Keyboard Culture: Associate Professor Tabea De Wille’s research at UL is helping to bring the world’s marginalised languages into the digital age

Associate Professor Tabea De Wille’s work begins with a simple idea: language is more than words; it shapes how people live, connect and take part in the world. She trained originally as a theoretical linguist and notes that this background shaped her research perspective: “You understand quite deeply how language is connected to culture and identity.” Her career took a decisive turn during seven years in the video games industry, where she specialised in commercial software localisation.

Commercial localisation involves adapting a software product for multiple regional markets. This requires modifications to user interfaces, compatibility with different writing systems and input methods, compliance with legal and payment requirements, and a range of technical adjustments that allow translation and cultural adaptation to happen. As Associate Professor De Wille puts it: “In order for a piece of software developed in English to work in French or Arabic or Chinese, you have to do things to the user interface. You have to make technical adaptations so that translation and cultural adaptation are even possible.”

This work exposed a global imbalance between the languages supported by digital technology and the languages people actually speak. Even the largest companies only cover a fraction of global linguistic diversity. “Most companies will do maybe five languages. If they are super good, maybe twenty. Google might do a few hundred. But we have nearly seven thousand languages in the world,” she says. She is clear about what qualifies as marginalisation that is being addressed in this context: “I’m not talking about some tribal language on New Guinea spoken by 50 people. I’m talking about languages spoken by millions of people in urban areas that are completely not part of the digital world.”

The disparity is not static. As she observes, “the further technology develops, the bigger the gap becomes,” because advances such as AI scale rapidly in major languages while many others still lack even the most basic digital tools. The outcome is a widening divide in who can participate meaningfully in digital life.

This gap shapes Associate Professor De Wille’s contribution to Translation Commons, an organisation working to reduce digital linguistic marginalisation. She is part of a team that builds digital keyboards for languages that are not ordinarily represented, beginning only when communities request support and working in close collaboration with those communities throughout. 

Designing a keyboard requires a mix of software engineering, linguistic analysis and interface design. Many communities have literacy and linguistic knowledge but not the specialised technical skills to create a keyboard layout that works on modern devices. Translation Commons provides this expertise, producing either tools that communities can immediately use or supporting communities in developing their own.

The effect on digital access is direct and transformative. “I can build you a keyboard right now. You can put it on your phone or desktop, download an app that acts as an intermediary between your phone and its applications, and then you have a functioning keyboard.” For many speakers, this is the first time their language can be typed natively on a device.

The consequences extend well beyond digital communication. Access to information is fundamentally shaped by language support. English speakers can easily find resources to learn skills such as programming, while speakers of many widely-used languages have almost no equivalent options. “If you are an English speaker, you can find two hundred different ways of learning the coding language, Python. If you speak one of the medium-sized Indian languages with millions of speakers, it is going to be very hard.”

Online knowledge repositories reflect the same imbalance. She cites research showing severe underrepresentation of large parts of the world in Wikipedia content and authorship. “The African continent is vastly underrepresented in terms of contributors. The internet doesn’t represent global knowledge, but largely Western knowledge.” This shapes what is recorded, who is represented and which perspectives are preserved.

Algorithmic systems further entrench this inequality. Artificial intelligence relies on substantial data to function accurately, and when languages have no digital presence, the systems that power major online platforms simply cannot process them. “If I want to do something with AI, I can worry about all the algorithms in the world. But if I don’t have data, then it doesn’t matter.” The consequences are extensive. Search engines misinterpret queries, user content is incorrectly moderated and automated systems fail to recognise linguistic patterns essential to understanding meaning.

The absence of data also prevents critical societal tools from functioning at all. “You don’t even have any sort of data to build anything that would allow you to do fake news detection, for example, or that would allow you to do hate speech detection.” This gap is not academic; it has direct political implications. When online platforms cannot identify coordinated misinformation or hate speech in a particular language, communities become vulnerable to manipulation, targeted harassment and political destabilisation. She references the example of the Rohingya, where online hate speech played a documented role in real‑world violence. “It has really big implications for actual global politics,” she says, pointing to the risks faced by communities whose languages are invisible to moderation and safety mechanisms.

Associate Professor De Wille’s current research seeks to change this by supporting communities to create their own digital tools. She is developing a platform that would allow speakers to generate their own keyboard code through AI. “A community could supply a sample text and design parameters. Through multiple steps, there would be AI generated keyboard code that gets pushed back to them for evaluation.”

Student projects at the University of Limerick have already produced workable prototypes, forming the groundwork for what she hopes will be a scalable resource. The overarching goal is to enable communities to bring their languages online independently, without requiring continuous outside intervention.

She also contributes to best practices and resource development through Translation Commons’ collaboration with the UNESCO Indigenous Language Decade (2022 – 2032). As a consulting expert, she assists in evaluating data and producing resources that help communities develop digital capacity responsibly and sustainably. Translation Commons’ contributions feed directly into UNESCO publications and projects.

Across her work, Associate Professor De Wille emphasises that digital participation must be shaped by the communities themselves. Her research and technical contributions illuminate the structural barriers that exclude widely spoken languages from the digital world and demonstrate how targeted interventions can create long‑term, community‑driven digital inclusion.