What language revival actually involves

There are somewhere around seven thousand languages currently spoken. Estimates vary on how many are endangered, and the commonly cited figures suggest that a large proportion have small and declining speaker populations, with a substantial number likely to cease being spoken within this century.

The reasons are consistent: economic pressure to speak a dominant language, education systems that operate in it, media, migration, and in many cases explicit historical policies that punished children for speaking their own language at school.

How a language actually stops

Rarely by the last speaker dying. That is the moment reported, and it comes at the end of a long process.

The mechanism is intergenerational transmission failing. Parents who speak the language do not use it with their children, usually for reasons that are entirely rational at the time — the dominant language leads to work, education and safety, and the smaller one is associated with disadvantage.

Once a generation grows up without it, the language has perhaps sixty years left as a spoken thing. The linguists' rough test is not how many speakers there are but how old they are and whether children are learning it.

A language with ten thousand elderly speakers and no child speakers is in worse condition than one with five hundred speakers where the children speak it.

What documentation does and does not achieve

Recording a language — grammar, dictionary, texts, audio — is valuable and it is not revival.

Documentation preserves information about a language. It does not produce speakers. A fully documented language with no speakers is a research resource, and there are many.

Documentation does become important later. Communities working to bring back a language that has stopped being spoken depend entirely on whatever records exist, which is why documentation carried out decades earlier, sometimes by people with mixed motives, can turn out to be the material a revival is built from.

What seems to work

Immersion education. The most effective approach where it can be implemented. Children taught through the language rather than about it, ideally from an early age.

The best-known models include Māori language nests in Aotearoa New Zealand, Hawaiian immersion schooling, and various programmes elsewhere. These have produced new generations of speakers, which is the only outcome that ultimately counts.

Master-apprentice pairing. A fluent speaker and a learner spending extended time together, doing ordinary activities entirely in the language, without translation. Used where speaker numbers are too small for a school.

Creating domains of use. A language survives where it is needed. Broadcasting, publishing, official use in local government, signage, workplaces, social media. If the only place a language is spoken is a classroom, it is a school subject rather than a language.

Community control. Programmes designed and run by the speaker community consistently do better than programmes designed for them. This is not primarily an ideological point — it is that community-led efforts are the ones people actually participate in, and participation is the whole mechanism.

The Hebrew case, which everyone cites

Hebrew is the standard example of a language brought back into everyday use after centuries of being primarily liturgical and literary.

It is also close to unrepeatable. It had an enormous continuously used written corpus, a population with religious literacy in it, a state that adopted it as an official language, a shared incentive among speakers of many different first languages to adopt a common one, and a level of institutional support that no minority language revival has ever had.

Citing it as a general precedent tends to obscure how unusual those conditions were.

What is at stake

The scientific argument is real. Each language encodes a particular grammatical system, and linguists study the range of possible human languages using the diversity that exists. Losing languages narrows what can be known about that range.

The knowledge argument is more specific. Traditional ecological knowledge — plant uses, seasonal indicators, place names encoding landscape information — is frequently held in the language and does not survive translation well.

But the argument that matters most to speakers is usually neither of these. It is that a language carries the humour, the songs, the way relatives are addressed and the way the world is described in a particular family and place. Losing it is a bereavement, and it is described that way by people going through it.

The realistic picture

Some revivals have worked. Several languages have more speakers now than they did forty years ago, and some have gone from no child speakers to a functioning school system.

Many will not, because the economic and demographic pressures are enormous and the resources are not there.

And the useful thing outsiders can do is fairly limited and fairly concrete: support the institutions that speaker communities are running, avoid treating languages as heritage objects to be curated by others, and recognise that the decision about whether to keep a language belongs to the people who speak it.