Sometimes It’s O.K. to Let a Language Die
How sad should we be when a language dies? Consider the case—call it callous or simply tough-minded—for saying, “Not at all.” A language, when it comes down to it, is a means of communication. When too few people are using it to communicate, it has stopped doing its principal job, and it may as well be allowed to die. It is, of course, a tragedy when languages are killed, especially when they are killed out of malice or bigotry. But death by natural causes—well, that’s just how things go.
Despite its ominous title, Sophia Smith Galer’s “How to Kill a Language: Power, Resistance, and the Race to Save Our Words” (Crown) is not a guide for would-be perpetrators of linguicide. Smith Galer has much to say against the case for putting languages out of their misery, but her focus is on a point with which even the most unsentimental can surely sympathize. Whatever the wisdom of letting dying languages die, the ethics of killing a thriving language are quite another matter.
Take Kurdish. The 2023 earthquake that left more than fifty thousand people dead in Turkey and Syria was particularly devastating in Turkey’s Kurdish-speaking southeast. Turkish administrators announced translation support for earthquake victims in several languages, including Russian, German, and Pashto—but not Kurdish, the language of some fifteen million Turkish citizens. Smith Galer tells us that the decision was a continuation of decades of official policy. It has been, at different times, illegal in Turkey to teach Kurdish, to publish books in Kurdish, even to speak the language while occupying public office. There were earthquake survivors who feared calling out for help in the language, in case the rescuers turned away.
Discover notable new fiction, nonfiction, and poetry.
Or consider Ukrainian. A long essay that Vladimir Putin wrote in 2021 characterized it as, at best, a regional variant of Russian—a sort of uppity, upstart dialect—whose speakers lacked a legitimate claim to national self-determination. Seven months after the essay appeared, Putin invaded Ukraine. Since 2022, wherever Russian forces have occupied Ukrainian territory, they have torn down Ukrainian road signs, and insisted on Russian as the language of schooling and administration. When Ukrainians see books in their language being burned, some of them are reminded of a longer history of persecution, such as Catherine the Great’s ban in 1784 on Ukrainian-language church services. (That’s one of several dozen events commemorated in “Linguicide,” a Kyiv public-history project made up of QR-coded plaques and a companion app.)
In these cases, I’d say, talk of “killing” is justifiable: the intended destruction of the language is part of a coördinated assault on a people’s collective existence and their claim to nationhood. But how far might one extend the term before it does start to seem an overstatement? Smith Galer writes powerfully about the experiences of Kichwa speakers in the Ecuadorian highlands. Here we see fewer overt acts by tyrannical states to suppress the language. What we find, instead, is a history of colonial contempt, of treating fluency in Kichwa as a marker of poverty and backwardness. And the death of the language comes from choices made by Kichwa-speaking Ecuadorians to stop speaking Kichwa to their children. One way languages gain and lose adherents is through calculations people make about what a given language can do for them.
Of course, those choices are not made against a background of equal power, which is why the situation is morally weighted. Still, the ideal case of a language chosen, or abandoned, under conditions of equal power is scarcely to be found. There is a fallacy in thinking of languages as brute and unchanging facts, laid out distinctly like elements in the periodic table. Languages are historical formations. They emerge, spread, splinter, standardize, contract, and disappear, much as peoples, states, and religions do, through forces like conquest, exchange, and chance.
More boldly, Smith Galer’s story of linguicide encompasses her own loss of the Emilian dialect—or, as she prefers to render it, Emigliän dialët—of northern Italy (“more like a kind of angry French”) that her mother and grandmother spoke to each other. She did not learn either standard Italian or the Emigliän variant for the simple reason that she was brought up in London by a family who had decided to speak to her entirely in English. That decision surely belonged to a wider history of migration, aspiration, and accommodation to the dominant language. Yet Smith Galer, citing the linguists Tove Skutnabb-Kangas and Robert Phillipson, suggests that linguicide might refer not just to the “outright killing of a language” but also to “letting a language die, or not supporting its coexistence”—which amounts to “covert language murder.”
