Student seeks to provide South Sudanese people with access to technology in their own language

Diana Alonso

Taban Ngunar speaks four languages: Swahili, Spanish, English, and his native language, Nuer. He even understands Dinka when he hears it. When asked how a young man can already master such different languages, he simply shrugs and argues that "you learn much faster when you're a child."

Neuroplasticity in childhood is undeniable. Another reality is that Ngunar is originally from South Sudan, a nation with immense linguistic richness. In a territory equivalent to a third the size of Mexico, some 65 indigenous languages coexist (mainly from the Nilotic family), which, along with Juba Arabic, used as a lingua franca, shape daily life and the identity of the entire population.

After gaining independence in 2011, following nearly 50 years of conflict, the South Sudanese government declared English its sole official language. The move aimed to distance itself from the Arab north and use this new language as an engine of development, capable of easing tensions between ethnic groups. This decision ignored the fact that the majority of the population spoke only indigenous languages.

"South Sudan is one of the countries with the highest levels of illiteracy; so many people simply don't understand English," Taban explains. This is compounded by the devastating effects of armed conflict on the country's educational infrastructure. Today, more than 2.8 million children (over 70 percent) are not in school.

Language is political, and Taban experienced this from a young age. To learn it, he first had to master Swahili and English, after moving to Kenya to continue his education.

Despite this, she never lost her connection to Nuer, the language of her community. She recalled that, due to her people's strong cultural roots, her family always made sure she learned to speak, read, and write the language correctly. "The attachment to tradition and family is something that Mexico and South Sudan can compare in […] What defines us is our cultural identity, where we come from; no one can take that away from me," she affirmed.

Taban's mind is constantly analyzing the forms and sounds of language. That's why, upon arriving in Mexico, one of the things that most caught his attention was hearing Spanish, a completely new language to him. He also discovered how much he missed speaking and hearing Nuer, a language he can now only speak during calls with his family.

Taban is one of three South Sudanese students supported by Project Habesha, an international humanitarian initiative that provides access to higher education in Mexico for young refugees. After completing an intensive Spanish program at the National Autonomous University of Mexico (UNAM), he began studying Engineering and Data Science at ITESO. Although Spanish still presents challenges, he is enthusiastic: "It's a wonderful school; I have many friends and classmates who help me understand and explain things to me when I need it."

"Democratizing access to information means developing technology in languages that people understand."

While in some countries technologies like artificial intelligence (AI) permeate daily life, in South Sudan basic digital resources remain largely inaccessible. The global technology industry rarely considers languages like Nuer or Dinka when developing services, tools, or platforms.

To address this digital divide, Taban, along with a group of volunteers, has begun developing tools capable of understanding and processing languages such as Dinka and Nuer. "What we are building are different tools around these languages, ranging from machine translation to large language models (LLM) and text-to-speech (TTS) systems," he explains.

Although companies like Google include some African languages in their translators, accuracy is poor and the information available online is incomplete or inaccurate. "What motivated me was the enormous digital divide that exists in the representation of South Sudanese languages," he adds.

For a machine to learn a language, it needs large volumes of text and audio to train on. However, in languages like Nuer, these resources are scarce or simply nonexistent. Therefore, Taban and his collaborators are working on creating linguistic datasets, which they compile with the help of native speakers and specialists, and then publish on open-source platforms. This way, researchers and developers from anywhere in the world can use them to create and improve technological tools in these languages. "In the future, we want to have many volunteers and significant participation from the South Sudanese diaspora community so they can contribute to the project."

The initiative is already beginning to have an impact. In addition to being used by South Sudanese developers interested in creating artificial intelligence tools for languages like Dinka, the databases promoted by Taban have also sparked the interest of researchers at the University of Surrey in the UK, who even offered to share linguistic materials collected over the last decade.

“There are more than sixty languages spoken in South Sudan. Almost none of them are usable in the software that people rely on every day. You can’t dictate a message in Nuer or Dinka, you can’t have a form translated, you can’t ask a device to read a document aloud.”

“When a language is missing in the digital world, its speakers are pushed into a second language to do ordinary things, or are excluded from doing them,” reads the portal of Dayom, the open-source technology lab of Taban and his team.

For the young man, studying at ITESO and leading this project is an opportunity to develop solutions that may one day benefit communities in his country. "In the future, we would like these models to be integrated into different digital services in South Sudan to help people who don't understand English or speak Arabic, for example, in areas such as healthcare."