Firefox 3 Language Coverage

Note: this isn’t the full picture; some people have been very helpful in pointing out stuff I’ve missed. Please continue to send in feedback, and I’ll post an updated version in a couple of days.

I have been attempting to figure out what percentage of the net population will get the upcoming Firefox 3 (June 17th!) in their native language (“heart language”). We’re doing 48 different localizations. I’ve attempted this before, but I have been limited by the quality of data available. No-one seems to have good statistics on the language breakdown of the net population.

So, I’ve taken net population figures for each country (232 of them; that number seems high, but I guess they have a generous definition of ‘country’) from the CIA World Factbook, and split them up by the language split in that country. This assumes, therefore, that the distribution of the net population in a country is in the same proportion to the languages spoken there. I’m sure for some countries that’s a bad assumption, but I still think the resulting data is better than what I had before.

Here’s a snapshot of the spreadsheet (.ods). Headline figures:

Category Count % Coverage
Firefox 2 44 87.7%
Firefox 3 45 88.1%
Firefox 2 and 3 together 48 89.2%
L10n projects with CVS access 63 93.7%
All localizations found, including unofficial 61 92.6%
IE 7 77 97.6%

Methodological notes: Many figures estimated. Unknown speakers allocated proportionally among the languages for which there is a figure. I don’t mean to insult anyone – if I’ve overlooked your localization, I apologise. Void where prohibited. Blame Canada.

OK, now the questions:

Why does FF3 say 45 and not 48?
I only have one column in my spreadsheet for each of English, Portuguese and Spanish. IE has two Portugueses, but only one Spanish and one English. If we split them and said that IE didn’t support British English or South American Spanish, we’d gain a large advantage which I don’t think would be really reflective of the truth. But perhaps some Spanish speakers in Latin America want to argue with me there :-)
Why is FF2+FF3 more than either FF2 or FF3?
FF3 has Indonesian, Sinhala, Albanian and Serbian (welcome!), which FF2 doesn’t have. FF2 has Bulgarian, Welsh and Persian, which FF3 doesn’t have (yet).
How come “L10n with CVS access” is more than “Total localizations found”?
Because some teams with CVS access don’t seem to have produced a localization yet, and that more than counterbalances the unofficial localizations I was able to find on addons and by searching the web.
Which language communities do we serve which IE does not?
Hello to native speakers of Belarusian, Frisian, Kurdish and Mongolian. We’re meeting your needs :-)
Why is our percentage lower than IE’s?
There are a couple of relatively big languages (as well as a dozen smaller ones) that we don’t have official localizations for yet: Hindi (2.20%), Vietnamese (1.48%), Thai (0.70%), Malay (0.66%), Bengali (0.47%), Tagalog (0.39%), Marathi (0.37%), Urdu (0.32%), …. They have at least 77 localizations, we only do 48, and because we’re a free software project, we don’t work down the language popularity list from the top :-) If we ever managed to do all the localizations they have, our additional localizations for smaller languages would give us 0.75 percentage points on them.
If Microsoft were reading, which language should they do next?
Balochi (0.51%), an Asian language spoken in Iran, Pakistan and Afghanistan.

10 thoughts on “Firefox 3 Language Coverage

  1. Gerv, they’re not “language packs”. We have full releases in all of those languages. Language packs are something you bolt on to a non-native release of Firefox.

    A lot of work goes in to the integration that makes a full language release and calling them “language packs” will make people think you’re discounting those efforts.

    - A

  2. Asa: it’s just loose language, no more than that. I didn’t mean to discount anyone. To be honest, I thought I’d get flak for combining the Portugueses and the Spanishes :-)

    The question I was trying to answer was “can someone get the Firefox (or IE) UI in their language”, regardless of the mechanism. So I lumped packs, addons, full builds, IE’s MUI Packs and Language Interface packs all together in one, and called them “language packs”. If there’s a better description which covers all these things, tell me and I’ll use it.

    Paul: Works for me…

  3. What does the curve look like in terms of time to market? How long does it take for IE to get to their %age of coverage, vs. FF or other browsers?

  4. I know it doesn’t count, but Firefox did have a Vietnamese language pack, but it was pulled. A sanitized version has been uploaded to Bugzilla, but not approved yet as an extension.

  5. Opera 9.5 has just been released in 30 languages; they are at 84.8%.

    shaver: You mean, how long after release, as opposed to our simultaneous release? According to the IE blogpost, it takes them 2-3 weeks to do Arabic, Finnish, French, German, Japanese, and Spanish, and up to four months for the rest. They don’t give exact dates for each – if I could find files with valid dates on them, I could plot a curve.

    Minh: You mean uploaded to addons.mozilla.org? It’s a shame approval is taking so long. I wonder if that’s a manpower thing. Perhaps now that we’re code complete, some other people will be able to help.

  6. Oops, sorry, 84.25%. I misread their download table. They don’t have the latest version in Afrikaans or Punjabi yet.

  7. It would have been interesting if F3 Download Day’s stats had been available by language, not just by country.