language loss essay

Speaking Truth to Israel Requires More Than Academic Freedom

Payangko, or Echidna ( Zaglossus attenboroughi )

Inside Amazon’s Union-Busting Tactics

Fighting for Reproductive Rights in Retirement

Can Ancient DNA Support Indigenous Histories?

Can Embracing Copies Help With Museum Restitution Cases?

Can Art Save the “Post-Apocalyptic” Salton Sea?

How Allocating Work Aided Our Evolutionary Success

Bringing Back the World’s Most Endangered Cat

The Shortcomings of Height in Politics

Griko’s Poetic Whisper

When a Message App Became Evidence of Terrorism

The Rise of Aunties in Pakistani Politics

For Families of Missing Loved Ones, Forensic Investigations Don’t Always Bring Closure

Albania’s Waste Collectors and the Fight for Dignity

Grappling With Guilt Inside a System of Structural Violence

Inside Russia’s Campaign to Steal and Indoctrinate Ukrainian Children

Coastal Eden

On the Tracks to Translating Indigenous Knowledge

A Call for Anthropological Poems of Resistance, Refusal, and Wayfinding

Buried in the Shadows, Ireland’s Unconsecrated Dead

Nameless Woman

A Palestinian Family’s History—Told Through Olive Trees

Can “Made in China” Become a Beacon of Sustainability?

The International Order Is Failing to Protect Palestinian Cultural Heritage

Spotlighting Black Women’s Mental Health Struggles

Being a “Good Man” in a Time of Climate Catastrophe

Cultivating Modern Farms Using Ancient Lessons

Imphal as a Pond

A Freediver Finds Belonging Without Breath

The Trauma Mantras

Why Are Languages Worth Preserving?

I met the last speaker of Naati on an empty stretch of beach on Malekula, an island in the South Pacific nation of Vanuatu. I had been hiking for hours along narrow paths through the hot, dense forest, wading across the occasional waist-high stream with my pack of recording equipment hoisted overhead. As I dropped my pack on the sand, a figure descended from the nearby cliffs and crossed the beach toward me.

We exchanged greetings in the local creole, and the conversation quickly turned to the topic of my unlikely appearance on these shores. I told the man, Ariep, that I was in the country to study one of its many Indigenous languages. When he learned I was a linguist, he excitedly shared that he speaks Naati.

Plunging several sticks into the sand and using them as reference points, Ariep explained the relationship between Naati and the other languages of the area. With a mix of pride and sorrow, he revealed that he is the last fluent speaker of Naati. Although a few of his family members have some knowledge of the language and make an effort to use it together, he fears that with his death, Naati will soon disappear.

Naati’s predicament is not unique. Of the roughly 7,000 languages spoken on the planet today, 50 to 90 percent are considered vulnerable to extinction by the end of the century.

Language is the cultural glue that binds communities together.

The crisis has received increasing public attention over the past decade, punctuated by lines such as “one language dies every two weeks” and illustrated by poignant tales of the death of a last speaker. In this UNESCO International Year of Indigenous Languages , as alarm bells sound and preservation efforts are celebrated, we should pause to ask: Why does it matter?

Should Naati’s fate concern the world? Ariep does not need Naati to communicate. Like many speakers of endangered languages, he is fluent in an impressive number of languages, including several native to his island, as well as the national language.

If we are heading toward a future in which we all speak one of a few large languages, isn’t that a good thing? Couldn’t it be a way to facilitate communication and level the playing field across nations? Is the desire to “save” these small languages purely sentimental—a romantic notion fostered by scholars in ivory towers of isolated peoples untouched by the exhausting rush toward globalization?

I argue “no.” As a linguist who has worked with endangered language communities in Canada and the Asia-Pacific, I know that language loss is a critical and urgent problem—not only for the speakers who lose their languages, but for everyone. Languages are a vital source of culture and identity for individual communities, and for the global community, languages are an invaluable source of information about human cognition. A linguistically diverse world benefits us all.

Consider what has happened to people whose language has been forcibly taken from them, supplanted by one of the larger, ostensibly more useful languages. This scenario has played out countless times across centuries at the hands of colonial powers or as a tool of national governments to suppress minority groups. It occurs around the world today in classrooms where children are punished or humiliated for using languages and dialects that deviate from an accepted standard.

The response of these communities has not been to celebrate the subsequent generations who speak English, Spanish, Swahili, or whatever the language of power might be. Rather, they decry this cultural genocide and, where possible, fight back against the theft of their linguistic heritage.

In Canada during the 19th and 20th centuries, the national government oppressed Indigenous people in part by removing children from their families and placing them in residential schools . In these spaces, children suffered a range of physical and mental abuses, including punishment for speaking their languages . These injustices severely disrupted the transmission of dozens of Indigenous languages, the majority of which are now endangered.

Today, despite a scarcity of resources to address numerous challenges after decades of persecution, Indigenous Canadian communities are making huge investments in reclaiming their languages. From the “language nest” in Tyendinaga Mohawk Territory in Ontario, where children are exposed only to Mohawk throughout the day, to the Nehiyawak language and culture camps in Saskatchewan, where families learn and share their Cree heritage, Indigenous language education across Canada is flourishing.

It would seem easier, cheaper, and infinitely more practical just to accept English (a language that is no less highly desirable internationally) and shift the resources elsewhere. The fact that people struggle to reclaim their languages despite the obstacles says something crucial about the value of language and the tragedy of loss.

Language is the cultural glue that binds communities together. Language loss is a loss of community heritage—from histories and ancestral lineages known only through oral storytelling, to knowledge of plants and practices codified through words unwritten and untranslated.

Lulamogi speakers in Uganda, for example, worry that as people forget the dozens of terms that describe methods of trapping and eating white ants—such as okukunia , okutegerera , and okubuutira —they will forget this important cultural practice. Also at risk are the phrases and associated customs for welcoming the agricultural seasons and washing the bodies of the dead.

In the words of Lulamogi language advocate Nabeeta Erusaniah: “It is like when a wall of a hut collapses, the ceiling does not remain standing. What keeps the social practices and a ritual standing is the language. Kill the language, and the shelter collapses too.”

Language loss is also a loss of community identity, collective purpose, and self-determination. While harder to quantify, such losses have real, detrimental effects on health and quality of life. Conversely, the ability of community members to speak their Indigenous language together enhances well-being.

In British Columbia, youth suicide rates are more than six times lower in Indigenous communities where at least 50 percent of the population speaks the native language. In Aboriginal and Torres Strait Islander communities of Australia, young people who speak an Indigenous language have lower rates of binge drinking and illegal drug use compared to nonspeakers, as well as a decreased chance of becoming victims of violence.

The disappearance of a language may seem like an unfortunate loss only to the people involved. However, the impact for all of us is real and substantial.

