Whose network? The Erzya language online: two spheres and one question

Between the 2010 and 2021 censuses, the number of people in Russia who reported command of the Erzya and Moksha languages fell from roughly 431,000 to 280,000 — a third lost in a decade (the census still lumps the two languages into a single "Mordvin" column, so the share of Erzya proper within these figures is unknown). UNESCO lists Erzya among endangered languages. Demography alone does not explain the decline: over the same period, native-language classes in schools became optional (2018); official record-keeping in Erzya remains non-existent; and the independent structures that had worked for the language for decades were brought under control or dismantled. This is the backdrop against which the digital sphere must be judged — the only domain where the Erzya presence has actually grown in recent years. The written record of the Erzya word goes back more than three centuries, to monuments researchers date to the late seventeenth century, and today that tradition is migrating from paper to the network. The only question is whose hands are doing the work, and on whose terms.

The digital space of the Erzya language today splits into two spheres that barely touch.

The first sphere is state-run and state-adjacent. Ogarev Mordovia State University, under a 2023–2024 grant of 8.5 million roubles, compiled a parallel Russian–Erzya corpus of 50,000 aligned sentences and in 2024 launched a neural machine translator as a Telegram bot — within the federal programme "Priority 2030". In July 2025, Erzya and Moksha appeared in Yandex Translate; that project has been run since 2023 jointly with the House of the Peoples of Russia and FADN — the very federal agency in charge of nationalities policy. Digital schooling belongs to the same sphere: the commercial platform iSmart, under an agreement with the Mordovian authorities, translated a set of school exercises into Erzya and Moksha. The tools are real and useful, but their framing is telling: official releases file them under "the Mordvin languages", their launches are synchronised with the symbolic calendar (the International Decade of Indigenous Languages, Russia's "Year of the Unity of Peoples"), and their existence depends on the political climate, not on the Erzya community. The state that made the language optional at school presents a translator as proof of its care.

The second sphere is independent: academic, volunteer, diasporic and international. The Erzya corpora built by the linguist Timofey Arkhangelskiy — 2.3 million running words of literary texts and another 830,000 from social media — are the product of independent scholarship; the social-media corpus, incidentally, is the best evidence that Erzya lives online in everyday use, not only in official reporting. The Erzya Wikipedia has travelled a generation-long road: the section left the incubator on 26 May 2008 — the same day as the Moksha and Sakha editions — wrote its six-thousandth article in May 2020, and as of 7 July 2026 holds 7,870 articles with 154,500 edits in its entire history. That is one article fewer than in early February: in the first half of this year the section did not grow, it shrank. All of it rests, by the community's own admission, "on the constant work of a few people": over the past 30 days only 16 users made edits, and two administrators remain. Alongside Wikipedia, Wikinews and Wikisource operate in Erzya. The diaspora plays a special part: only about 28% of speakers live in Mordovia itself, the rest are scattered from Samara to Tallinn, and for them the network is practically the only Erzya-speaking environment there is. Emigration has produced a digital life of its own: the Erzya of Ukraine recorded audiobooks in their native language, and in 2020 the journal ĚRZÄŃ VAL began publication in the Latin script — a Latin orthography for Erzya was developed that same year. This sphere exists without the patronage of the Russian state — and it is precisely this sphere that has taken the blows of recent years.

The independent sphere also has an international wing, rarely mentioned inside Russia. At the University of Tromsø, the language-technology laboratory Giellatekno, together with Jack Rueter, a Helsinki-based scholar of Finno-Ugric languages, has spent decades building open computational morphology for Erzya: an analyser that parses the grammar of words in running text, with all code published openly (the work is summed up in Rueter and colleagues' 2020 paper "Open-Source Morphology for Endangered Mordvinic Languages"). The same Tromsø platform hosts a free online dictionary of Erzya and Moksha — and it is no accident that it lives at valks.oahpa.no: valks is simply the Erzya word for "dictionary". On these foundations, the Apertium project maintains an open machine-translation system between Erzya and Moksha — a detail worth savouring: between "dialects of a single Mordvin language", no machine translator would ever be needed. This entire infrastructure — open code, free licences, servers outside Russian jurisdiction — is built on the principle opposite to the first sphere's: it cannot be managed, and it cannot be switched off by a ministerial decision.

