True Digital Inclusion: The Power of Local Languages in Technology

[Hassan_Wurie]

When we talk about making digital tools accessible and impactful in Sierra Leone, language is a foundational pillar that cannot be overlooked. At the recent Orange Social Venture Prize event, a comedian shared a powerful example of students at Upline School being taught in Temne instead of English. This resonated deeply with me because it highlights how learning in one’s native language can dramatically improve understanding, retention, and engagement.

While English remains vital as a global lingua franca, it is our native languages—Krio, Fula, Temne, Mende, and others—that carry cultural nuances, lived experiences, and authentic ways of thinking. Historically, major waves of knowledge and innovation have always involved translation—from Greek, Arabic, Persian, to Kemet—shaping how civilizations learn and grow. Now, the question is: what if Sierra Leone leads a new wave by translating and developing digital knowledge in our local languages?

In my work, I have contributed to building open datasets and language models for Fula and Krio, including one of the first Krio Text-to-Speech models. These projects, hosted on Hugging Face at Pullo-Africa-Protagonist (Hassan Wurie Jalloh), aim to democratize access to digital content and tools in languages people speak every day. The goal is to create technology that understands and serves our people, not just imported solutions.

However, there are challenges. Developing technology in local languages requires expertise in linguistics, computing, and cultural context. Data scarcity, lack of standardized orthographies, and limited funding are real barriers. We also need more local talent empowered and encouraged to participate in these projects.

I invite this community to share insights and experiences:

What successful projects have you seen that prioritize local language inclusion in digital tools?

What are the technical and social challenges we must overcome to expand this work?

How can government, academia, industry, and communities collaborate to build sustainable language tech ecosystems?

In what ways can we inspire and equip the next generation of Sierra Leonean innovators to lead this field?

Our native languages are not just a means of communication—they are gateways to knowledge, innovation, and empowerment. By integrating them deeply into our digital public goods and AI projects, we can build a Sierra Leone where technology truly belongs to everyone.

Let’s explore how we can make this vision a reality. Your thoughts?

Hassan Wurie

[Techlinkus]

Thank you for this insightful post, Mr Hassan. I found it powerful and deeply relevant.

One point that really stood out to me is how language shapes access to digital tools. From our experience building the Healthy Mother App, we’ve seen that while translating text into local languages like Krio or Temne is helpful, many users still struggle to read even in their own languages. Most people never formally learned to read or write in Krio or Temne, because those languages weren’t part of traditional schooling.

That’s why I strongly believe voice technologies like text-to-speech and speech-to-text will play a crucial role in true inclusion. For our app’s chatbot feature, which allows pregnant women to ask health questions and get instant responses, we’re exploring how to make it more accessible through voice input and voice output in local languages.

Your work on Krio Text-to-Speech is incredibly relevant here, and I’d be genuinely interested in collaborating with you to explore how we could integrate that into the Healthy Mother App. Imagine a mother simply speaking her question in Krio and hearing back a trusted response, no reading or typing needed. That’s what digital public goods should be about meeting people exactly where they are.

Let’s connect further I’d love to learn more about what you’ve built and how we might work together.

Ibrahim Success Swaray

[8086]

Ah, interesting stuff! Really fascinating.

This ties directly into a core focus of mine. Through our work at EasySTEM, where we develop EdTech solutions, we constantly encounter the challenge of bridging the linguistic divide which exists for tech solutions deployed. In many parts of Sierra Leone, especially in rural areas, you find entire communities where no one speaks Krio, let alone English.

This raises a fundamental question: How can we ensure children can learn in their local language, without English becoming a barrier to knowledge acquisition? Personally, I envision a hybrid learning model, where both the local language and English are used in parallel. We’re actively working towards this, although translation efforts are incredibly expensive. As you rightly mentioned, data is the first and biggest hurdle to overcome.

Just last week, my co-founder, Lovetta, returned from a pre-deployment site visit for our offline servers in several rural villages. She reported that even Krio was not spoken in some of these communities, so you can imagine the situation with English. The schools and community gatherings even teach in Limba et. al!

I also see this linguistic divide impacting how we design service delivery tools. A few years ago, while I was learning the Qur’an, I noticed that my karmokoh’s (religious teacher for context) WhatsApp was filled entirely with voice messages, not a single line of text. He can’t type in English, Krio, or even Fula (his tribe), but he can communicate fluently through voice.

That experience made me reflect more broadly: WhatsApp is just a messaging app, yet it became accessible to my karmokoh simply because it allowed him to communicate through voice. But what about platforms that require more than just sending bidirectional messages? How do we design health apps that enable users to listen instead of read? Or government service delivery apps that allow citizens to interact without needing to be literate in English or Krio?

This raises an important design challenge: How do we build tools that are not only linguistically inclusive but also tailored to how people actually communicate and access information in their daily lives? Voice-based interfaces, localised audio content, and culturally-aware design principles should not be optional; they should be foundational.

I am aware of interesting efforts being done across for language translation, from Ghana NLP’s recent Krio, Temne and Mende TTS models to other around the ecosystem. It will be interesting for us to have conversations around improving data collection.

@vipul @alpha_accurate do you think language data collection is work we can include in our activities for DPG?

Also, @Hassan_Wurie, it will be interesting for us to discuss more on the great strides you have taken. At EasySTEM, we are currently considering efforts around mass-scale data collection (We have compute till early next year to use . And I think it will be interesting to see how the DPG community can be utislised for such