This impact goes beyond the losses of particular bits of information, like Indigenous names for medicinal plants yet to be classified by scientists outside a community, or concepts and worldviews reflected in the words and structures of one language that do not have parallels in another. Understanding language is vital to understanding human cognition. Each language is a piece of the puzzle that we need in order to determine how language works in the mind. With each missing piece, we are further from seeing the full picture.

Analyzing language patterns has real implications for our lives.

Languages may appear to differ wildly from one another, but they are all variations on a theme. Like a field of flowers, the individual plants may vary in height and color, but they all have stems and petals.

Your language may have “tall trees” or “trees tall.” It may ask, “Where is the dog?” or “The dog is where?” It may thank you in one syllable or in many. Regardless, whether your language is spoken or signed, it draws on a limited set of forms and structures, and it uses them in consistent and predictable ways.

The remarkable similarities across languages suggest that there is some cognitive capacity that underlies all human language, directing how language develops and setting the boundaries for what is possible. The goal of contemporary linguistics is to describe and model this system—in essence, to figure out how language works.

For example, languages contrast greatly in their number of consonants, from the six in the Papuan language Rotokas to the 122 in the Southern African language ǃXóõ. There are enough commonalities among sound systems, however, that if linguists know your language has 20 consonants, we can make a fairly good guess as to what many are likely to be, and we can be almost certain of others that would not occur. In terms of sentence structure, all languages use the three basic elements subject, object, and verb. Although these can be ordered in different ways, about 80 percent of known languages put the subject first, while only about 1 percent put the object first.

Analyzing these patterns is far from an esoteric academic exercise; it has real implications for our lives. The more we understand about how language functions, the better equipped we are to improve our therapies for communication disorders and our methods for language teaching.

This knowledge contributes to technological innovation as well. Research on sound patterns is used in creating speech synthesis software, while models of grammatical structure aid in developing linguistic components for artificial intelligence.

Understanding language in turn gives us a window into cognition. Observations about the strikingly similar ways that children acquire language, across languages and cultures, provide insight into how the brain develops. Psycholinguistic experiments involving language production, comprehension, and recall tasks reveal clues to how the mind organizes information.

Early models of grammar were based primarily on a few large, mostly European, languages that Western scholars knew or could easily access. Imagine the deficiencies if the research stopped there. It would be like basing an understanding of plants on a neighborhood vegetable garden or of animals on a trip to a petting zoo.

Recent reports on gender bias in medical testing have revealed that therapies tested on men do not necessarily work for women. Studies of racial bias in the tech industry have shown that applications such as facial recognition, which were trained on images of white people, do not necessarily work for people of color. When it comes to language, what if our models are proven incorrect by a previously undocumented cluster of languages in the Amazon? A theory of human language must account for the language of all humans.

However, taking into account all languages, or even a representative sample, is a huge challenge. Thousands of languages are undocumented or only very poorly described, and no one—neither linguists nor speakers—understand how they work.

Documenting a language thoroughly is a major undertaking involving years of collaboration between the members of a speech community and linguists (who may or may not be speakers themselves). Given the rapid rate of language loss in the world today, many languages are in danger of disappearing before they have been documented, taking with them irreplaceable information about human cognition.

The very limited documentation we have of Naati reveals that the language has a sound called a “bilabial trill.” These trills were once considered impossible speech sounds, but now linguists know that they are common in the languages of Malekula.

As I watched Ariep turn back toward the cliffs that day on the beach, taking with him a wealth of linguistic and cultural knowledge, I wondered, Does Naati contain other features that could challenge our understanding of language?

What can the many undocumented languages teach us about language structure and cognition, about the richness of our cultures and traditions, about our very humanity? For the sake of the speakers of endangered languages, for the sake of us all, we must preserve the world’s languages as we search for answers and work to ensure linguistic diversity for generations to come.

Anastasia Riehl is a linguist with a Ph.D. from Cornell University. She is currently the director of the Strathy Language Unit at Queen’s University in Kingston, Ontario, Canada. Her work with endangered languages has taken her to Indonesia and Vanuatu, as well as to the city of Toronto, where she directs a project to document the endangered heritage languages of immigrant communities.

Stay connected

Facebook , Instagram , LinkedIn , Threads , Twitter , Mastodon , Flipboard

Y ou may republish this article, either online and/or in print, under the Creative Commons CC BY-ND 4.0 license. We ask that you follow these simple guidelines to comply with the requirements of the license.

I n short, you may not make edits beyond minor stylistic changes, and you must credit the author and note that the article was originally published on SAPIENS.

A ccompanying photos are not included in any republishing agreement; requests to republish photos must be made directly to the copyright holder.

We’re glad you enjoyed the article! Want to republish it?

This article is currently copyrighted to SAPIENS and the author. But, we love to spread anthropology around the internet and beyond. Please send your republication request via email to editor•sapiens.org.

Accompanying photos are not included in any republishing agreement; requests to republish photos must be made directly to the copyright holder.

Become a Supporting Member

For 15 years, Longreads has published and curated the best longform writing on the web—and we wouldn’t exist without supporters like you. Give today and ensure that quality journalism continues to flourish.

Thank you for your contribution!

The Best of the Web—in Your Inbox

Every day we scour the internet for the best longform writing, and every day we send you our editors' picks. Join 100,000 newsletter subscribers—and don't miss that story everyone is talking about.

• Daily Updates
• Weekly Top 5

Join Longreads today!

Register with Longreads for free and get access to our editors' picks collecting the best stories on the web, as well as our award-winning original writing.

Newsletters

Our privacy policy can be found here.

Thank you for registering!

An account was already registered with this email. Please check your inbox for an authentication link.

Longreads : The best longform stories on the web

Disappearing Language: A Reading List on Losing Your Native Tongue

Share this:.

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)
• Click to share on Tumblr (Opens in new window)
• Click to share on Pocket (Opens in new window)
• Click to email a link to a friend (Opens in new window)
• Click to share on Mastodon (Opens in new window)

This story was funded by our members. Join Longreads and help us to support more writers.

By Pardeep Toor 

English was the first language my newborn heard after his birth in October 2021, probably something medical the midwife said, or congratulations from a nearby nurse. My wife and I were speechless, focusing only on our son’s blue skin, piercing screams, and block of black hair that overwhelmingly confirmed he was indeed ours.

As first-time parents, our son instantly became the exclusive lens through which we viewed our world. I should’ve gotten new windshield wipers to make sure we reached home safely. Next time, I will take my elevated bilirubin levels seriously. My wife swore to get the spot on her retina checked again. She couldn’t remember the last time we’d dusted underneath our bed. We’d prepared nine months for this shift, but it still shook us. Our lives were only necessary to sustain his life.

We couldn’t stay speechless for long. The nurses eventually checked on us less frequently. Poking and prodding clinicians dissipated, leaving us with the humming overhead tube lights and beeping in the hallway. It was our turn to talk to our son, but what would we say? 

English wasn’t the first language for either of us. My wife is a native Spanish speaker and I exclusively spoke Punjabi for the first six years of my life. We both acquired English through our respective educations in Colombia and Canada. We promised to give our son both our languages, despite failing to acquire them from each other. 

English is our essential language, a primary means of communication that allows us to thrive as a couple, while simultaneously pulling us away from our native languages and cultures. Each spoken syllable of English is a leap away from our rolling “Rs” in Punjabi and Spanish. I’m not demonizing English, rather recognizing the challenge of its dominance in our lives. 

In the past four months, we’ve obsessed over speaking our respective languages to our son. It’s turned into a game: My wife will say something to him in Spanish, I’ll ask what she said, she tells me in English, and then I translate it into Punjabi and say it louder and faster back to our son as if his comprehension is a race we’re each trying to win. It’s partly in fun — but also stems from a sincere apprehension. We’re trying to pass down our languages while preserving them in our own lives.   

Regardless of our efforts, English will inevitably become the common language that my wife and I share with our son. It’s the only way we can talk to him without isolating each other. In doing so I’m afraid I’ll continue to lose my native Punjabi and our son will forever lose something he could have had. 

This struggle isn’t exclusive to our family. The loss of language has been extensively explored in the following essays. 

The Pain of Losing Your First Language (Kristin Wong, Catapult , December 2021) 

This essay outlines the suffering Wong endured since foregoing her native language as a child for the sake of assimilating into America. Wong does a phenomenal job of incorporating linguistic research in her analysis, giving academic weight to her regrets. The balance of personal narrative and analysis of English language adoption and native attrition flows through the essay as studies confirm Wong’s feelings, yet don’t free her from longing for her first language. “I wonder what Cantonese words my brain pushed out when I started speaking mostly English at age six. And is attrition limited to words? What else did I lose to assimilate?”

During her pregnancy, Wong commits to re-learning her native Cantonese so she can pass it on to her child. However, despite her efforts, she feels the impossibility of her task. 

Like learning how to spell only, the more I look at Jyutping, the more the words start to make sense. But part of me knows better. I’ll never speak Cantonese the same way I would have if I’d never stopped speaking it to begin with. Like a phantom limb, the memory of my first language stays with me even with it gone, but that’s all it is: a memory. It occurs to me that trying to relearn this language is the embodiment of my bicultural identity. The American in me is determined to reclaim the Chinese part of myself.

Why Do I Write in My Colonizers’ Language? (Anandi Mishra, Electric Literature, March 2021)

Part of the struggle with English has always been its foreignness. The language either forcefully invaded foreign lands or immigrants willingly chose to move west. Mishra addresses the colonial legacy of English and how that makes her feel “queasy,” while also recognizing the language’s modern dominance in India. 

For my family, friends, relatives, and teachers, English was seen as a language of access. It could land you better jobs, remove limitations, and open up avenues. English speakers were high achievers, often conflated with the colonizers who ruled over us for about 200 years. It was ironic that the language of our colonizers was seen as aspirational, something that could lift us out of the discomfort that our parents’ mid-level jobs put us through. In reading all the subjects at school in English, we were made to understand that English was the language of possibilities. 

Mishra reluctantly accepts the realities of the English language in India and her own life. But acceptance can also be an acknowledgment of adaptation — it’s not one or the other, English or Hindi, but hybridity that can hopefully be respectful to native languages and English’s injection into them. 

Translation as an Arithmetic of Loss (Ingrid Rojas Contreras, The Paris Review , June 2019)

Rojas Contreras opens this essay by acknowledging, “When you live between languages, the conversion of meaning is an arithmetic in loss.” Thoughts generated in one language come out awkwardly in English, or sometimes not at all. This ultimately leads to a feeling of “being understood sufficiently, rather than fully.”

This loss between languages catalyzed Rojas Contreras to write her debut novel, Fruit of the Drunken Tree , in English, even though she thought of it in Spanish. She constructed the sentences in Spanish in her mind but then immediately translated them into English on the page. 

Why didn’t you write the novel in Spanish? This is a question I get all the time. Language is one of the things you sacrifice when you migrate. I wanted to be true to the toll of that sacrifice by making visible what exactly was being lost.

My wife loves Rojas Contreras’s writing. She sees both her own Spanish and English in the syntax. It’s how she sees the world, in a constant state of translation from one language to another. By seeing Rojas Contreras’s translated language on the page, my wife sees her worldview being expressed as a reality and feels understood as an immigrant in the United States. 

Teaching Yourself Italian (Jhumpa Lahiri, The New Yorker, November 2015) 

Lahiri, an internationally renowned fiction writer, started writing from scratch in Italian to escape the personal weight of the English language. 

Why am I fleeing? What is pursuing me? Who wants to restrain me? The most obvious answer is the English language. But I think it’s not so much English in itself as everything the language has symbolized for me. For practically my whole life, English has represented a consuming struggle, a wrenching conflict, a continuous sense of failure that is the source of almost all my anxiety. It has represented a culture that had to be mastered, interpreted. I was afraid that it meant a break between me and my parents. English denotes a heavy, burdensome aspect of my past. I’m tired of it.

Lahiri opted not to re-engage with her native Bengali, which she spoke with an accent and admitted to not knowing how to read or write. She started anew with Italian, placing herself in a linguistic exile, far removed from her familial past in Bengali and professional life in English. Native languages, and the projected cultural expectations that come with them, can be as burdensome as chasing a distorted and romanticized memory or history. Lahiri’s fresh start in Italian vanquishes the obligation of chasing ghosts from her past and allows her to forge a novel identity in a new language. 

Ghosts (Vauhini Vara, The Believer , August 2021)

What happens when all languages fail? When you’re unable to express yourself no matter how many languages you can speak? That’s where the future comes in. Vara, unable to express grief after losing her sister, turned to an algorithm to complete her thoughts. 

I felt acutely that there was something illicit about what I was doing. When I carried my computer to bed, my husband muttered noises of disapproval. We both make our livings as writers, and technological capitalism has been exerting a slow suffocation on our craft. A machine capable of doing what we do, at a fraction of the cost, feels like a threat. Yet I found myself irresistibly attracted to GPT-3—to the way it offered, without judgment, to deliver words to a writer who has found herself at a loss for them.

What followed was an experiment in human-computer interactions. Vara feeds words into Generative Pre-Trained Transformer 3 (GPT-3)  and the remaining text is predicted, with the details Vara provides about her sister determining how the narrative ends — technology offering a borderless universal language with infinite memory.