The linguist David Crystal, in his indispensable scholarly survey “Language Death,” has a term for the sort of thing that happened to Smith Galer and her dialët: cultural assimilation. The pressure to assimilate linguistically can be, he points out, either “top down” (laws, incentives) or “bottom up” (fashion, peer pressure). But it is hard even for dispassionate surveys to avoid lapsing into an elegiac mode in which these important distinctions are flattened into a single tragedy. It is one thing to be indignant about a Kenyan schoolchild asked to wear a sign reading “I AM A DONKEY” for speaking Kikuyu. But how should we feel about a Kichwa speaker who moves to Lima and starts living her life in Spanish in response to the demands of the labor market, especially when the Kichwa speaker doesn’t see the switch as involving any great loss?
The idea that we have a special duty to prevent the death of languages is instantly compelling to anyone who loves languages; it has a good deal in common with that other conservationist instinct directed at giant pandas and polar bears. The essential argument runs something like this: language death is very bad, and therefore we must do everything we can to prevent it. But does that conclusion follow? And is the premise true?
Smith Galer briskly gives us many reasons to accept the premise. The loss of a language, she writes, is the loss of “the wellspring of information that flows from it—its sayings, its means of identifying and perceiving the world, its character. Languages aren’t dictionaries, but encyclopedias, containing entire worlds of often irreplaceable information.” A compound called prostratin, she recounts, was discovered to be a potential treatment for H.I.V. by a scientist who knew Samoan well enough to have spoken to tribal healers in Samoa about the healing properties of a certain tree. Speakers of the Iwaidja language in northern Australia have long had different words for two distinct dolphin species; modern zoology wouldn’t have taken so long to recognize their distinctness if scientists had, Smith Galer writes, “bothered to speak to the Iwaidja people, or learn their language.” There are also documented links, she tells us, between a language’s health and the mental health of the communities that speak it. Communities in Canada that could “hold a conversation in their Aboriginal language experienced low to zero youth suicide rates. In communities where fewer than half could speak their language, suicide rates were six times higher.”
In general, Smith Galer’s empirical examples tend toward the optimistic. The clinical value of prostratin, for one, has yet to be established. And if languages can encode traditional knowledge they can also encode and perpetuate traditional error. We need no exotic examples to see that the folk category of “fish” (in English as in many other languages)—as something that, understandably enough, included whales, dolphins, and porpoises—was genuinely improved upon by the scientific term “mammal” and its associated taxonomies. Nor is factual error the only thing a language can preserve. Languages also encode values—gender hierarchies baked into pronouns, honorifics that perpetuate caste distinctions—that liberal-minded activists may not be eager to preserve as valuable pieces of linguistic heritage.
One may wonder, too, about those vaunted correlations between language health and a community’s psychological well-being. Isn’t it at least as plausible that some more basic social or economic fact explains the health both of the language and of the community? If that’s true, we shouldn’t be so confident that communities will be happier if only we enrolled their members in language lessons.
Even if we grant Smith Galer the premise that language death is always a terrible thing, it doesn’t follow that we ought to do everything in our power to stop it. That’s an extremely demanding standard that we adopt in no other policy domain. The death of human beings is also a very bad thing, but all health systems have to make hard choices about how they use their finite resources. There is no reason for language death to be an exception to the general principle. Even where the death of a language would be acknowledged as a tragedy, what we should do—what costs we should be willing to bear—in trying to stop it must remain an open question.
The interventions that are easiest to get behind are the ones aimed at the lowest-hanging fruit. There is a great deal that states and other coercive institutions could simply stop doing: they could stop banning the teaching of Kurdish in schools that want and have the resources to teach it; they could stop rejecting the names Kurds want for their children because they are not on the list of approved ones; they could stop denying Ukrainian defendants interpreters in court.
Individuals, too, may find that they have more power than they realized. Immigrants might continue to speak their native languages to their children without worrying that it might affect their acquisition of their new country’s dominant tongue. (The current research suggests that bilingualism poses no threat to speech development.) Where a school offers the heritage language of your family, consider allowing your children to study it. Maybe, other things being equal, choose a school for your child that offers a program in your native language.
What happens when other things are not equal? When it’s a matter not of killing a language but of striving, officiously, to keep it alive, even though it’s no longer being reliably transmitted at home? Smith Galer praises a master-apprentice program in the U.S. that partners younger adult learners of Karuk (an Indigenous language from northwestern California) with one of the very few fluent speakers and pays the learners a fifteen-dollar-an-hour stipend for their participation. She deplores the fact that the American federal government has earmarked only a hundred and eighty million dollars for language revitalization since 2005—a relatively paltry sum. Still, once numbers enter the discussion, as they must, there are uncomfortable questions that can’t be evaded.
David Crystal refers to a 1992 controversy about whether it was appropriate for the Academy of Mayan Languages, in Guatemala, to devote resources to a reforestation project. Shouldn’t we take more seriously the claim, made by defenders of the decision, that improving the community’s economic security was the most effective route to saving the language? Crystal mentions this episode as an aside in “Language Death,” but anyone who hopes to move from plaintive elegy to policy proposal must make such questions central. If saving a language takes money, how should that money be spent? Perhaps the Karuk people would prefer to use their own judgment to decide whether they want to spend it on language-learning programs or on direct support. That argument is sometimes made cynically, by people who’d rather no money were given to anyone for any reason at all. But these are real questions that organizations working in the field have to face.
Passionate preservationists—whether they hope to save a rare species of newt, or a beloved old railway station threatened with demolition, or coal-mining jobs in a Rust Belt town—share with campaigners for endangered languages a habit of presenting their case as if it were a contest between a real value on their side and nothing at all on the other. Preserve this language (or species or building or mine) or else surrender to indifference, cynicism, and nihilism. But is that ever the choice we face?
Every dollar dedicated to a language-revitalization program could, in any realistic budget, go toward something else: a school, or a hospital, or even another language program. The language program may well be of real value. But the value of the hospital or the school we might instead build is also real. To pretend otherwise, and to accuse everyone who points to these difficult trade-offs of bad faith, is to duck a confrontation with the deepest problems of politics and policy.
Isaiah Berlin famously observed that there is no social world without loss. We want and value many things that cannot be had at the same time. Say we want people to be free and also to be equal and to live in happy, unalienated communities. What if some people’s exercise of freedom renders things less equal, or less happy and more alienated? Berlin thought it a contemptible sleight of hand to insist that such freedom was not “true freedom.” To hope to honor all our values, he believed, is not simply overoptimistic but incoherent. What makes some political choices tragic is not that we have failed to think hard enough about how to have it all but that there is, in fact, no way to have it all. Politics, then, is the always unsatisfying project of managing loss; at its best, it consists of doing this without self-deception.
Berlin’s thesis lands with special poignancy when we consider the question of language loss. When French people today complain about the diminished international prestige of their nation and language, and wince when their teens talk about “e-mail” rather than the approved “courriel,” they might remember that their ideal de-Englished French was itself a construction built on the graves of the dozens of languages and regional forms of speech which dominated much of rural France well into the nineteenth century. The historian Eugen Weber wrote influentially of how peasants who spoke languages like Occitan, Breton, or Basque became—or, more honestly, were made—Frenchmen. Weber’s account does not play down the violence and humiliation of the process. Smith Galer herself notes that, as late as 1949, one could find children wearing slates around their necks with the degrading confession: “Je parle Breton” (“I speak Breton”). These were not cases of languages being allowed to die a natural death.
But Weber’s account is also a reminder that the project of linguistic homogenization had a republican, and not merely an administrative, justification. During the Revolutionary period, Abbé Grégoire held that a true republic could not flourish if its citizens could not understand one another; in the Third Republic, the push for universal primary education in French explicitly framed it as the condition for civic participation. Kings may have wanted a single language for the convenience of their scribes and tax collectors, but republicans wanted it for the sake of the citizens.
The French republican project did, in fact, succeed in creating a public sphere where a farmer from Brittany and a lawyer from Provence might read the same newspaper, serve on the same jury, and vote in the same election, without some Parisian smart-ass claiming not to understand what they were saying. What was erected on the site of that linguicide could be something of real value. A similar story might be told of Italy after unification, or of the ambitious effort to find a lingua franca to connect the vast and diverse islands of Indonesia without giving the country’s dominant Javanese even more power than they already had.
There are, of course, other models of languages coexisting, with varying levels of amity. The Flemish-French divide in Belgium has not exploded in tumult but has nevertheless left the country in a state of institutional paralysis. French thrives in Quebec, supported by some extremely assertive language legislation, but the threats of secession in the province are a persistent nuisance. And in India, whose constitution recognizes twenty-two languages, the ill-starred attempt in the nineteen-sixties to make Hindi the sole language of the union foundered in the face of violent resistance from speakers of southern languages such as Tamil, leaving the country with a “three-language formula” governing its school system.
The Indian case speaks to both the possibility of a genuinely multilingual society and its risks. Linguistic diversity is not a good acquired at no expense to the societies that have chosen it (or been forced to choose it) over homogeneity. Indian multilingualism, no less than French or Italian homogeneity, is a ledger with its own column of costs—communal tensions, constant incomprehension—to be weighed against its undoubted benefits.
If India has managed to avoid a single national language, there has been a considerable amount of linguistic consolidation within language groups: what’s called Hindi is really a bureaucratic attempt to unify a quite disparate group of languages and dialects spoken across the north. Tamil itself is no less a political construction, with public discourse and education carried out in a self-consciously archaic form of the language only half-intelligible to many speakers of its multifarious vernaculars. (My own first language, notionally a dialect of Tamil, takes a good deal of its vocabulary from Malayalam; my attempts to speak the standard forms of both languages nearly always occasion laughter.)
Not even those linguistic underdogs Kurdish and Ukrainian are exceptions to this pattern. In Syria and Turkey, the standard form of Kurdish is Kurmanji, written in Latin script; in Iran and Iraq, the standard form of Kurdish is Sorani, written in an Arabic-based script. Both standardizations have tended to push aside smaller, related tongues such as Gorani, which had been a literary language in its own right, and Zazaki, whose classification as a variant of Kurdish or as a separate northwest-Iranian language remains contested. Ukrainian, meanwhile, was built on the prestige of the Poltava-Kyiv tongue of the Middle Dnipro region; Rusyn, which enjoys protected minority-language status in Slovakia and Poland, is treated within Ukraine as a mere dialect. Kichwa trails a similar history. Spoken today by hundreds of thousands, it descends from the Quechuan language carried north by Inca imperial expansion. In southern Ecuador, the Cañari were conquered by Quechuan speakers, and their own language went extinct. We can condemn the perpetrators of linguicide without assuming that their victims must themselves be blameless innocents. It’s graveyards all the way down.
What would a preservationist victory actually look like? It may look like contemporary Wales. Some six decades ago, the writer and politician Saunders Lewis was predicting that the Welsh language would be dead by the year 2000. Instead, it is now an official language of Wales, taught throughout the public-school system, broadcast on its own television channel, used in courts, in the national assembly, on road signs, and in public documents as a matter of routine. It is spoken, to a greater or lesser degree of fluency and often mixed with bits of English, by around a sixth of the population of Wales. The Welsh government hopes to see the number who speak it almost double, so that the language will have a million speakers by 2050.
The health of Welsh has much to do with the fact that it is being transmitted, particularly in north Wales, by parents to children in the ordinary, unself-conscious way one associates with any living vernacular. Welsh is still not, in the permanent sense, “safe”; English exerts a constant pressure. But it is a good deal better off than Scottish Gaelic. Fluent speakers of Scottish Gaelic number at around seventy thousand at most, and the great majority of those who use it as their vernacular are elderly. Intergenerational transmission in the home is now the exception, even in the traditional Gaelic heartlands of the Western Isles. What competence there is among the young appears to be almost entirely the product of institutional intervention. It is a substantial achievement of Gaelic’s defenders that there are children who learn Gaelic at a Gaelic-language school in Glasgow or Edinburgh, from teachers who themselves learned the language as adults—all in cities where Gaelic is spoken by almost nobody outside the classroom. But this is a long way away from even the precarious health that Welsh now enjoys.
One cannot avoid the uncomfortable image of a wildlife sanctuary, where endangered animals live in some barely adequate simulacrum of a wild existence but are constantly monitored for the disease or the excessive predation that might finish them off. Keeping a species alive even under these conditions is, again, an achievement. Yet it is a lesser order of achievement than that much rarer case, where the thing comes back not as a museum exhibit but as a living population reproducing itself unassisted—like the gray wolves reintroduced into Yellowstone.
The one case of revitalization that comes anywhere close to the Yellowstone analogy is Hebrew. Smith Galer’s account of its revival makes the scale of the accomplishment clear: a language with few, if any, native speakers for the better part of two thousand years, heard mostly in synagogue and study hall, was deliberately and programmatically turned, by a generation of inspired nation-builders, into the first language of a modern state, Israel. And somehow, the process managed to produce a nation where children curse, dream, and throw tantrums in Hebrew without anyone feeling for a moment as if they were trapped in a historical reënactment.
Hebrew has been contemplated with some envy by campaigners in other societies. There are, for instance, still Indian nationalists who fantasize about a Sanskrit revived as the common language of India; perhaps there are people in Cornwall or the Isle of Man wondering if, with the right will (and pots of money), we might not see a Cornish or Manx revival to match it. We won’t. The success of Hebrew is obviously due to a convergence of circumstances historically unprecedented and unrepeatable. And is there a linguistic community that would want the experience of the European Jewry reprised, even if it would mean a revitalized language? And, if Hebrew now thrives, what of Yiddish? Or Ladino, the Iberian-descended language of the Sephardic Jews, once widely spoken in Thessaloníki?
One of the historical fallacies that an honest preservationist must avoid is presenting the conflict as one between an eternal, authentic linguistic heritage and a rootless, contingent modernity. For surely the heritage was once the rootless, contingent upstart; we forget this fact simply because it won. Were there not earlier languages in western Britain supplanted or absorbed by Welsh or its Brittonic precursors, not unlike the better-attested displacement of Occitan by French? Everything old, after all, used to be new.
To acknowledge this contingency—as wildlife conservationists must acknowledge that the fossil record is full of creatures that went extinct long before there were any humans around—doesn’t mean abandoning the worthy instincts that animate Smith Galer’s book and others in its genre. We can sympathize with her when she writes that “of the rich seven-thousand-plus body of thoughts, feelings, vocabularies, and grammar systems that exist today, well over three thousand languages will go. No funerals will be held; no cemeteries will fill. Some languages will leave behind dictionaries, novels, hours of radio; others will leave nothing save for a relative’s turn of phrase someone still remembers with affection.” And yet the language of bereavement has its limits.
What are the dying really owed? It rather depends on whether we are thinking of the old man lying peacefully on his deathbed after a long life well lived or the child run over by a drunk driver. If the latter fills us with rage and despair, the former need not. We don’t have to conclude that the old man was failed by medicine, or that there was a conspiracy to kill him. Nor need we demand that his death must be prevented at any cost, or that he be kept alive on a respirator for another decade with no thought to what it would exact from everyone around him. But to say that he should be allowed to die does not mean we shouldn’t grieve his loss. He is owed comfort, dignity, and the assurance that he won’t be forgotten by those who loved him, that he will be genuinely mourned.
Freud once helpfully distinguished that healthy thing, mourning, from melancholia—the mourning that never ends, grief that is unable to let the lost object go. The literature of linguistic preservation is a literature of mourning when it calls for languages to be documented, honored, and, in the rare instance where a revival might be a real option, revived. At its most culpably self-deceiving, it comes closer to melancholia: a refusal to recognize when the time for salvation is past and the most we can hope for is salvage. In acknowledging that not all languages can or should be rescued, we do not love the dying languages less. In a world where resources are scarce and values conflict, it may be the only way to love them wisely. ♦