These are my responses to questions I am tired of answering repeatedly.
The best answer I can give so far is our paper on Kyrgyz NLProc, particularly sections 3 and 4.
Not yet.
NLP researchers often work with languages they don’t speak. No one expects those developing models for hundreds of languages, or even working with extinct languages, to be fluent in all of them — so why is this an issue for me? 🙂
That said, I only take criticism on this topic from those who have already contributed to Kyrgyz NLP. Otherwise, it’s just gatekeeping, which is counterproductive given the current state of the field.
Of course, there are limitations. I can’t work on tasks requiring manual evaluation of colloquial usage, syntax nuances, or similar expertise. In such cases, I either avoid those tasks or collaborate with native speakers and language experts.
I do research for the sake of research — mostly for the fun of it, not profit. It’s a rare privilege to work on something simply because you can and because you care. I’m also writing my doctoral thesis at KSTU n.a. I. Razzakov, and that’s the extent of it. Maybe I should care more about political implications, but I don’t. See also my response to the last question.
Actually, I work on Russian NLP as well!
Again, I don’t care about such labels. I research for the sake of research — mostly for fun and personal interest. It’s a privilege to work on something purely because you care. Also, as mentioned, I’m writing my doctoral thesis at KSTU n.a. I. Razzakov. That’s all there is to it. For more context, see the response to the last question.
This is by far the most common question. There are several reasons.
Sentimental. Soviet philologist Roman Timenchik once said that it’s a privilege to work in a field that allows you to express gratitude to entire cities, countries, and peoples. For him, this meant writing a book about Venice in Russian poetry. For me, NLP offers a similar opportunity. I want to create something meaningful in Kyrgyz NLP as a gesture of appreciation for a country and people who have given me so much — both inspiration and important moments in life.
Exploration. I’ve always been interested in working with agglutinative languages. A few years ago, I even compiled an awesome-azeri-nlp list of language resources for the Azerbaijani language. Kyrgyz is a perfect candidate to delve into this area.
Opportunity. Kyrgyz is a low-resource language, which means there’s a lot to do for a computational linguist and NLP practitioner. This is an excellent chance to learn by doing and make meaningful contributions across various subdomains of NLP.
Collaboration. Working on a language you don’t speak naturally requires collaboration. I genuinely enjoy working with the local AI community, contributing where I can, and learning together. It’s a rewarding experience both professionally and personally.
Basically, trying to do something pretty while I can.