The nine stories completed by artificial intelligence in this piece are something new, void of human attachment. Culture in an algorithm. Is this how people will one day express themselves and understand their upbringing? Then what remains of the cultural nuances embedded in our native languages? Are one or two or three languages enough for our son? Are they the right languages? What if they all fail him? The fear of losing our language and culture to algorithms and English inspires us to transmit what we have left amidst our loss. 

Pardeep Toor ‘s writing has appeared in the Best Debut Short Stories 2021: The PEN America Dau Prize, Catapult, Electric Literature, Southern Humanities Review, Midwest Review, and Great River Review.

Editor: Carolyn Wells 

Support Longreads

By clicking submit, you agree to share your email address with the site owner and Mailchimp to receive marketing, updates, and other emails from the site owner. Use the unsubscribe link in those emails to opt out at any time.

We've recently sent you an authentication link. Please, check your inbox!

Sign in with a password below, or sign in using your email .

Get a code sent to your email to sign in, or sign in using a password .

Enter the code you received via email to sign in, or sign in using a password .

Subscribe to our newsletters:

Sign in with your email

Lost your password?

Try a different email

Send another code

Sign in with a password

Four Things That Happen When a Language Dies

This World Mother Language Day, read about why many say we should be fighting to preserve linguistic diversity

Kat Eschner

Languages around the world are dying, and dying fast. Today is International Mother Language Day , started by UNESCO to promote the world's linguistic diversity.

The grimmest predictions have 90 percent of the world's languages dying out by the end of this century. Although this might not seem important in the day-to-day life of an English speaker with no personal ties to the culture in which they’re spoken, language loss matters. Here’s what we all lose:

1.  We lose “The expression of a unique vision of what it means to be human”

That’s what academic David Crystal told Paroma Basu for National Geographic in 2009. Basu was writing about India, a country with hundreds of languages , at least seven major language families and rapid language loss.   

The effects of that language loss could be “culturally devastating,” Basu wrote. “Each language is a key that can unlock local knowledge about medicinal secrets, ecological wisdom, weather and climate patterns, spiritual attitudes and artistic and mythological histories.”

Languages have naturally risen and fallen in prominence throughout history, she wrote. What makes this different in India as well as throughout the world is the rate at which it’s happening and the number of languages disappearing.

2. We lose memory of the planet’s many histories and cultures.

The official language of Greenland, wrote Kate Yoder for Grist , is fascinating and unique. It’s “made up of extremely long words that can be customized to any occasion,” she writes. And there are as many of those words as there are sentences in English, one linguist who specializes in Greenlandic told her. Some of those, like words for different kinds of wind, are disappearing before linguists get the chance to explore them. And that disappearance has broader implications for the understanding of how humans process language, linguist Lenore Grenoble told Yoder. “There’s a lot we don’t know about how it works, or how the mind works when it does this,” she said.

Yoder’s article dealt with the effect of climate change on language loss. In sum: it hastens language loss as people migrate to more central, “safe” ground when their own land is threatened by intense storms, sea level rise, drought and other things caused by climate change. “When people settle in a new place, they begin a new life, complete with new surroundings, new traditions, and, yes, a new language,” she wrote.  

3. We lose some of the best local resources for combatting environmental threats

As Nancy Rivenburgh wrote for the International Association of Conference Interpreters, what’s happening with today’s language loss is actually quite different from anything that happened before. Languages in the past disappeared and were born anew, she writes, but “they did so in a state of what linguists call ‘linguistic equilibrium.’ In the last 500 years, however, the equilibrium that characterized much of human history is now gone. And the world’s dominant languages—or what are often called ‘metropolitan’ languages—are all now rapidly expanding at the expense of ‘peripheral’ indigenous languages. Those peripheral languages are not being replaced.”

That means that out of the around 7000 languages that most reputable sources estimate are spoken globally, only the top 100 are widely spoken. And it isn’t just our understanding of the human mind that’s impaired, she writes. In many places, indigenous languages and their speakers are rich sources of information about the world around them and the plants and animals in the area where they live. In a time of mass extinction, that knowledge is especially precious.

“Medical science loses potential cures,” she writes. “Resource planners and national governments lose accumulated wisdom regarding the management of marine and land resources in fragile ecosystems.”

4. Some people lose their mother tongue.

The real tragedy of all this might just be all of the people who find themselves unable to speak their first language, the language they learned how to describe the world in. Some find themselves in the unenviable position of being one of the few (or the only ) speakers of their mother tongue. And some, like many of Canada’s indigenous peoples, find their language in grave danger as the result of a campaign by government to stamp out their cultures.  

This loss is something beyond all the other losses, linguist John Lipski told Lisa Duchene for Penn State News: “Imagine being told you can’t use your language and you’ll see what that undefinable ‘more’ is,” he said.

What can you do about all this? Educate yourself, to start with. The Smithsonian's annual Mother Tongue Film Festival takes place every February in Washington, D.C. And projects like  National Geographic 's " Enduring Voices " are a great place to learn about endangered languages and their many speakers, and UNESCO's own website is another resource.  There's still hope for some of these languages if we pay attention. 

Get the latest stories in your inbox every weekday.

Kat Eschner | | READ MORE

Kat Eschner is a freelance science and culture journalist based in Toronto.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 16 December 2021

Global predictors of language endangerment and the future of linguistic diversity

• Lindell Bromham   ORCID: orcid.org/0000-0003-2202-2609 1   na1 ,
• Russell Dinnage 1 ,
• Hedvig Skirgård 2 , 3 ,
• Andrew Ritchie 1 ,
• Marcel Cardillo 1 ,
• Felicity Meakins 4 ,
• Simon Greenhill   ORCID: orcid.org/0000-0001-7832-6156 2 , 3 &
• Xia Hua   ORCID: orcid.org/0000-0003-3485-789X 1 , 5   na1  

Nature Ecology & Evolution volume  6 ,  pages 163–173 ( 2022 ) Cite this article

35k Accesses

51 Citations

879 Altmetric

Metrics details

• Language and linguistics
• Macroecology

A Publisher Correction to this article was published on 03 February 2022

This article has been updated

Language diversity is under threat. While each language is subject to specific social, demographic and political pressures, there may also be common threatening processes. We use an analysis of 6,511 spoken languages with 51 predictor variables spanning aspects of population, documentation, legal recognition, education policy, socioeconomic indicators and environmental features to show that, counter to common perception, contact with other languages per se is not a driver of language loss. However, greater road density, which may encourage population movement, is associated with increased endangerment. Higher average years of schooling is also associated with greater endangerment, evidence that formal education can contribute to loss of language diversity. Without intervention, language loss could triple within 40 years, with at least one language lost per month. To avoid the loss of over 1,500 languages by the end of the century, urgent investment is needed in language documentation, bilingual education programmes and other community-based programmes.

