Beginning with the Cretan dialect, a new collaboration between the Greek government, ElevenLabs and the Athena Research Center aims to record and digitally preserve forms of Greek shaped over centuries by geography, migration and history.
In Greek families, a single word can reveal an island, a village or an entire family journey.
The Greek spoken by a grandmother from Pontus may sound different from that of a grandfather from Crete. A Cypriot household in London, a Cappadocian family in northern Greece and a Greek-speaking community in southern Italy may all carry forms of the language shaped by centuries of migration, isolation and contact with neighboring cultures.
Now, Greece is turning to artificial intelligence in an effort to ensure that some of those voices are not lost—or excluded from the technologies through which people increasingly communicate.
The Greek government has entered into a strategic partnership with the artificial intelligence voice company ElevenLabs covering public services, tourism and the preservation of Greece’s linguistic heritage.
The agreement was signed on May 28 at the Maximos Mansion by Minister of Digital Governance and Artificial Intelligence Dimitris Papastergiou and ElevenLabs co-founder and CEO Mati Staniszewski, in the presence of Prime Minister Kyriakos Mitsotakis.
The Athena Research Center and its Institute for Language and Speech Processing will serve as the Greek scientific partners for the linguistic portion of the initiative.
According to the Athena Research Center’s announcement, the first phase will focus on Cretan, with the possibility of adding other Greek dialects in the future.
The initial step will be to record the natural speech of dialect speakers. Those recordings are intended to form the foundation of a digital voice capable of reproducing the sounds and characteristics of the dialect.
The details are still being developed. The partners have not announced when recordings will begin, how speakers will be selected, which parts of Crete will be represented or which dialect will come next.
What has been announced, however, is the beginning of a national effort to bring forms of Greek that are poorly represented in modern technology into the age of artificial intelligence.
More Than One Way to Speak Greek
Modern Standard Greek is the language taught in schools, used in government and heard across national media. But it represents only one part of a much larger linguistic landscape.
The Greek language has never been spoken in exactly the same way everywhere. Mountains, islands, migration routes, foreign rule and long periods of separation allowed communities to develop their own vocabulary, pronunciation and grammar.
According to the Academy of Athens’ Research Center for Modern Greek Dialects, regional forms of Greek had already begun taking shape during the Byzantine period. Most modern Greek dialects developed from the Hellenistic Koine, the common form of Greek that spread widely during the centuries following Alexander the Great.
There is one especially striking exception. Tsakonian, still spoken in a small area of the eastern Peloponnese, is generally regarded as the only surviving descendant of ancient Doric Greek rather than of the Koine from which most other modern varieties emerged.
The dividing line between a dialect and a local idiom is not always simple. Linguists generally use the word “dialect” when the differences are substantial enough to make understanding difficult for Greek speakers from elsewhere. Other regional forms may be described as idioms or local varieties.
In everyday conversation, however, Greeks use “dialect” more broadly to describe the distinctive speech of Crete, Cyprus, the Ionian Islands, the Dodecanese, northern Greece and numerous other regions.
Together, these forms record the movement of Greek communities across the Mediterranean, the Black Sea and Asia Minor.
Why the Project Begins in Crete
Cretan is a logical place to launch the initiative because it remains a living form of Greek with a strong oral tradition, a large community of speakers and an extensive diaspora.
It is also a dialect with an important literary history.
During the period of Venetian rule, Crete became the center of a major literary and theatrical flowering. Writers developed a literary language based on the speech of the island, producing works including Georgios Chortatsis’ “Erophile” and Vitsentzos Kornaros’ “Erotokritos.”
Written during the Cretan Renaissance, “Erotokritos” contains more than 10,000 rhyming verses and became one of the most widely loved works of Greek literature. Its language helped carry Cretan speech far beyond the island.
The Center for the Greek Language notes that nearly all the principal features associated with modern Cretan can already be found in literary works of the 17th century, suggesting that the dialect had developed its recognizable form centuries earlier.
Cretan should not, however, be described as a language that has already disappeared or is spoken by only a few people.
Researchers working on the first digital linguistic tools for Eastern Cretan describe it as a living but technologically under-resourced dialect. It continues to be spoken across the island and among Cretan communities abroad, although its use varies considerably by region, generation and social setting.
The problem is not that Cretan has vanished. It is that digital systems are usually trained primarily on Standard Modern Greek.
A Cretan speaker may be understood perfectly well by family and neighbors while being misunderstood by an automated transcription service, voice assistant or artificial intelligence system.
From Pontus to Cappadocia
The future expansion of the project could open a much broader chapter in the documentation of Greek linguistic history.
Pontic Greek developed among Greek communities on the southern shores of the Black Sea and in neighboring parts of the Caucasus. Following the upheavals and population movements of the early 20th century, it was carried into Greece and throughout the diaspora.
Today, Pontic is spoken primarily in northern Greece and within Pontian communities around the world. Its long history in the Black Sea region left it with influences from Turkish, Armenian, Georgian and Russian, while retaining older Greek vocabulary and grammatical features.
