Nordic cooperation to protect minority languages in the age of AI 

The forces of modernity and globalization have strengthened to gradually cause vulnerability for spoken languages. In the times of AI, people around the world now face unprecedented pressure to adopt to common languages used in government, commerce, technology, entertainment, and diplomacy. 

These perspectives are among the topics to be discussed when ASTIN (the Workgroup for Language Technology in the Nordic Countries), in collaboration with SFI NorwAI, organize a conference in Trondheim on language technology for the languages of the Nordic region to foster cooperation across national borders and languages on November 5th-6th.  

―Because some minority languages are shared and all share challenges in our region, we are convening a Nordic conference to learn from each other, postdoc fellow Lars Bungum explains.  

Bungum acts as local organizer in tandem with NorwAI colleague Karolina Storesund. 

A shared understanding

―We aim to foster a shared understanding between the academics, policy makers, language councils and users of language technology on where we stand and where we need to go, says Lars Bungum.  

 ―We will learn from experiences in the Faroe Islands, Iceland, Scandinavia, and Ireland, as well as larger European efforts to create large language models for its regions. After luminous talks, a panel discussion will connect the lessons learned and summarize the findings, he says.  

Borealium and poster session

The conference organizers will also launch a new language technology platform for small languages in the Nordics called Borealium: It brings together language technology products, solutions and resources for small languages in the Nordics in an audience-friendly format.  

Recipes on how to use the products must be in a language that people without high technical competence can understand, and in many of the Nordic minority languages 

A threat

Historically, many languages die out gradually as successive generations of speakers become bilingual and then begin to lose proficiency in their traditional languages. According to Encyclopedia Brittanica, linguists estimate that of the world’s approximately 6,900 languages, more than half are at risk of dying out by the end of the 21st century.  

Arguably, the advent of Large Generative Language Models exacerbates this homogenizing effect, as the world languages American and Chinese languages dominate the training material for the time being. The Language Council of Norway who is one of the organizers of the conference, says in their invitation that the impact by LLMs has been greater and more sudden that expected.

Kick-off 

The conference kicks off with a key note by Associate Professor Jussi Karlgren at the University of Helsinki who will give an overview of how learning and generative language models have, in the past two years, transitioned from engineering experiments to components in everyday consumer products. Thus, some of the optimism shown by language technologists working with learning models since the fifties has been justified by this real-world impact. However, important research into the capabilities and limitations of these models is still needed, according to Jussi Karlgren.  

Also interesting among a multifaceted agenda is a report on the current state and prospects of language technology for less resourced languages by Steinþór Steingrímsson at the The Árni Language Magnússon Institute for Icelandic Studies. The report was initiated by last years’s Iclandic presidency of the Nordic Council of Ministers to compile a report of the Nordic minority languages and the languages of the island nations in the Nordic region.  
 
Professor Jon Atle Gulla at NorwAI will talk on the experiences six months after NorwAI launced six language models in Norwegian, and Magnus Sahlgren, head of research at the National Language Understanding at AI Sweden will bring a European perspective and discuss whether the Nordics are catching up or falling behind in the development of LLMs.

Nordic coop

The conference is a cooperation between ASTIN (Arbetsgruppen för språkteknologi i Norden) with members UiT The Artic University of Norway, Språkrådet vid Institutet för språk och folkminnen in Sweden, Dansk Sprognævn in Denmark and  Språkrådet, NorwAI in cooperation with NorwAI, Norwegian Research Center for AI Innovation. People from all the Nordic countries and including Greenland and Northern Ireland are participating.  

The conference is an open event free of charge, financed by the Nordic Council of Ministers, the intergovernmental body for cooperation in the Nordic region.   

See the program 
 

PUBLISHED: 2024-10-29