The story of Erzyan Mastor ('Land of the Erzya') needs to be told accurately here, because it is a model of what is happening. The newspaper was founded in 1994 by the Erzya national movement and for three decades remained the last periodical in Mordovia outside the authorities' control: it was prosecuted for "extremism" in 2007–2009 (neither the Supreme Court of Mordovia nor the Supreme Court of Russia upheld the charges), Roskomnadzor, the state media regulator, tried to close it in 2015, and in 2009 the chairman of the Foundation behind it was detained for handing out copies at the Fourth Congress of Finno-Ugric Peoples in Saransk. When the paper could not be closed, the system was replaced from within: in 2020 the founders were removed from the Foundation for Saving the Erzya Language, the paper's publisher: Nuyan Vidyaz (Yevgeny Chetvergov), who had registered the Foundation in 1995 and edited the paper for many years, and Kshumantsyan Pirguzh, the reviver of Raśkeń Ozks, the traditional all-Erzya prayer gathering, and a former inyazor — the elected elder of the Erzya people. Independent media describe what happened outright as a takeover of the Foundation and the newspaper; they also report a press campaign against movement figures in the government-controlled republican media. By 2020 the paper's print run had fallen to 200 copies. The final blow came in October 2023: mass FSB searches in the "Kirdiyur community" case, with the 82-year-old Kshumantsyan Pirguzh and the 89-year-old Nuyan Vidyaz among those detained. The digitised archive of Erzyan Mastor survived on Wikisource — the paper had managed to adopt a free licence back in 2017, still under its old editorial team, thanks to wiki volunteers. Today that digital repository is very nearly all that remains of thirty years of independent Erzya journalism.

Comparison with the neighbours confirms the diagnosis, with one correction. Yes, Komi has what the Erzya lack: the permanent FU-Lab laboratory in Syktyvkar, a corpus of over 50 million running words (twenty times the Erzya one), an online library of 6,500 works with parallel reading, electronic dictionaries and a spell-checker; both Mari Wikipedias — Meadow and Hill — are larger than the Erzya one, and Mari and Udmurt entered Yandex Translate earlier than Erzya and Moksha and have already gone through a quality-improvement cycle. But FU-Lab is a unit of the state-run House of Friendship of Peoples — the same state sphere, only older and more systematic. The Komi model provides tools; it does not give the language a voice. The contrast within the Erzya–Moksha pair is instructive too: in the state sphere the two languages move strictly in tandem — added to Yandex simultaneously, their university corpora compiled in parallel — whereas the volunteer energy is almost entirely Erzya: the Moksha Wikipedia, launched on the very same day in 2008, has remained several times smaller. The state operates with the administrative category of "the Mordvin languages"; living grassroots movements belong to each language separately, and Erzya's is markedly stronger. Most telling of all is the Veps example: the Veps Wikipedia, for a language with almost no speakers left, was at one point twice the size of the Erzya one — "Erzyans today lack cohesion even in supporting internet projects," the Erzya wiki community itself observed with bitterness. It is not the number of speakers that is decisive; it is the number of people willing to work — and an environment that lets them. The Erzya have had those people taken away: some removed from their posts, some searched, some driven into exile.

So what do the Erzya lack in the digital sphere? Not merely a spell-checker, keyboard layouts, speech synthesis and an online library — that technical list is accurate but secondary. What is missing is an independent institutional footing beyond the reach of FADN or the FSB: a durable structure with open data, a distributed team, and resources mirrored outside any single jurisdiction. The experience of recent years has proved that any Erzya-language infrastructure inside Russia exists exactly as long as the state permits it — and that an archive placed under a free licence in time outlives the takeover of its own newsroom.

Valks was built with this lesson in mind: an open dictionary, free data, independence from grant cycles and administrative favour. Our priorities follow from the list of gaps — machine-readable dictionary data for developers, the spoken word, phraseology, links to independent corpora — and the preservation of what the community has already created. The digital sphere will not save the language by itself. But it is the only sphere where the Erzya word can exist today without asking anyone's permission.