Preserving Heritage: AI Learns Greece’s Endangered Regional Languages and Dialects

by · Greek City Times

Stay connected to Greek City Times for Free on Google News
Click NOW

ATHENS — Of the roughly 7,000 languages spoken worldwide, only a fraction possess the digital footprint required to train modern artificial intelligence models. To prevent regional varieties from falling silent, researchers in Greece are executing an ambitious digital rescue mission—leveraging state-of-the-art technology to introduce Greek dialects to AI training models before they disappear.

Wassim Hamidouche (AI for Good Lab) and Gianna Andronopoulou (Microsoft, right) at the presentation of the Greek part of Lingua.

At a landmark presentation hosted at the Museum of Cycladic Art, Microsoft unveiled the foundational results of the Greek chapter of Lingua—a global program designed to record, map, and digitally preserve languages with a minimal digital footprint.

The Digital Gap and Linguistic Inequality

The meteoric rise of Large Language Models (LLMs) has revolutionised access to human knowledge, but it has also exposed a severe gap in global linguistic inequality. Because AI models train primarily on existing digital texts, unwritten, local, or marginalised languages are systematically excluded from the technological ecosystem.

According to UNESCO data, only about 1,000 of the world’s 7,000 languages have any form of digital presence. “Of these, fewer than 100 have the kind of presence in the digital environment that meaningfully contributes to training large language models,” noted Wassim Hamidouche, a researcher at Microsoft’s AI for Good Lab.

To bridge this gap, Hamidouche’s team launched the “Bring Your Own Language into the LLMs” toolkit, providing researchers with an end-to-end framework to record, map, and classify endangered dialects.

Five Greek Projects Secure Critical Tech Support

The Greek expansion of the Lingua program operates as a strategic collaboration between the Ministry of Culture and the Ministry of Digital Governance. The first round of the initiative has backed five vital linguistic preservation projects:

  • Greek Romani Preservation: The ARSIS program received a € 50,000 grant to record and preserve the spoken language used by Roma communities across Greece.
  • Computational Dialectal Atlas (CoDAG): Developed by the University of Crete to digitally map regional phonetic and grammatical variances.
  • The LALIES Project: Spearheaded by the Athena Research Centre to document nine Modern Greek dialects, ranging from Pontic and Cypriot to the unique Apeiranthos dialect of Naxos, alongside a comprehensive documentation of the Pomak language.
  • T.S.A.K.O.N.I.A.: Managed by the Tsakonia Archive to digitally preserve one of Europe’s oldest living languages.
  • Istorima Archive: A massive digital repository of oral histories capturing the live testimonies of dialect speakers, including Cappadocian, Arvanitika, Pontic, and Tsakonian.

“The first round marks the starting point for building an ecosystem that will bring researchers and the state together,” stated Gianna Andronopoulou, General Manager of Microsoft Greece, Cyprus, and Malta.

Rewriting the Fate of Unwritten Languages

Linguists have long quoted the famous academic maxim that “a language is a dialect with an army and a navy,” meaning institutional power dictates which idioms survive. In the modern era, that power belongs to data. If a language lacks data, it ceases to exist to an AI model.

In Greece, the standardisation of school-taught Modern Greek historically marginalised regional tongues as local folklore. However, technology is now reversing this trend. Experts emphasise that digital recording empowers the communities themselves.

“Digital tools are necessary, but languages survive as long as there are speakers,” explained Stella Markantonatou, emeritus researcher at the Athena Research Centre, highlighting the ongoing importance of fieldwork.

By building a comprehensive digital archive, Microsoft’s Lingua program aims to democratize AI training data. The initiative ensures that future conversational models can interact with users not just in standard state languages, but in the historic, rich dialects of their ancestors.


BBC puts world spotlight on Sfyria: Greece’s disappearing whistled language

Stay updated with the latest news from Greece and around the world on greekcitytimes.com.
Contact our newsroom to share your updates, stories, photos, or videos. Follow GCT on Google News and Apple News.