that's a good summary. – Tim Post May 12 '15 at 15:00
@TimPost, Just like you said "these aren't just sites in different languages, each localized Stack Overflow has its own culture". You're absolutely right about that. :) – Eko Junaidi Salam May 16 '15 at 6:41
adding a per language site is at least better than nothing... we just need to fix the major problem of decentralized informations, may be by implementing a feature to link similar per language topic (let say a little frame under the question for similar answered questions in other languages) – intika Jun 25 '15 at 17:27
@TimPost according to this meta answer in a bilingual SE site people are apparently already improvising the functionality I suggested. I guess doing this with flags would be much more robust. – p4ssword Jul 8 '15 at 20:19
As Tim said, using subdomains its a good idea. They can work as "sub-sites" . For example the profiles are linked to those sub-sites so no extra registration, etc is required. The only problem I see is that someone with high privileges in SO, could also have high privileges in sub-sites (not a big deal I think). Many people here can speak more than one language, so its more natural that way, IMHO. – lepe Aug 24 '15 at 1:32
While we are making analogies to Wikipedia, wouldn't similar cross language linking be useful as well for cross-language "duplicates"? – Hjulle Oct 20 '15 at 13:57
Consider countrycode.example.com, since domain.com is an active commercial domain. – user2943160 Sep 5 '16 at 3:17

Look at all these duplicates in English and in this same site:

and I could go on forever.

Directly addressing your points:

  • Comparing programming languages with human languages is... mind bending. Those who speak wildly different languages than English like Japanese or Russian will tell better than me for sure.
  • The point about small languages is a perfect solution fallacy.
  • Two equal questions in different languages cannot be duplicates, at least in 2015, when machine translation is still... fragile.
I edited my question. I don't think you are really addressing my points. You are arguing against general points that I didn't make in that form. It appears to me that you just skimmed the question without looking at the details. If you think this is a duplicate, flag it as such. – p4ssword Jul 8 '15 at 12:10

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .