Language abandonment

Why can a language disappear when almost every speaker wants it to survive?

Ask the members of a community whose language is fading whether they want it to survive, and the answer is very often an emphatic yes. Ask them whether they are raising their own children in it, and the answer is very often no. The two answers come from the same people, sincerely, in the same conversation.

That combination is the thing to explain. It is easy to account for a language dying under suppression — punishment in schools, displacement, prohibition — and that history is real. But languages also disappear where nobody is forbidding anything and where the speakers' stated preference is overwhelmingly for survival. A story about coercion cannot cover that case. Something else is doing the work.

Start by asking what a language is worth to one person. Not culturally — practically. What does knowing it buy you?

Almost all of its value is in who else knows it. A language you alone speak has no communicative use at all. Its usefulness scales with the number of people you can use it with, and with what those people can do for you. This is the crucial asymmetry: the benefit of speaking it accrues to you only through others' choices, while the cost of acquiring and maintaining it is borne entirely by you.

Now put a parent in that position. They face a genuine allocation problem. A child's early years are finite; schooling, employment and public life run in the majority language. Every parent knows their child must have that one. The question is whether to also invest heavily — and it is heavy, requiring consistent household use, not occasional exposure — in a language whose future usefulness depends on how many other parents make the same investment.

Ask what the individually sensible answer is when you are uncertain about the others. If the language survives regardless, your child gains a modest amount from your effort and loses little by your abstaining. If it does not survive, your effort was largely spent and your child arrives at adulthood having had less time in the language that pays. Either way, on a narrow reading, abstaining looks defensible. And a collectively catastrophic outcome is assembled entirely from individually defensible choices.

That is the structure of a collective action problem, and it is worth naming precisely: it is not that people are selfish. The parents in question typically do value the language intrinsically and are willing to bear cost for it. It is that the return on their sacrifice is contingent on a coordination they cannot observe or enforce.

Push one step further and something sharper appears. Because value rises with the number of speakers, the game has more than one stable resting point. A community where everyone expects transmission to continue is stable — each parent's investment is rational because the others' is too. A community where everyone expects the language to fade is also stable, for exactly the mirrored reason. Both are self-fulfilling, and which one a community occupies is a fact about shared expectations at least as much as about shared preferences.

That explains the puzzling speed of the collapse. Shift does not proceed as a gentle linear decline; communities often go from near-universal transmission to almost none within two or three generations. If you model each speaker's choice as depending on the current share of speakers and the language's perceived status — as Abrams and Strogatz did in a much-discussed 2003 model — you get exactly that: a threshold, and below it a rapid slide to zero. Their model is deliberately crude, treats languages as competitors for speakers, and has been fairly criticised for ignoring stable bilingualism, but the qualitative point survives its simplifications.

There is also a pivot in the chain that makes the collapse hard to reverse. Joshua Fishman's work on reversing language shift places the decisive link at intergenerational transmission in the home. Everything downstream of it — signage, media, adult classes — can be present and impressive while the home link is broken, and the language still ends. This is why revival efforts that target schooling alone so often disappoint, and why the interventions with the strongest record aim at the youngest children in a language-saturated setting, as with the Maori language nests established from 1982.

Two corrections are worth making.

The first is that this argument does not replace the history of coercion; it sits underneath it. Suppression works precisely by altering the payoffs and the expectations the argument above depends on — lowering the perceived status of the language, breaking the domains in which it was useful, and teaching parents that their children will be punished for it. The mechanism explains why the damage outlives the policy by decades: the equilibrium shifted, and removing the coercion does not shift it back.

The second is that a collective action problem is not a prophecy. Its distinguishing feature is that it is solvable by changing the strategic situation rather than by changing anyone's values. Anything that makes other people's commitment visible and credible does real work: a school where the language is the medium rather than a subject, an employer that requires it, a legally guaranteed domain in which it must be used. Each of these raises one parent's expected return on the same sacrifice they were already willing to make. That is why the successful cases — Hebrew's revival, the partial recoveries in Wales and New Zealand — combine institutional guarantee with early-childhood immersion, rather than relying on exhortation.

• Why stable long-term bilingualism persists in some regions and collapses in others with similar demographics.
• What "reversing language shift" achieves when fluent adult speakers are already gone.
• How writing systems and standardisation can help a language and, by privileging one variety, harm its dialects.

Every term the collection defines is gathered in the glossary.