Localization under a government umbrella [LWN.net]

December 17, 2008

This article was contributed by Ivan Jelic

In an era of wider governmental adoption of free software, the Serbian authorities decided to take a different approach toward the affirmation of GNU/Linux and free software in the business sector and the general public. Instead of direct adoption of free software and open standards, Serbian authorities decided to fund several localization projects with the goal of helping to improve the competitiveness of free software on the Serbian IT market.

The first information about the government's plans to help the localization of Free Software appeared in December 2007, when several of the Serbian media reported about the issue. Shortly after the news was revealed, the official press release (Google cached page, since the site was changed with no resources in English at the moment) from the Serbian Ministry of Telecommunications and Information Society was published, giving all the details that were available to the public at the moment.

In short, February was set as a deadline for the first results, which meant localized versions of Ubuntu, Fedora, Mozilla Firefox, Thunderbird and OpenOffice.org. The projects were funded by the ministry and delegated to the several Serbian computer science faculties for organization and implementation. All of them, except the Ubuntu localization team, showed their first results in March at a presentation organized by the ministry. Ubuntu was late since the localized version was planned for the LTS (Long term support) release which came out in April. Shortly after Ubuntu 8.04 was released localized Ubuntu ISOs appeared on project servers.

Ubuntu was known as a distribution which didn't have a localized installer or characteristic Ubuntu software translated in Serbian. In order to provide better localization, people from Faculty of Electrical engineering in Belgrade forked Ubuntu and named the new distribution cp6Linux. Cp6Linux was recognized as symbolic way to write "SerbLinux" since cp6 can be understood as "Serb" in something that might be considered as Cyrillic "leatspeak". The development team never confirmed this though. "Linux for human beings who speak (only) Serbian" is packaged in three flavors: Home, School and Business. Beside this way of packaging, the cp6 development team customized visual identity and adopted a user interface to make it more friendly for users coming from Windows.

The most important task and the purpose of cp6's existence is not entirely completed, but the situation compared to a vanilla Ubuntu installation is a lot better. The live disk bootstrap interface and the live system installer are translated into Serbian. System tools and package managers are also localized, but translations of package descriptions and configuration messages are missing. The graphical configuration tools shipped with Ubuntu, like restricted-manager, are translated too, so it seems that cp6 2008 (which is the first and so far the only version) is basically targeting localization of the GUI applications and tools. The cp6 team produced a 52-page Creative Commons licensed User manual (CC-NC-SA), covering the most important features in using and installing cp6Linux 2008.

The Fedora localization team (Google translation) took different strategy and decided to produce localized flavors of Fedora, with no forks and branding. The Serbian Fedora localization community was quite well organized and productive before, so the first thing that people for Faculty of Organization Sciences in Belgrade did was getting in touch with translators who already worked on Fedora. According to them, 19416 of 32480 strings in total were localized already, and they've localized 98% of 19500 unlocalized strings, which leads us to the total score of 99% localized strings.

Almost 100% of localization strings in real life mean localized configuration tools, package management GUIs and installation interface. YUM and package descriptions, similar to cp6Linux, remain untranslated. Most of the work was done on Fedora 8, which is available for download from project servers, with Serbian localization and settings out of the box. There is no information about ISOs or localization details for Fedora 9 or 10 on the project website.

Mozilla products were localized by the people from Electronic Faculty in Niš. As in the case of Fedora, project organizations continued existing efforts. The final result, for GNU/Linux and Windows, are Cyrillic and Latin versions available for download from the project website (Firefox 2.0.0.12 and Thunderbird 2.0.0.9).

Back in Belgrade, localization of OpenOffice.org was delegated to The Faculty of Mathematics. Again, the project continued existing efforts and took over the coordination of the official Serbian translation team. The first steps toward a localized OpenOffice.org dated back to 2001 when a group of Serbian free software users got together for a big translation marathon organized by ICT Tower, a local OSS oriented software company. Sadly, without any external support, they failed to keep interest in the project and translations were never updated. The second big push was in the summer of 2005 when Novell gave some money to the "prevod.org" group for improving Serbian localization in SUSE. Following the OpenOffice release 2 "prevod.org" members returned to keeping up with GNOME translations, and once again the OpenOffice.org translation was left unmaintained.

"In December 2008 the Ministry of telecommunications and information society Republic of Serbia started four projects for free software localization." explains Goran Rakic, Serbian OpenOffice.org native language project lead. According to Rakic, the biggest achievements of the project are localized releases of 2.4, 2.4.1 and 3.0 with continuity. "We did large QA and localization quality is better then ever", he states. Project statistics show distribution of more than 30,000 localized installations via the project site and more than 3000 in just one week after the 3.0 release. Rakic reveals that localized OOo is used inside government too, with some large deployments and many more to go. Rakic looks into the future saying that the "Ministry and Faculty of Mathematics in Belgrade signed contract for three years with option to extend and we are just one year in it. I can say that future looks bright for all current and new OpenOffice.org users in Serbia."

