In a pioneering collaboration, linguists from Vitus Bering Kamchatka State University and Russia’s Higher School of Economics have embarked on a crucial project to develop a parallel corpus for the Koryak language. This ambitious initiative, deeply rooted in machine learning technologies, seeks to integrate the speech of Kamchatka’s indigenous inhabitants into modern software products, safeguarding their linguistic heritage for generations to come. The establishment of this electronic database promises to meticulously document linguistic material, ensuring its accessibility regardless of the demographic shifts within the traditional homelands of Koryak speakers.
The Koryak language currently faces a grave threat of extinction. Its remaining speakers primarily reside in remote villages across northern Kamchatka, with younger generations increasingly adopting Russian as their primary language. This shrinking natural environment for communication necessitates innovative approaches to preserve Koryak’s rich vocabulary and grammar, making digitization an indispensable step towards integrating the language into contemporary daily interactions.
A significant technical hurdle for the project lies in the acute scarcity of original source material. Unlike globally dominant languages such as Russian or English, which boast billions of textual documents ripe for machine learning analysis, the entire body of written Koryak texts numbers only in the hundreds of pages. This extreme data deficit renders standard neural network algorithms largely ineffective. To circumvent this, scientists are deploying specialized language models specifically adapted for sparse data sets, combining these with painstaking manual and automated morphological tagging. Essential initial texts and audio recordings are meticulously gathered during annual field expeditions, a testament to the dedication driving this preservation effort.
The resulting linguistic corpus represents a highly structured repository of data, where each word is precisely classified by its grammatical form and contextual usage. Olga Rebkovets, acting rector of Kamchatka State University, highlights the scalability of these solutions, noting, “Language corpora are tools that unlock practical opportunities for language speakers within the digital realm.” The invaluable experience gained from this project is slated for application across other regions of Russia, providing a blueprint for supporting regional languages in the development of machine translators, voice assistants, and interactive chatbots.
The ultimate objective of these researchers is to cultivate a comprehensive digital ecosystem where users can fluidly employ the Koryak language in messaging apps and various mobile services. Early successes are already evident, with the educational mobile application “Koryak tuyu” (Koryak Word) publicly available. Further developments are actively underway, including a localized weather forecast service, an online dictionary complete with audio files, engaging animated projects, and a graphic novel, the publication of which is anticipated by late 2026.
Beyond technological development, the research also delves into neuro-linguistic aspects. In collaboration with the Higher School of Economics’ Center for Language and Brain, Kamchatka specialists are conducting hardware-assisted analysis of articulation among speakers of Koryak, Alutor, and Itelmen languages. Simultaneously, the international LexTALE digital test, designed to assess vocabulary, is being adapted. The statistical data amassed from these combined efforts is expected to form the basis for future federal support programs aimed at Russia’s indigenous minority languages, with policies informed by the actual number of speakers and the preservation status of various dialects.