A recent article in a newspaper has Chris Hofmann stating something along the lines of that the absence of any Indic language other than Gujarati and Punjabi from Firefox3 was due to the lack of “motivated” volunteers. Runa has a blog entry responding (in part) to that.
I met Chris during foss.in last year and have been interacting sporadically on mail since then on. He comes across as a person who “gets it” where “it” being the notion of collaborating within a process towards producing open source software. However, this time he did make a broad sweep at existing Indic communities and comes across as extremely rude.
The first point that would be important to note is that across various Free and Open Source Software Projects, the Indic language communities are fairly robust. So, when it comes to Firefox, the stats provided by Gervase are slightly odd. Attempting to reach out to the largest segment of language consumption communities makes sense for a web browser project since it is an artifact of immediate and visible consumption. The discussion around the stats can be found here, here and here. There is a side note from Axel who states that [he is] “generally skeptical on recruiting localization teams. To me, users benefit from a growing community, and they get hardly anything from a one-time effort”
The second point is that a large number of projects tend to fall into the box of being too much interested in localisation and not much in anything else. With the number of constitutionally accepted languages in India, it does make for a great case for herding a diverse set of cats and getting UI bits in the local language. The paradox here is that L10n is just not string crunching. A large segment of the languages have no existing equivalent terms for the technical jargon. Thus, the volunteers-translators-language_maintainers have to be careful in choosing words that convey the essence or, in some cases coming up with words that make sense. This work also includes paying close attention to branding and trademark issues (most of the FOSS trademarks are non existent in the local language) as well as ensuring a consistency of translations across teams working on various modules. If it were mere crunching of strings, theoretically it could be done by machine translations or a translation memory system too.
L10n is thus not merely translation but generation of content. And it is this content that determines how successful a FOSS project / product would be for the language communities. However, L10n is a finite state of affairs. Teams can work with focus and incrementally attain 100% localization after which it turns into a sustenance phase of translation. The projects have to figure out a way of “what next” and start putting stakes in the ground towards doing that. This is the exact place where I don’t see Mozilla or OpenOffice.org put in much work in India. Translation is a beginning towards getting (what_is_called) traction and then one has to build onto that into bigger things. A common thread of complaint that was around from foss.in last year was that the Project Days had a “condescending” approach during the presentations. A part of this can be traced to the lack of will and / or enthusiasm in getting beyond L10n into larger areas of contribution into the core of the projects.
It isn’t always the motiviation of the volunteers that cause languages to slip. Most of the time it is how the project treats its contributors. There’s a lesson in there somewhere.