The years I spent as a graduate student in Madison were some of the most enjoyable in my life, mainly because I had lots of free time and very few responsibilities. I ended up spending almost as much time, if not more, on unofficial hobbies as I did on my Ph.D. Most of my free time was probably spent on lattice, but I also spent a lot of time on various initiatives related to technologies enabling the use of Bengali on computers. One of the first internet communities I was part of was an ad hoc group of fellow Bangla computing enthusiasts under who banded as a collective named Ankur (originally called bengalinux).
I don’t quite remember exactly how I got involved with Ankur, but the impetus was an idea I had had for a while, which was to create an online archive of public domain Bengali literature, similar to Project Gutenberg. In theory, this was just a matter of typing up whatever I had the time and inclination for, but in practice, there were many technical challenges.
The time was the early 2000s, and the internet had not yet started using Unicode, although it was clear to those technically inclined that it was going to be the future. ISCII had never really caught on. Popular bengali websites, such as anandabazar.com, were using proprietary solutions, often tied to particular browsers (making snapshots of their websites taken at the time completely useless now). A company whose name I have forgotten had come out with CD containing all the works of Rabindranath Tagore, but it also came with an ad hoc encoding.
I had by that time already been influenced by the ideas of Richard Stallman, and had no intention of using proprietary technologies. One reasonable solution was Bangtex, which was a LaTeX extension that could be used to typeset Bengali. It was open source, but could only produce PDF files, not HTML, and more importantly, it also used an ad hoc encoding that was specific to the font being used, making it not a worthwhile investment if one wanted to be future-proof.
It was quite clear even at the time that Unicode was the “right” solution in terms of encoding Indic languages. The idea of Unicode itself was not novel (the Indic language encodings were based on the earlier ISCII encoding), but now Microsoft was heavily invested in it (presumably as part of their plans to expand into non-European territories), which meant that it was very likely to be widely adopted. What was still missing was a functional implementation of all the pieces that would make the technology work seamlessly on all platforms.
The earliest version of my Bengali archive that I could find on the Wayback Machine has a good summary of the issues, which have all been resolved since then, with some contributions from the Ankur group along the way. The most challenging problem was the lack of a useable (let alone good) font. There was a font called Code2000 which covered almost all Unicode characters including Bengali, but did not have much in terms of OpenType layout support (for conjucts and other visual features). Microsoft was distributing a Devanagari font called Mangal, but their eventual Bengali font Vrinda was not yet available. So, even though I had no experience with typography or any other kind of design for that matter, I decided to try my hand at creating a font myself, which turned out to be not too difficult. The name of the font was “Likhan”; the font shapes were created using a nice open source software called Pfaedit (since renamed as FontForge), with Opentype layout tables added using Microsoft’s Visual Opentype Layout Tool (VOLT).
Another practical issue at the time (and even now, to some extent) was the problem of typing in Bengali. I wrote a couple of web-based tools to convert phonetic roman input into Unicode Bengali, versions of which are still available here. This was my first non-trivial project with Javascript, so it was fun.
Not many people visited my literature archive, and even fewer people contributed, so it has basically languished, although I have always kept some version of it available. Fortunately, the wiki movement has really taken off since then, and wikisource has filled the gap that I had tried to fill, in a much more systematic and collaborative way. I still try to experiment with my site when I can (especially now that Google has essentially perfected OCR), and during the COVID-19 lockdown in 2020, I re-designed it as an exercise to learn about Github Pages.
The font I had designed, and to a lesser extent the Bengali input tool I had created, turned out to have more substantial, if short-lived, influence on the wider world. It put me in touch with Ankur, and sundry people from around the world whom I would not have known otherwise.
As I said, I don’t exactly remember how I got in touch with
bengalinux, but it definitely happened before December 2002. The
bengalinux-core mailing list archives start in September 2002;
Likhan is mentioned by Sayamindu in October, and
I seem to have become a member by November. I think what happened was
that Sayamindu came across Likhan and asked me to join both the Free
Bangla Font project and
bengalinux, and this happened sometime in October.
Fonts were a big focus area in those early days. The rest (implementation in Free Software systems, or even Windows for that matter) was not really in our hands and also beyond our skill set, although we did what we could in terms of testing and reporting bugs, especially in Pango and Qt, and influencing the development of the Unicode standards pertaining to Indic languages. Likhan, although not a good font by any stretch of the imagination, nonetheless became somewhat popular as our other offerings were equally bad. Eventually, more professional-looking Bengali Unicode fonts started being released. I suspect many of them were created by copying the glyphs from other non-Unicode fonts without permission and pasting them on a template Unicode font, but the copyright owners did not bother to take legal action even if they were aware, and as a result the dearth of Bengali Unicode fonts subsided. My second attempt at making Bengali fonts led to a font named Jamrul, which was released in 2004. It was a much better font than Likhan in my opinion, but nobody else was interested in it by then. I kept using it myself for many years, before giving it up in favour of the Google Noto fonts.
Ankur, as bengalinux was eventually rebranded as, was home to some interesting software projects as well (such as an English to Bengali translator based on a part-of-speech tagger), but eventually the focus turned to internationalizaton and localization, or in other words, translating software interfaces to Bengali. This was never really a passion of mine. I had little interest in the outcome unless the results were something I would use, and the initial focus was on translating the GNOME environment, which made sense because it was the de facto choice for new users, but it was also something I rarely used because I had by that time settled on KDE as my environment of choice (it still is). Secondly, as fun as it was to see menu items in Bengali, it didn’t really help me in any way because I was perfectly comfortable with the original English. Without any additional incentive, it was difficult to sustain interest in something I would not use myself.
Still, I got into it for a while, translating KDE because that way I would at least see the results of my work. I did put in a lot of effort, with help from a few others, but eventually stopped when I finished my postdoc and moved on to other things. Somewhat sadly, no one else seems to have has taken up the task of translating KDE to Bengali in the more than ten years since then.
Eventually, the online community of Ankur fizzled out. Partly this was because people moved on with their lives, and partly because we had accomplished what we had started out to do, which was to help enable Bengali on the web and on Free Software platforms; today, working with Bengali is as easy on GNU/Linux systems, if not easier, than it is on Windows or Mac OS. What remained was routine and boring work, perhaps important, but not as attractive to the hobbyists who started the group.
My only regret is that I never physically met anyone in the group other than Sayamindu, and never got to know them closely enough to know what motivated them to get involved in the first place.