There are roughly 7,000 languages, and the distribution is extraordinarily lopsided: a few dozen account for most of humanity, while thousands have fewer than a thousand speakers each. Around 40 percent are commonly classed as endangered, and the usual estimate is that one is lost every few weeks.
How a language dies
Almost never by its speakers disappearing. The usual mechanism is language shift across generations.
| Stage | What happens |
|---|---|
| 1 | A dominant language arrives with economic or political weight |
| 2 | The community becomes bilingual |
| 3 | The heritage language retreats to home and ceremony |
| 4 | Parents raise children in the dominant language |
| 5 | Remaining speakers are elderly; no new ones appear |
| 6 | The last fluent speaker dies |
Stage four is where it is decided. Parents choose the language they believe will serve their children, and when schooling, work and media all run in the dominant language, that choice is rational — which is why coercion is rarely needed and why reversal is so hard.
The measure that matters is transmission, not speaker count. A language with 50,000 adult speakers and no children learning it is in worse condition than one with 2,000 speakers that children are still acquiring at home. Counting speakers without asking their ages consistently overstates how secure a language is.
Hebrew is the exception
Hebrew ceased to be an everyday spoken language around the third century CE. It remained in continuous use for liturgy, scholarship and correspondence — read and written by many, spoken natively by nobody — for well over a millennium.
From the late nineteenth century a deliberate revival movement, associated above all with Eliezer Ben-Yehuda, pushed it back into daily use. The conditions were unusually favourable:
| Factor | Why it mattered |
|---|---|
| Continuous written record | an enormous corpus survived intact |
| Widespread literacy in it | many could already read Hebrew |
| No competing shared language | immigrants arrived speaking many languages |
| Strong ideological drive | revival tied to a national project |
| Schools in Hebrew | children acquired it as a first language |
| State backing, later | institutional support at every level |
That combination has not recurred. Modern Hebrew also differs from Biblical Hebrew in ways that reflect its revival — a restructured tense system, European-influenced syntax, and a great deal of coined vocabulary.
Partial revivals
| Language | Where | Method |
|---|---|---|
| Welsh | Wales | Welsh-medium schools, broadcasting, official status |
| Māori | New Zealand | kōhanga reo language-nest preschools |
| Hawaiian | Hawaii | immersion schools from preschool upward |
| Irish | Ireland | compulsory schooling, Gaeltacht support |
| Cornish | Cornwall | reconstructed from written records |
The language-nest model — elders and young children together, with the heritage language the only one used — has been the most transferable idea, adopted from Māori practice in Hawaii and elsewhere.
Welsh shows both the possibility and the limit: speaker numbers have stabilised and Welsh-medium education has grown considerably, but the language has not returned to dominance in daily life outside particular regions.
What documentation does and does not do
Recording an endangered language is genuinely valuable. It preserves knowledge, supports later revival work, and gives a community material to teach from. Cornish was reconstructed largely from written sources.
But a documented language is not a living one. Archives do not transmit a language; households do. Every successful revival has ultimately depended on getting children to acquire the language from people who use it, and documentation serves that goal rather than replacing it.
Why it matters beyond sentiment
Each language encodes distinctions others do not make and knowledge — ecological, medical, navigational — held nowhere else. Linguistic theories built on a handful of well-studied European languages have been repeatedly corrected by data from small languages that behave differently.
And for the communities themselves the loss is not abstract. A language carries the material that gives a community its continuity, and losing it removes direct access to everything recorded only in it.
Dictionaries for languages the internet forgot
NDT Studio publishes offline dictionaries for Cheyenne, Ojibwe, Mizo, Hmong and other under-resourced languages. Free, and no connection needed.
Open the dictionary appFrequently asked questions
How many languages are endangered?
Estimates commonly hold that around 40 percent of roughly 7,000 languages have too few speakers to be secure, with one lost every few weeks on average. Most of the loss is concentrated among languages with small speaker populations and no institutional support.
Has any language ever been successfully revived?
Hebrew is the clear case — from no native speakers for centuries to several million today. Others have been partially revived, including Welsh, Māori, Hawaiian and Cornish, where speaker numbers have risen without the language returning to full community dominance.
What actually kills a language?
Rarely the disappearance of its speakers. Far more often, the community shifts to a dominant language over two or three generations because that language carries economic and educational opportunity. The final stage usually involves parents choosing not to pass the language to children.