Similar content being viewed by others

The network nature of language endangerment hotspots

English language hegemony: retrospect and prospect

A revised digital edition of Wurm & Hattori’s Language Atlas of the Pacific Area

As with global biodiversity, the world’s language diversity is under threat. Of the approximately 7,000 documented languages, nearly half are considered endangered 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 . In comparison, around 40% of amphibian species, 25% of mammals and 14% of birds are currently threatened with extinction 9 . The processes of endangerment are ongoing 10 , with rates of loss estimated as equivalent to a language lost every one to three months 7 , 11 , 12 , and the most pessimistic predictions suggesting that 90% of the world’s languages will be lost within a century 13 . However, unlike biodiversity loss 14 , predictions of language loss have not been based on statistically rigorous analysis. Here we provide a global analysis to model patterns of current and future language endangerment, and compare the predictive power of variables representing some of the potential drivers of language loss. Our analysis has three key features. First, we examined a broader set of influences than previous studies, encompassing demographic factors, linguistic resources, socioeconomic setting, language ecology, connectivity, land use, environment, climate and biodiversity (Table 1 ). Second, we addressed major statistical challenges of large-scale comparative analyses, by simultaneously accounting for phylogenetic non-independence, spatial autocorrelation and covariation among variables. Third, our models incorporated demographic and environmental variables that can be projected into the future, allowing us to make predictions of future patterns of language endangerment in time and space.

While language change and shift are natural processes of human cultural evolution, the loss of global language diversity has been massively accelerated by colonization and globalization. Many factors contribute to language endangerment, some of which are specific to particular regions, language groups or languages. The historical context of each language, such as patterns of colonial expansion, and particular political climates, such as support for bilingual education, are expected to have substantial impacts on language endangerment patterns 10 . In addition to specific historical and local influences, there may also be widespread general factors that contribute to language endangerment, which can be used to identify languages that may come under increasing threat in the future. For a dataset containing 6,511 languages (over 90% of the world’s spoken languages), we analysed 51 predictor variables that target different aspects of language maintenance 15 , including language transmission (for example, whether a language is actively learned by children or used in education), language shift (for example, connectivity, urbanization, world languages) and language policy (for example, provision for minority language education, official language status). We also included variables that have been associated with language diversity, including features of climate and landscape. Clearly, any list of threatening processes will be incomplete, and the requirement for globally consistent data will fail to capture important influences on language vitality that operate at regional or local levels. Our aim is not to provide a comprehensive picture of language endangerment but a useful exploration of the influence of a selection of potential impacts. Broad-scale quantitative studies are therefore a complement to more focused qualitative studies on language endangerment and loss.

Understanding global threats to language diversity requires that we develop a macroecology of language endangerment and loss 16 . A macroecological approach has many advantages: it allows evaluation of a large range of factors that influence language vitality; formal testing of general patterns above the signal of individual language trajectories; statistical comparison of the explanatory power of different models, accounting for covariation of cultural, socioeconomic and environmental factors; and a way of avoiding the confounding effects of spatial distribution and relationships between languages 17 . Although threats to linguistic diversity, shaped by social, cultural, political and economic influences, often differ from processes threatening biodiversity 18 , the analytical challenges associated with studying global patterns of endangerment are common to biologists and linguists 17 , 19 , 20 , 21 . Here we use global analysis to illuminate some of the complex interactions of extrinsic factors threatening language diversity, and use this understanding to predict the fate of the world’s languages over the next century.

Results and discussion

Current patterns of endangerment.

We use an endangerment scale based on EGIDS, which incorporates a range of factors including domains of use and intergenerational transmission 22 , 23 . We describe languages that are losing first-language (L1) speakers as ‘Threatened’, those with only adult speakers and no child learners as ‘Endangered’ or those with only elderly speakers as ‘critically endangered’, and languages with no L1 speakers as ‘Sleeping’ (the term preferred by many speakers of endangered languages 1 , 24 , 25 : Supplementary Table 1 ). Of the 6,511 languages in our database, 37% are considered threatened or above (which we will refer to generally as ‘endangered languages’); 13% of these are no longer spoken (sleeping). The areas of greatest absolute number of endangered languages are in New Guinea, Central America, Himalayas and Yunnan, and regions between Central and West Africa (Extended Data Figs. 1 and 2 ), but this pattern may largely reflect diversity 17 : where there are more languages, speaker populations and geographic ranges tend to be smaller, potentially resulting in more endangered languages. Areas with the highest proportion of their languages endangered include Australia, North China, Siberia, North Africa and Arabia, North America, and parts of South America (Fig. 1 ). Areas with the greatest language loss to date are in Australia, South America and USA (Extended Data Fig. 2 ).

Each hexagon represents approximately 415,000 km 2 . The coloured bars show the predictors of level of endangerment identified in the best model for a global language database of 6,511 languages, and for each of 12 regions any additional influences on patterns of language endangerment (see Supplementary Data 3 ). Dark grey areas on the map do not have data for all the independent variables in the best model for language endangerment level. Language distribution data are from WLMS 16 ( worldgeodatasets.com ).

Predictors of language endangerment

Our analysis seeks the best set of variables, from 51 candidate variables, to explain variation in endangerment level (the dependent variable), over and above covariation due to relationships between languages, spatial autocorrelation and contact between language distributions, and allowing for interactions between predictor variables and region. We reduced the number of variables by grouping variables according to their pairwise correlations, identified independent variables with significant predictive power on a proportion of the data (training dataset), then evaluated the fit of the model on the remaining data (test dataset). We then estimated model parameters on the full dataset (see Methods for details).

Our best-fit model explains 34% of the variation in language endangerment (comparable to similar analyses on species endangerment 26 , 27 , 28 ). These variables cannot provide a full picture of the processes threatening language diversity, as there will be many other important factors that cannot be included due to lack of appropriate and consistent data with global coverage, or because of the idiosyncratic nature of processes of language endangerment and the influence of historical factors that cannot be captured in a broad-scale model. For example, patterns of human migration and past episodes of population expansion and contraction will not be captured fully in contemporary language distribution data. Furthermore, language endangerment and loss is an ongoing process, and there may be historical factors that caused dramatic reduction in L1 speakers that will not be captured in current values of socioeconomic variables, such as massacres of Indigenous populations or ethnic groups, punishing people for speaking their language and separating children from parents. Patterns of language endangerment may at least partially reflect past influences, such that current predictors might not fully capture important processes that resulted in the current endangerment status (a phenomenon known in conservation biology as extinction filter effect 29 ). Because of these unavoidable limitations, no study of this kind can aim to comprehensively describe factors affecting vitality of all of the world’s languages. But by identifying contemporary factors that are significant predictors of current patterns of endangerment at a global scale, we contribute to the understanding of the complex interaction of factors contributing to language endangerment.

