I am Balbino Tchoutzine, a 4th-year Computer Engineering student at ENSPY (École Nationale Supérieure Polytechnique de Yaoundé), Deep Learning enthusiast. If you want to know more about me and my journey in AI, you can read more about it here. But today, I want to talk about something else: my mother tongue, Medumba.
What is Medumba?
Medumba is the language of the Ndé Department, in the West Region of Cameroon. It is spoken in Bangangté, the department's main city, as well as in a dozen surrounding villages: Bakong, Bangoulap, Bahouoc, Bagnoun, Tonga and many others. There are approximately 210,000 speakers.
It is my family's language. The one I heard every day at home, spoken by my parents.
Yet I don't really speak it myself.
School takes up all the space
French first, then English. That's the reality for many African students: mother tongues stay at home while academic languages occupy all intellectual space. I don't experience it as a heavy regret, it's simply the reality I grew up in.
But over time, a desire took root. Not necessarily the desire to speak Medumba fluently overnight, but something bigger: the desire to preserve it. For future generations. And for myself too.
The turning point: NLP for African Languages
When I started getting seriously interested in AI, I discovered a trend gaining momentum across the continent: using NLP (Natural Language Processing) to give digital existence to languages that had been invisible until now.
Initiatives like Masakhane, a pioneer in sub-Saharan Africa. Or GalsenAI, the Senegalese community doing remarkable work for Wolof (translator, speech synthesis, language models). Students, researchers, and passionate individuals proving it's possible, and actually doing it.
And closer to home, at ENSPY itself, seniors from Polytech have done the same with Yemba, a language from the Menoua department (Dschang, West Cameroon), and others like Lambda Africa. That one really hit home for me.
I told myself: Medumba deserves this too. And I can do it.
What I want to build
My concrete goal is an automatic translator for Medumba, integrable into platforms or chatbots accessible to the general public.
Not necessarily groundbreaking on paper. But a useful, concrete, accessible tool, something that puts the language forward and solves a real problem.
The real challenge: data?
Have I already started? Honestly, not really yet. And the main obstacle isn't the technical side. Skills can be acquired, resources are plentiful and mostly free.
I'm currently following the Google DeepMind AI Research Foundations program, an excellent immersion into large language models and their NLP applications, mainly focused on Africa. I highly recommend it if you're a computer science student.
The real challenge is digital data. Written resources in Medumba are limited and hard to find online. But the data exists. It's alive:
- In the mouths of 210,000 speakers
- In families (starting with my own)
- In villages, everyday conversations, ceremonies
Collecting and transcribing it takes time and resources. But it's entirely doable.
My message
If you're a student and your mother tongue has no or few digital tools: JUST START.
Not to solve a grand problem from day one. Just to build something useful, something that puts your language forward. And if your language already has some resources, go further, make it accessible, build tools that bring it to life.
AI today offers real, concrete means to do this. Let's take advantage of it.
Done is better than perfect.
This project is just beginning. So is this blog. But both are moving forward, and I'm taking you along for the ride.