It is very hard to give a general conclusion about the implementation and impact of these projects. First of all, the public was never informed of any study related to the use of localized versions of any software in Serbia. So it's impossible to predict how many users might directly benefit from those activities. The only numbers that we can use for any sort of analysis are download statistics, which doesn't necessarily reflect the real amount of acceptance or everyday use of localized programs and distributions.

Contributions and translations from the Faculty of Organization Sciences have gone upstream, and cooperation with the Fedora translation team seems to be established and functioning according to the information on the Serbian team page. On the contrary, it seems that the Cp6Linux translations didn't go upstream, since there are no noted contributions on Launchpad. As in the case of Fedora, communication and cooperation is managed on the Serbian Mozilla localization team wiki. OpenOffice is the only project that actually took over coordination of the localization team, at least officially. Speaking of distributions, in both cases GNOME is being used as the default desktop environment, which has a strong and devoted localization community whose work was packaged in cp6Linux and Fedora in Serbian. GNOME translation is not a part of government funded activities, though.

In the meantime, the Faculty of Technical Sciences from Novi Sad started to work on Alfresco localization, and the results are available on the Alfresco Forge page.

This non-typical approach to free software from the government was motivated by the expectation that localization will become another recommendation for the Free Software adoption in Serbia. According to Mr. Nebojša Vasiljevic, assistant of the Minister of Telecommunications and Information Society for Information society, in his interview for GNUzilla magazine (issue 36). He also said that those project are not part of any strategy involving switching to free software in governmental institutions.

(Log in to post comments)

Balkanization of translation efforts

Posted Dec 18, 2008 15:08 UTC (Thu) by nim-nim (subscriber, #34454) [Link]

The younger generations have been taught after the war to hate the other sides, thus hoping for students in academia to work with each other is illusory.

Balkanization of translation efforts

Posted Dec 18, 2008 15:58 UTC (Thu) by epa (subscriber, #39769) [Link]

It's the same mess on Wikipedia, which has Croatian, Bosnian, Serbian, and Serbo-Croatian. I don't get it.

Balkanization of translation efforts

Posted Dec 18, 2008 21:16 UTC (Thu) by robilad (guest, #27163) [Link]

I'd expect people familiar with Linux distributions to get it much better then those not familiar with them. ;)

Balkanization of translation efforts

Posted Dec 18, 2008 16:15 UTC (Thu) by grakic (guest, #55670) [Link]

Differences are not that subtle as it may seem.

Automatic transliterating from Latin to Cyrillic is not possible for several reasons (you will have to keep English text like command names or file paths untranliterated, and there are complex differences in spelling involving digraphs). If you translate in Cyrillic, than you can transliterate to Latin but it's hard to find young people in Croatia who know how to write in Cyrillic.

Even if we can solve that, rich morphology of Serbo-Croatian make things much harder then in US-GB-AU case. For example, word gender of some terminology localization is different, and changing gender requires changing form of almost every other word in a sentence.

Some grammar verb forms are different, and not to forget, Croatian is using Ijekavian and Serbian is Ekavian and there is no rules you can use to switch from one into another. (You can write dictionaries and exception lists, but that is on a totally different scale)

We are cooperating on personal and meta level, sharing tools, experience and resources like documentation and other for several years now.

Balkanization of translation efforts

Posted Jan 13, 2009 16:48 UTC (Tue) by forthy (guest, #1525) [Link]

it's hard to find young people in Croatia who know how to write in Cyrillic

I wonder why. Cyrillic is not so difficult to learn, after all. I can read Cyrillic and Greek, even though I don't understand the languages, takes a few hours to learn them all. I can understand why young people in Vietnam can't read old Vietnamese texts written with Chinese characters, but for someone speaking an Serbo-Croatian dialect, having access to all texts in all those dialects is much easier (~30 characters vs. >5000 characters).

But maybe it's just me. My mother once set me a bounty to learn Fraktur, to have access to the older books in our family library, and it took me an afternoon to get halfway fluent in reading it (and get the bounty). The difference between a Latin script and Fraktur is certainly less than between Latin and Cyrillic, but the experience broke a dam, and I understood that learning other alphabets is fairly easy. I added Greek and Cyrillic letters to my alphabets, and then stopped due to lack of further sources (that was all way before the Internet).

Balkanization of translation efforts

Posted Dec 18, 2008 21:33 UTC (Thu) by jspaleta (subscriber, #50639) [Link]

Being of Croatian decent, I am so very pleased to actually see the term Balkanization actually being used in the context of the Balkans again instead of of general term for fragmentation.

I applaud your effort to help re-capture the term as something inherent to the Balkan peninsula. The world uses the term Balkanization far to glibly to refer to any destructive process which fragments a group into hostile combatants over some perceived resource or idea. That must stop. It's time that the Balkan people and their descendents take back the word from the rest of the world.

We must stand together and fight for our shared cultural heritage as a group of people who cannot stand together to fight for our shared cultural heritage.

-jef

Balkanization of translation efforts

Posted Dec 18, 2008 22:32 UTC (Thu) by nix (subscriber, #2304) [Link]

And this is why jspaleta should not be banned.

That last line had me laughing harder than anything else I've read this
week...