Which languages are truly about to die?
We hear a great deal about dying languages. The phrase turns up in headlines, documentaries and fundraising appeals, usually attached to a striking number and a photograph of an elderly speaker. But which languages are genuinely on the verge of extinction, and how would anyone actually know?
An intriguing and somewhat irreverent article published on Weirdworm identified seven languages it described as hanging on by a thread. Of the seven, most are exactly the candidates a linguist would expect. They have long histories. They are tied to a limited geographical area. They have been slowly absorbed or dominated by a larger neighbour. That is the classic recipe for language extinction, and it has been cooked thousands of times.
The number everyone quotes
Roughly 7,000 languages are spoken today. Estimates of how many will survive the century vary wildly, from a pessimistic half to a catastrophic ten per cent, and the spread tells you how soft the underlying data is. UNESCO maintains a classification running from vulnerable through definitely, severely and critically endangered to extinct, and even it depends heavily on self-reported national data of very uneven quality.
The distribution is the part that gets lost. Language diversity is not spread evenly. Papua New Guinea alone hosts more than 800 languages. A handful of countries account for the majority of the world's linguistic variety, and they are, almost without exception, the countries with the least money for documentation.
Speaker counts are the wrong metric
This is the mistake most popular coverage makes. A language with 500 speakers sounds doomed and a language with 50,000 sounds safe, and both intuitions can be badly wrong.
What actually predicts survival is intergenerational transmission. Are children learning it at home, before school, from people who are not trying to teach them anything? If yes, the language reproduces itself. If no, it is finished, and the only remaining variable is how long the last cohort lives.
A language with 500 speakers in a community where every child grows up speaking it is in better health than one with 50,000 speakers who are all over sixty. The second has a bigger number and no future. Linguists call it a moribund language, and the term is unsentimental for a reason.
How a language actually dies
The process is almost always the same, and it is almost never violent in the final stage.
- Bilingualism arrives. The community adds the dominant language for work, school and administration. Nothing is lost yet. Bilingual communities can remain stable for centuries.
- The domains shrink. The heritage language retreats to home and ritual. The dominant language takes school, commerce, media, government.
- Prestige inverts. The heritage language starts to feel rural, backward, an obstacle. This is the decisive step and it is psychological.
- Transmission stops. Parents speak the dominant language to their children. It is a loving decision, made for the child's prospects, and it is fatal.
- The last speakers. A generation of semi-speakers who understand but do not produce. Then nobody.
Note what is absent from that list. No massacre, no ban, no burning of books. Colonial suppression and punitive schooling did enormous damage historically, but in the present the killer is usually a rational parental calculation about employment.
The prestige argument that turned out to be false
The belief driving step four is that a child burdened with a minority language will fall behind in the dominant one. It is one of the most thoroughly refuted ideas in applied linguistics. Bilingual children do not suffer academically for it, and there is a substantial body of research suggesting the bilingual brain handles executive function and attention switching rather well.
The tragedy is the timing. By the point research reaches the community, the parents who made the decision are grandparents and the grandchildren do not speak the language. Reversing it means teaching young adults a language their own family stopped passing on, which is possible but enormously harder than never stopping.
What revival actually requires
Hebrew is the case everyone cites, and it is the worst possible model, because it succeeded under conditions almost impossible to reproduce: a state, a coordinated ideological project, mass immigration in which the revived language was the only common tongue, and a liturgical tradition that had kept the written language alive for two millennia.
More instructive is Welsh, or Hawaiian, or Maori. Each combined the same unglamorous ingredients: immersion schooling from early childhood, a broadcast presence, official status that made the language useful rather than merely permitted, and a generation of parents willing to speak it to children who would answer in English at first. It takes decades and the results are partial. But endangered languages are not saved by dictionaries and archives. Those preserve a record. Only children preserve a language.
The honest answer to the question
Which languages are truly about to die? Overwhelmingly, the ones you have never heard of, in places that are never filmed, whose last fluent speakers are in their seventies and whose grandchildren are pleasant, capable people who cannot understand them. There are hundreds of these. Databases like the Endangered Languages Project and the catalogue at Ethnologue track what can be tracked. The list of seven is a good read. It is also, unavoidably, a rounding error.