Five predictors of language endangerment are consistently identified at global and regional scales: L1 speakers, bordering language richness, road density, years of schooling and the number of endangered languages in the immediate neighbourhood. Each of these predictors highlights a different process in language endangerment; taken together, they paint a picture of the way interactions between languages shape language vitality.

Number of first-language (L1) speakers is the greatest predictor of endangerment. It is important to emphasize that not all small languages are endangered, and that language loss does not necessarily result from a reduction in number of people in a particular culture or population, but often occurs when people shift from using their heritage language to a different language 1 , 30 . Therefore the multilingual setting in which each language is embedded (referred to as the language ecology) plays a key role in endangerment, by influencing whether speakers shift to another language or adopt additional languages in their multilingual repertoires 31 . Our results suggest that direct contact with neighbouring languages, as reflected in the number languages with overlapping or touching distributions, is not in itself a threatening process. In fact, languages whose distributions are directly in contact with a greater number of other autochthonous languages have lower average endangerment levels (Fig. 1 ). This may reflect a common observation that communities in regular contact with speakers of other Indigenous languages may be multilingual without necessarily giving up their L1 language 31 . If ongoing language contact was a threat to language vitality, then we might expect that more isolated languages, such as those on islands, would be less endangered, but this is not the case (Supplementary Fig. 7 ). Similarly, we find no evidence that barriers to human movement that might be expected to reduce contact between nearby speaker populations, such as steep or rough terrain, are associated with reduced endangerment. We conclude that being in regular contact with speakers of another language does not in itself usually endanger Indigenous language vitality. Instead there are other more complex social, economic and political dynamics influencing language endangerment that may co-occur with language contact but are not synonymous with it.

A language is more likely to be endangered if a higher proportion of languages in the region are also endangered, suggesting that, in addition to language-specific threats, there are also widespread factors that influence language vitality across a region. One such factor is the density of roads in the neighbourhood surrounding each language (Fig. 1 ). One interpretation of the association between road density and language endangerment is that roads increase human movement and thus bring people into contact with speakers of other languages, and this may result in language shift. However, our results suggest that the association between language endangerment and roads is unlikely to simply reflect language contact. If language contact always generated language shift and loss, then we would expect languages with a high degree of contact with other languages to be more endangered. In fact, we find the opposite: languages whose distribution overlaps or meets many other languages are less endangered (Fig. 1 and Supplementary Data 3 ). Furthermore, if contact between speakers of different languages was a driver of language loss, then we would expect landscapes that inhibit movement to reduce language contact and show lower levels of endangerment, but none of the other connectivity variables, such as altitudinal range, landscape roughness or density of waterways, show consistent association with language endangerment. The association with roads is neither simply a result of socioeconomic shift, as other indicators of development (for example, GDP, life expectancy) are not associated with language endangerment, nor is it a reflection of increasing urbanization, land use change or increase in built environment (Supplementary Fig. 7 ). Instead, road density may reflect connectivity between previously remote communities and larger towns, with increase in the influence of commerce and centralized government. Lack of roads has been cited as a protective factor in maintaining Indigenous language vitality, as it may limit the spread of ‘lingua francas’, such as Tok Pisin in Papua New Guinea 32 . The association between road density and language endangerment may reflect movement of people in two directions, as people move from their traditional homelands into larger population centres, and outsiders move into previously isolated communities, both of which have been implicated as threats to Indigenous language vitality 33 . For example, access to new employment opportunities (such as a shift from rural work to factory or construction work) may result in shift away from heritage languages to dominant languages of commerce 34 , 35 , 36 . Roads can aid the spread of ‘lingua francas’ or languages of central governance 37 .

There is consistent global support for higher average levels of schooling being associated with greater language endangerment (Fig. 1 ). The association between schooling and language endangerment cannot be interpreted as a side effect of growing socioeconomic development, because years of schooling is a much stronger predictor of endangerment patterns than other socioeconomic indicators. Instead, it is consistent with a growing number of studies showing a negative impact of formal schooling on minority language vitality, particularly where bilingual education is not supported or, in some cases, is actively discouraged 38 , 39 , 40 . Yet having a minority education policy is not globally associated with reduced threats to language diversity, possibly due to variation in the extent and manner of provision of bilingual education for speakers of minority languages. For example, the Bilingual Education Act of the United States (1968) was primarily concerned with improving access to mainstream education for students from non-English speaking backgrounds by using heritage language as a bridge to English acquisition, rather than being designed to allow students to maintain their first language 41 .

The spatial scale of the variables reflecting education policy and outcomes cannot capture variation within countries. Reliable statistics on average years of schooling are, for most parts of the world, only available as national averages, even though years of schooling may vary within a country, particularly between socioeconomic groups, or when comparing rural and urban populations. However, we note that the same effects have been reported in local-scale studies: for example, in a remote northern Australian Indigenous community, increased number of years schooling is associated with reduced use of Indigenous language elements across all generations, from elders to children 42 . Collection of regional data on variation in number of years of schooling would allow the generality of this relationship to be tested at a range of spatial scales.

Similarly, our data on education policy is necessarily coarse grained, which may mask some patterns at local scales: national legal provision may not reflect use of minority languages in schools at a local or regional level. For example, in China, the Regional Ethnic Autonomy Law (1984) promotes learning both regional languages and Mandarin Chinese, but the policy is not translated into educational practice evenly across all regions due to lack of resources in some languages, or local emphasis in some places on students from minorities learning the centralized language of governance and commerce 39 . The same bilingual education policy may invigorate minority languages in some areas, but result in greater emphasis on education in the dominant national language in other places 43 . More fine-grained analysis at regional level is needed to examine the influence of minority languages in classrooms on language diversity and vitality.

Our results not only identify global threats to language vitality, but also reveal differences in threatening processes in different regions. For example, in Africa, language endangerment is associated with greater areas of pasture or croplands, potentially reflecting language shift associated with subsistence change (for example, as hunter–gatherer societies adopt the languages of neighbouring pastoral or agricultural groups 44 ). Climate has the strongest association with language endangerment in Europe, with endangerment levels increasing with temperature seasonality, reflecting patterns of language erosion in Arctic regions. These regional patterns are ideal foci for future studies of language endangerment: while the current study is constrained to predictors that are globally relevant and consistently measured for all regions of the world, a targeted study could focus on variables considered important at regional scales, such as land use and subsistence in Africa, population density change in Oceania, or climate in Europe and Central and Eastern Asia (Supplementary Fig. 7 ).