For descendants of Pontian refugees, the dialect is inseparable from songs, mourning traditions, storytelling and family memory. Its survival is therefore not simply a linguistic matter, but part of preserving the history of a people uprooted from their ancestral homeland.
Cappadocian Greek carries a different history of survival.
The dialect developed in Greek-speaking communities in central Anatolia, where generations of close contact with Turkish produced one of the most unusual forms of Greek. Turkish influence entered not only its vocabulary but aspects of its grammar and sentence structure.
Following the population exchange between Greece and Turkey, Cappadocian speakers settled mainly in northern Greece. The dialect declined rapidly as refugee communities adopted Standard Modern Greek, and it was at one point believed to have effectively disappeared.
The Academy of Athens now records Cappadocian as surviving in a limited number of communities in northern Greece. Researchers have also begun creating digital records and language tools for Cappadocian and other endangered varieties from Asia Minor.
These dialects are linguistic records of coexistence, separation and forced migration. Their Greek foundations remain visible, but so do the centuries during which their speakers lived beside other peoples and languages.
A Living Link to Ancient Doric
Tsakonian occupies a unique place in the Greek linguistic world.
Spoken in villages around Leonidio, Tyros and the eastern slopes of Mount Parnon, it differs so sharply from Standard Modern Greek that some linguists classify it as a separate language rather than merely a dialect.
The Academy of Athens describes Tsakonian as the only surviving descendant of ancient Doric Greek.
That does not mean a modern Tsakonian speaker talks exactly like a Spartan of antiquity. Languages change continuously. It does mean that Tsakonian followed a different historical path from nearly every other form of Greek spoken today.
Its survival in a small and geographically concentrated population makes it one of the most significant candidates for careful recording and digital preservation.
Greek Beyond the Borders of Greece
The linguistic geography of Hellenism also extends well beyond the modern Greek state.
Griko, sometimes called Grecanic or Southern Italian Greek, is still spoken in communities in Calabria and the Salento region of southern Italy. It developed over centuries within Greek-speaking communities surrounded by Italian and regional Romance languages.
The precise balance between its ancient Greek and later Byzantine origins remains a subject of scholarly debate. What is clear is that Griko represents a surviving Greek linguistic presence in southern Italy and is only partially understandable to speakers of Standard Modern Greek.
Cypriot Greek offers another example of a dialect sustained both in its homeland and through migration.
It is the everyday spoken language of most Greek Cypriots and remains widely used within Cypriot communities in Britain, Australia, Greece and elsewhere. The dialect also has a long documented history. Many features recognizable in present-day Cypriot already appear in medieval texts, including the chronicles of Leontios Makhairas from the 15th century.
For diaspora communities, these varieties often survive in unexpected ways. A second- or third-generation Greek may speak limited Standard Greek yet recognize the local expressions, pronunciations and rhythms heard in childhood from parents or grandparents.
The dialect becomes the emotional language of the family, even when it is no longer used fluently.
Teaching Machines to Hear the Whole Language
The challenge for artificial intelligence is straightforward: machines learn from data, and regional dialects generally have far less recorded and transcribed material than national standard languages.
The Athena Research Center has already developed or assembled resources for numerous Greek dialects and language varieties, including Cretan, Cappadocian, Cypriot, Aivaliot, Lesbian, Aperathiot and Messenian.
Its work also includes Arvanitika, Pomak and Gheg Albanian, which have long been spoken within Greece but are not dialects of Greek. Their inclusion reflects the wider linguistic diversity found within the country.
Recent research has shown how far current technology still has to go. A 2026 study of automated speech recognition found that leading multilingual systems struggled to accurately transcribe several Greek dialects, particularly those that differ most strongly from Standard Modern Greek.
Building better systems requires real recordings, carefully checked transcriptions and the participation of people who understand how the dialect is actually spoken.
Artificial intelligence can process and reproduce speech, but it cannot decide by itself which village, generation or local variation represents an entire dialect. That work requires linguists, cultural organizations and speakers from the communities involved.
A Project Still at the Beginning
The partnership with ElevenLabs has been formally announced, but it remains in an early stage.
The government has not yet disclosed the project’s budget, timetable or the number of speakers who will be recorded. It has not said whether the resulting recordings and digital models will be freely available to researchers and the public.
Questions also remain about consent, ownership and representation. A digital voice trained on community speech raises important questions about who controls that voice and how it may be used in the future.
Those details will determine whether the initiative becomes a lasting cultural resource or simply an interesting technology demonstration.
The first goal is nevertheless significant: to record human voices before the knowledge carried within them grows harder to find.
ElevenLabs says the collaboration will begin with Cretan before adding other dialects. The company has said the broader initiative may eventually include regional forms spoken by only a few hundred people in isolated communities.
Mitsotakis described the project as an opportunity for AI to capture Greece’s linguistic wealth and keep it available for future generations.
For Greeks abroad, that idea carries particular meaning. The language of the diaspora has never been only the standardized Greek taught in a classroom. It is also the Greek of villages, islands, refugee settlements and family kitchens—the words that traveled in suitcases and survived in homes thousands of miles away.
