Berlin – Since 2021, Wikimedia Deutschland has been coordinating a transnational project that promotes greater language diversity in the digital space. The aim is to preserve endangered languages and improve access to knowledge for underrepresented language communities. The project was made possible by the financial support and expert guidance of the Arcadia Fund.

At the heart of the project are local software teams in Indonesia, Nigeria, Brazil and Ghana, who are working with Wikimedia Deutschland to develop technological solutions that make it easier to edit Wikidata and, in particular, to add language data (also known as lexicographical data). Wikidata has been developed by Wikimedia Deutschland since 2012 and is more than just an open database. It is a knowledge graph that links and correlates information in many languages. Wikidata’s global community consists of more than 12,000 active volunteers who regularly contribute new data on a wide range of topics and provide Wikidata with reliable sources.

In Indonesia, the Wikicollabs software team developed the user-friendly tool Lexica. It helps people to add words from their languages to Wikidata in a structured way, even without prior knowledge of linguistics or data modelling. Lexica currently supports 32 languages, including seven local languages such as Javanese, Banjar, and Minangkabau.

In Nigeria, the Igbo Wikimedians User Group is working to build technical knowledge in Africa with its Wiki Mentor Africa training program. Their goal is for African contributors should be able to develop and improve tools for Wikidata themselves, in their languages, for their needs. Since the program was launched, over 1,300 community members have been reached through hackathons, mentoring, and training.

The Wikimedia Brasil team is developing a new version of the QuickStatements editing tool. This allows many changes to be made simultaneously in Wikidata. The new version 3.0 is intended to be simpler, more accessible, and more powerful.

In Ghana, the One Click Info team is working with the Dagbani Wikimedians User Group on a multilingual browser extension. It displays short, reliable summaries from Wikidata entries. For example, if you highlight an unfamiliar term on a news page, you can get an explanation from Wikidata in the language of your choice at the touch of a button.

As a knowledge graph with currently over 117 million entries, Wikidata is not only readable by humans, but also by machines. This is a crucial prerequisite for underrepresented languages to become visible in digital applications and thus find their way into the digital space.

Most digital applications, including generative AI, from translation tools to chatbots, usually only work for languages that are already well-represented on the internet – examples include English, Spanish, Arabic or several variants of Chinese. The human rights organization The Why has found that 89% of all online content is written in just 10 languages, while over 7,000 languages are spoken around the world. According to the Internet Society Foundation, 20% of the world’s population cannot access information online due to language barriers. The project aims to close this gap with a locally anchored approach and a decentralized development model.

Following the first successful start-up phase, the focus of the second phase will be on scaling and professionalizing the decentralized teams. To make this possible, the project managers are currently looking for new financial support.

The aim remains to place software development in the hands of local communities in the long term because free knowledge needs a strong and diverse infrastructure.