Predicting future language loss

If a language is no longer being learned by children, we can use demographic information to predict when, in the absence of interventions to increase language transmission, there will be no more living L1 speakers. We can combine the current L1 speaker population size with endangerment score (which tells us the relative generational distribution of L1 speakers and whether the number of L1 speakers is declining; Supplementary Table 1 ), and use demographic information on age structure of the population (Supplementary Table 8 ) to predict how many L1 speakers will be alive in the future (see Supplementary Methods 5 for details). Our analysis is conservative in that it only considers change in L1 speakers in languages identified as having reduced transmission to younger generations (see Supplementary Table 1 ): we did not model change in speaker number for languages currently considered to be stably transmitted, even though they may become endangered in the future.

Our model predicts that language loss will at least triple in the next 40 years (Fig. 2 ). Without intervention to increase language transmission to younger generations, we predict that by the end of the century there will be a nearly five-fold increase in Sleeping languages, with at least 1,500 languages ceasing to be spoken (Fig. 3 ). Some parts of the world stand out as ‘hotspots’ of future language loss, with the greatest absolute loss of languages predicted to occur in the west coast of North America, Central America, the Amazon rainforest, West Africa, north coast of New Guinea and northern Australia (Extended Data Fig. 4 ). After 80 years, the model predicts additional areas of language loss in Borneo, southwest China and areas around the Caspian Sea. The greatest proportional loss of languages is predicted to occur in the Arctic, interior plains of Northern America, temperate areas of southern Chile and the Sahara (Extended Data Fig. 5 ).

a , b , The red shading represents the differences between the predicted values at present and the predicted values in 40 years, for the absolute number ( a ) and proportion of languages ( b ) per hex grid, based on generational shift and demographic transition in L1 speakers. c , Proportion of languages predicted to become Sleeping (EGIDS ≥ 9) in the next 40 years. See Supplementary Table 1 for information on endangerment scales. Language distribution data from WLMS 16 ( worldgeodatasets.com ).

a , Current and predicted proportion of languages that are endangered (EGIDS 6b–8b) or Sleeping (no living L1 speakers, EGIDS 9–10). b , c , Current and predicted number of endangered (6b–8b) ( b ) and Sleeping (9–10) ( c ) languages according to the current level of language documentation. Each violin gives the probability distribution of the number or proportion of languages that are predicted to be endangered or Sleeping, with the dot showing the mean and the whisker showing the standard deviation. Each dashed line shows the number or proportion of languages that are currently endangered or Sleeping. This figure projects current levels of documentation for each language, hence does not reflect future documentation efforts of threatened languages.

In addition to demographic shift, our model also identifies predictors of language endangerment that are likely to change over time. For some of the variables associated with language endangerment, such as average years of schooling, we lack an adequate predictive model that is global in extent but would allow for regional variation. However, there are some variables identified as significant predictors of language endangerment at regional levels, such as land use and climate, for which we can predict future values on the basis of current trends (see Supplementary Information 5.2 ). For example, we can use climate change models to predict future values of climate variables at all points of the globe, and we can use information on rates of change in land use in each grid cell to project possible future values for land use variables in that grid cell. Clearly, such predictions should be regarded as possible values only, and all such future projections are subject to caveats: for example, we chose a mid-range climate model so the future values could be higher or lower depending on the effectiveness of global climate change mitigation strategies, and the land-use projections are based on the average rate of change in the last few decades, although local factors may cause those rates of change to either increase or decrease in the future. But it is a useful exercise to add climate and land use to the predictive model to illustrate the potential for forward prediction of variables impacting endangerment status. The results of the predictions based on generational shift and demographic transition are shown in Figs. 2 and 3 . Predictions that are additionally adjusted for change in climate and land-use variables show qualitatively the same results (Extended Data Figs. 2 – 5 ).

Safeguarding language diversity

The crisis of language endangerment has prompted worldwide efforts to recognize, document and support language diversity 45 , reflected in the UNESCO International Decade of Indigenous Languages, beginning in 2022. Every language represents a unique expression of human culture, and each is subject to idiosyncratic influences of their specific history and local sociopolitical environment. By identifying general factors that impact language vitality, or areas at greatest risk of language loss, we may be better placed to direct resources for maintenance of language diversity.

In biology, ‘extinction debt’ describes the inevitable loss of species that are currently persisting with inviable populations or insufficient habitat 46 , 47 . For languages, ‘extinction debt’ arises from reduced intergenerational transmission. Languages currently spoken by adults but not learned as a first language by children will, without active intervention and revitalization, have no more L1 speakers once the current L1 speakers die. Using information on intergenerational transmission for each language combined with demographic information, our model predicts that the greatest increase in endangered languages will coincide with areas of greatest language diversity, in particular New Guinea and Central America (Fig. 2a ). However, some regions are predicted to lose a greater proportion of their current language diversity, such as the Great Lakes region of North America, the northern Sahara and eastern Siberia (Fig. 2 ).

We emphasize that these predictions are not death knells, but possible outcomes in the absence of investment in language vitality. For example, while our model predicts Alutiiq (Pacific Gulf Yupik {ems}) in Alaska to increase in endangerment level, the community has instituted a language revitalization programme that may counter the predicted trend. Identifying external factors associated with language endangerment can focus attention on areas where language vitality might become threatened. For example, some areas with the greatest predicted increase in road density, such as Nigeria, Papua New Guinea and Brazil 48 , are predicted by our model to have the highest potential loss of languages (Extended Data Fig. 4 ). Since increasing road density also has negative impacts on biodiversity, focusing mitigation efforts on areas of increasing road density may be beneficial for both language vitality and biodiversity 49 , 50 .

In addition to identifying correlates of language endangerment that are likely to change in the future, such as land use, we also identify factors that are open to intervention to reduce loss of language diversity. Currently, more years of formal schooling are associated with greater rates of language endangerment (Fig. 1 ). Research suggests that bilingual education, where students learn part or all of the curriculum in their first language, typically results in greater overall academic achievement without sacrificing proficiency in the dominant national language 51 , but emphasis on high-stakes testing for competency in the national language can contribute to erosion of heritage language proficiency 42 . Having provision for bilingual education enshrined in legislation, or official recognition of minority languages in government or in education, is not sufficient to reduce language endangerment (Supplementary Fig. 7 ). Implementation requires genuine commitment to bilingual education, and support from community members who can bring heritage language to the classroom. The benefits of providing support to enhance Indigenous language vitality, in terms of wellbeing 52 , 53 and socioeconomic outcomes 54 , are likely to far outweigh the costs. Implementation of support for Indigenous language vitality at all levels of governance and within speaker communities is urgent, given the predicted loss of L1 speakers who can aid language vitality and transmission (Fig. 3 ).

