Major Browsers: UI Language Coverage Stats

With the Firefox 4 release almost upon us, Google having quietly shipped Chrome 10 last week, and IE 9 released yesterday, it’s time for another comparison of how the different browser makers support language communities. Here’s the headlines:

Browser Language
Firefox 4 (predicted) 73 93.7%
Firefox 3.6 68 95.1%
Chrome 10 45 91.6%
Opera 11.0 40 86.0%
IE 9 38 89.7%

For those just tuning in, these stats come from an large spreadsheet where I take the internet population of each country, divide it as best I can among that country’s language communities, and thereby try and predict what percentage of the world’s internet users have first language support in the UI of each browser. The standard disclaimer: I have adopted a possibly controversial and idiosyncratic definition of what is a language; please don’t beat me around the head with it, but focus on the big picture.

There’s a load of interesting stuff behind the headlines:

  • There are 37 languages whose speakers are out of luck if they go and see Microsoft, but who we cater for. Every single translation we have is done by volunteers in our community. The power of open source!
  • Firefox 4.0 has more languages than Firefox 3.6 but a lower percentage. That’s because, sadly, we were unable to get Serbian and Vietnamese on board for the Firefox 4.0 final release. Vietnamese accounts for 1.32% of the world’s internet population, and Serbian another 0.25%. These two languages account for the drop, and also mean that our language set is not a strict superset of those provided by IE (and why the figure above is 37, not 35).
  • However, since Firefox 3.6 we have gained Akan, Asturian, Breton, Bosnian, Gaelic, Armenian, Lugandan, Mailithi, Sepedi, Songhay, and Zulu. A strong African contingent there :-)
  • IE gets to a higher percentage than Opera with fewer languages primarily because Opera does not have an Arabic version (worth 3.2%).
  • Between Chrome 8 and 10, they seem to have lost Tagalog, Kazakh and Hebrew, which makes a big difference to their percentage (-2%). It could be that these languages will come along post-release; I don’t have a window into their process.
  • None of the browsers now has a Tagalog version, despite it being (as far as I can see) a very large market. Do Philippines computer users just use the English versions?

Feel free to look at the numbers and post about anything interesting you find :-) Updates to the data which splits a country’s internet population into language groups would be particularly welcome.

14 thoughts on “Major Browsers: UI Language Coverage Stats

  1. Vietnamese will ship as Beta for Firefox 4, afaik.

    In the Philippines, most software is English, so imo we should do an en-PH but Tagalog would be good too. Problem is that Tagalog doesn’t have terms for many of the technical innards of browsers, etc.

  2. Gen: Good to see you online; I hope all is well with you and your family.

    Chrome have done a Tagalog localization in the past so perhaps we could look at that to see what choices they made about terminology. I guess we shouldn’t do it if no-one will use it, but I suspect a good proportion of Filipinos don’t speak any English.

    I asked Axel about Vietnamese; he’s not sure what’s going to happen.

  3. …(and why the figure above is 37, not 35)…

    Shouldn’t this be the other way around?

  4. Stas: No, I don’t think so. “The figure above” is the figure 37 in the first bullet point. 73 – 38 = 35, but IE has two languages that we don’t (Serbian and Vietnamese) so we must have two more that they don’t, making the total 37. I did a count using the spreadsheet and it came to this value too.

  5. Commented in .planning on ‘vi’, fwiw.

    Technical detail, they came in basically the day after RC went to build. Just very unfortunate timing.

  6. Have you got stats for ‘all browsers combined’? Based on what you’ve said above this must be at least 75 languages and 95.1%

    Sorry for being lazy, but I can’t get the spreadsheet to import into google docs (my main office suite).

  7. Ian: the spreadsheet says: 99.33%, 97 languages for all versions tracked (and I delete old versions periodically), 95.55%, 77 languages for all the latest versions. If Firefox had localizations in Georgian, Serbian, Swahili and Vietnamese, it would be a strict superset of the languages offered by every other browser. And we have had three of those four in the past. We should find out who Chrome’s Swahili l10n team are and borrow them :-)

  8. Swahili isn’t easy historically. Nor is Tagalog, we had a project for that, but they couldn’t really agree on the variant to use, and thus lost momentum, I guess. It’s also a largely en-US-using region.

    The African contingent is awesome contribution by the Anloc project, which is led by Dwayne, btw.

  9. > I suspect a good proportion of Filipinos don’t speak any English.

    Yes, but how much overlap is there between that portion and the portion who have access to the internet? (I don’t have the answer to this; I’m just raising the question.)

    My understanding (which admittedly has not been updated very recently) is that the distribution of wealth in the Philippines is kind of like that in India — there are people at many different levels ranging all the way from squatting on a burning mound of garbage right on up to multinational-CEO class, and everything in between. I don’t know how well-represented the non-English-speakers are in levels with enough means to worry about whether they have access to the internet.

    Geography may also be relevant. The distribution of languages is certainly not even across all the islands, and I very much doubt that the distribution of internet access is either, but I don’t know how these distributions relate to one another.

    The other thing is, Tagalog is a non-first language for most of the non-English-speakers. (The role of Tagalog in most of the Philippines is comparable to the role of high Latin in medieval Europe.) However, supporting more than a hundred different Austronesian languages spoken in the Philippines is unlikely to be practical in the near term, and even the dozen or so major regional lingua franca would be a fairly significant undertaking. If you’re going to pick one endemic language to support purely for the sake of Filipinos, Tagalog (or “Filipino”, which is essentially Tagalog in an official capacity) is an obvious choice.

    Cebuano in theory has almost as many speakers as Tagalog (depending on whose figures you look at), probably fewer of whom also speak much English I would guess, but I don’t have any idea how many of them have internet access.

  10. Quite cool. Another bias to take into account in your stats. Internet users != Browser Desktop users. There are countries where users never have access to a desktop computer and exclusively use a mobile phone browser. That would drastically changed the coverage in terms of percentage.

  11. I see that in Internet population by country you list Malay as No 3. I think most of us prefer an English UI even if we are reading content in Chinese, English or Malay. I think Gen is trying to make arrangements with the Ubuntu translation community (which does have Malay translations) to share resources.