25 years of Viquipèdia: a lesson for Catalan in the age of AI

Viquipèdia turns 25, and the debate on multilingualism that shaped it is key to Catalan in the face of artificial intelligence.

communityA tool can use Catalan and still fail to reflect the cultural reality of the people who speak it. It can offer information in our language while leaving out references, experiences and knowledge that belong to our community. This observation is as relevant today, in the face of artificial intelligence, as it was at the beginning of Viquipèdia. And it is a reminder that the presence of Catalan in the digital world is a challenge with a long history, one that must be taken up again with every technological transformation.

The Catalan-language Wikipedia, Viquipèdia, is celebrating its 25th anniversary this year. The Octubre Centre de Cultura Contemporània in Valencia hosted, from 6 to 8 November, a special edition of the Viquitrobada, organised by Amical Wikimedia, which brought together the community of editors to share projects and address the main socio-digital debates.

Joan Abellà, Director General of Accent Obert, took part in these sessions with a reflection on the current challenges facing the Catalan language and culture in the face of artificial intelligence. An intervention that linked the creation of the Gran Enciclopèdia Catalana (Great Catalan Encyclopaedia) with the founding debate on Viquipèdia’s multilingualism and the questions AI raises today, starting from a question that has reappeared for sixty years: how can we take part in building universal knowledge from our own language and experience?

An encyclopaedia to explain the whole world

It is not a new question. In the mid-1960s, Max Cahner and Enric Lluch proposed a project that seemed manageable: to translate and adapt into Catalan the Italian encyclopaedic dictionary Garzanti. The initial proposal was to turn that two-volume work into a three-volume one, adding Catalan proper names and the vocabulary of Fabra’s dictionary. It would have provided an encyclopaedia in Catalan by drawing on an existing work.

But in 1965, with Jordi Carbonell leading the first editorial team, the promoters abandoned that approach and chose to produce an encyclopaedia of their own. Translating and adapting the Garzanti would have meant starting from a selection of topics, sources and criteria of relevance established in another cultural context. Creating the Gran Enciclopèdia Catalana made it possible to decide, in their own voice, what needed to be explained, in how much depth and on the basis of what knowledge.

The ambition was to explain the whole world in Catalan. To do so, it was necessary to bring together specialists, write articles and establish scientific and technical terminology. Betting on a work of their own thus expanded the language’s capacity to produce and transmit knowledge in every field, while contributing a distinct perspective to shared knowledge.

Connections that multiply knowledge

Decades later, Viquipèdia would face a similar dilemma: did editions in other languages have to translate the articles of the English one, or could they create content of their own? The linguistic debate arose naturally just as Wikipedia was taking its first steps. On 27 January 2001, twelve days after its launch, a contributor who had come across an article in German asked on the mailing list what people thought about multilingualism.

A practical question, but only in appearance: which languages should that project include? In March of that same year, the first non-English editions were launched, Catalan among them. But the existence of these editions opened up a second discussion: what relationship should they have with one another?

In October 2002, in a conversation about moving the English edition to the address en.wikipedia.org, this question was made explicit: a Polish editor argued that contributors wanted to take part in the project as partners, not as translators. Jimmy Wales, co-founder of Wikipedia, then formulated an idea that still defines its architecture: a single project with many languages, whose editions are not mere translations.

That debate expressed a desire to combine shared principles (neutrality, openness and encyclopaedic rigour) with the autonomy of communities to incorporate their own topics, sources and approaches. Diversity expanded the knowledge available; circulation between languages, including through translation, made it possible to share it. Language thus became a way of accessing knowledge and a tool for producing it.

A challenge renewed and multiplied by artificial intelligence

Today, artificial intelligence puts this question back on the table: how do we ensure that Catalan is also a language from which the knowledge these tools offer is built? It is not enough for artificial intelligence to speak to us in apparently correct Catalan. We need to guarantee both linguistic quality and its ability to interpret our context and the variety of registers and experiences that make up a shared and diverse culture.

The precedent of Viquipèdia helps to guide this ambition. For the Catalan language and culture to take full part in the development of AI, we need quality content, criteria to evaluate results and communities able to identify gaps and propose improvements. Editors, creators, researchers, institutions and organisations hold knowledge that is essential to define what it means for these tools to work well in Catalan.

Twenty-five years of Viquipèdia show that the universality of knowledge is built from diversity: what we know and explain from here expands what all of us can come to know. But they also remind us that achieving it takes an organised community, shared tools and continuity. In the face of artificial intelligence, the will to take part is not enough: we need an effective capacity to have an impact.

This work connects with Accent Obert’s updated mission: to defend our way of being in the digital world. To do so, we need to connect existing knowledge and efforts, sustain a collective response and take the needs of the Catalan-speaking community to where the decisions that shape its future are made. That is how we can make it possible, also in the face of artificial intelligence, for us to keep creating, sharing and transforming the world with our own voice.

Subscribe to the newsletter

Stay up to date with Accent Obert news and content.