We emphasize that our analysis is focused on L1 speakers who learned the language as children, reflecting continuity of language transmission over generations. A language classified as ‘Sleeping’ (no L1 speakers) may be spoken as an acquired (L2) language in a multilingual context, as a reflection of ethnic identity or through revitalization (which may ultimately generate new L1 speakers). Language revitalization benefits from documentation, such as texts, dictionaries and grammars. Our future predictions give cause for concern that within 80 years there could be 1,500 or more languages that will no longer be spoken, yet a third of these currently have little or no documentation (Fig. 3 ). The majority of these languages currently have living L1 speakers, so there is still time to increase documentation based on the expert knowledge of fluent first-language speakers 55 , and to support communities to re-invigorate intergenerational language transmission 56 .

The loss of language diversity results from a complex network of factors, particularly those associated with colonization, globalization, and social and economic change. While identifying correlates of endangerment does not provide a full picture of the loss or erosion of any particular language, it does contribute to a general ‘theory of language loss’ 38 , 57 . A key difference between species and language endangerment patterns is that while many correlates of species extinction risk are intrinsic features of species biology (such as low reproductive rate or specialist diet 58 ), we have considered only ‘external’ factors, which frame the context in which languages persist. But external factors, unlike species traits, are amenable to manipulation. Some identified predictors of language endangerment may act as ‘red flags’, highlighting areas that would benefit from interventions to support language vitality (such as regions where road networks are expanding rapidly) or prompt finer-grained analysis of potential impacts (such as educational policy). Our study highlights the critical level of under-documentation of language diversity (Fig. 3 ), showing that without intervention, we might lose a substantial proportion of language diversity without having ever adequately documented how those languages represent unique expressions of human cultural diversity 59 . Investing in speaker communities to provide them with the support they need to encourage language diversity and vitality will bring measurable benefits in terms of social justice, community wellbeing and cultural engagement 53 , 54 , 55 , 60 .

Language data

We used data on L1 speakers, geographic distribution, endangerment level and relationships for 6,511 languages classified as ‘spoken L1 languages’ 17 , 61 , 62 (see Supplementary Methods for details of data and variables). We give the standard nomenclature according to the ISO 639-3 three-letter language identifiers in Supplementary Data 1 , and throughout this document we give the ISO code in curly brackets at the first mention of a language. Nine ‘world languages’ were included only as factors potentially influencing language vitality (see Supplementary Table 2 ) but were otherwise excluded from all language-level analyses. There are several schemes for evaluating and categorizing the risk of language loss 63 , 64 , most of which target indicators of language vitality, such as intergenerational transmission, official recognition, domains of use, and level of documentation and resources 23 , 65 (Supplementary Table 1 ). We based our analysis on EGIDS because it provides the most comprehensive coverage for our data (Supplementary Methods 2.1.2 and Fig. 1 ). Signed languages were not included in this analysis due to insufficient information on number of L1 signers, distributions and endangerment status for the majority of the world’s signed languages (Supplementary Information section 2.1.6 ).

Many previous analyses of global patterns of language endangerment relied on speaker population size and geographic distribution as proxies of endangerment status 4 , 20 , 66 . While low speaker number is the ultimate outcome of endangerment, current population size may not always provide a reliable indicator of language vitality or risk of loss 67 , 68 . Small localized languages may be stable and vigorous, for example some Papuan languages are confined to one or a few villages with only hundreds of speakers, yet are not considered endangered (for example, Neko {ISO 639-3: nej}, Mato {met}), and large widespread languages are not secure if they are not being reliably transmitted to younger generations (for example, Domari {rmt}, an endangered Indo-European language with over a quarter of a million speakers). Using population and range size to represent endangerment also conflates endangerment and diversity: range and population size correlate with number of languages per unit area 17 , so an area with more languages may, all things being equal, also contain a larger number of endangered languages 4 , 20 . Our analysis emphasizes global trends and general patterns over specific language trajectories or local histories. Use of global databases provides an overview of language diversity and vitality, but it is not possible to verify current speaker numbers, endangerment and distributions without expert knowledge of each individual language. Some regions or language families may be less well represented in global databases (for example, Australian languages have patchy representation and would benefit from expert revision on speaker numbers and endangerment levels). Furthermore, there is often no clear line between a dialect and a language, and this can result in variation in assigning L1 speakers to languages (Supplementary Methods 2.1.2 ). Our results should therefore be interpreted as providing general patterns and broad-brush predictions rather than specific detail on particular languages.

Predictor variables

We included ten broad categories of variables to describe key extrinsic factors that influence language vitality (Table 1 ). Variables were either recorded per language, as a weighted average across the language area or national values, or for a 10,000 km 2 ‘neighbourhood’ around the language (see Supplementary Methods for details). For each language, we recorded the reported number of L1 speakers, endangerment level (Supplementary Table 1 ), distribution 62 , level of documentation 61 , whether the language has official recognition in any country, or is officially recognized as a language of education. We characterized the ‘language ecology’ by the diversity of languages in the surrounding area, the number and proportion of endangered languages in the area, the relative representation of speakers compared to nearby languages, and whether it occurs in a country (or countries) that has one of nine ‘world languages’ as an official language (Supplementary Table 3 ). We recorded levels of educational attainment and education spending at national level, as well as the presence of a general provision for the use of minority languages for instruction in all or part of formal schooling, and whether each language is recognized for use in education (Supplemenary Tables 5 and 6 ). Socioeconomic context is represented by Gross Domestic Product per capita (GDPpc), the Gini index of income inequality and life expectancy at 60 years of age (Supplementary Tables 5 and 7 ), noting that these national averages do not capture variation between groups or areas within each country (see Supplementary Information 2.4 ).

To represent the environmental context of each language, we included variables representing population density, climate, land use, biodiversity loss, connectivity and ‘shift’ (that is, the rate of change in land use, population, built environment) (Table 1 ). Because language loss is often a result of language expansion replacing autochthonous languages, we included measures of connectivity: density of roads and navigable waterways (which encourage human movement) and landscape roughness and altitudinal extent (which discourage human movement). To indicate human impact on the natural environment, we included ‘human footprint’ (which summarizes anthropogenic impacts on the environment 69 ) and measures of biodiversity loss. We included factors previously shown to be correlates of language diversity: mean growing season, average temperature, temperature seasonality and precipitation seasonality (we did not include species richness because biodiversity patterns are not significantly associated with language diversity above and beyond these climatic covariables 17 ). To model the impact of changing landscape and environment, we included rates of change in urbanization, population density, land use and human footprint 69 .

The variables we included vary in their degree of spatial resolution. For variables concerning legislation and policy (for example, provision for minority language education), data is typically available only at country level. For some socioeconomic variables, such as life expectancy, there is regional data for some countries, but most areas only have country level data, so for consistency we used national averages provided by global organizations such as the World Bank and World Health Organization (Table 1 ). For environmental variables, such as temperature seasonality, we averaged values over all grid cells in the language distribution area, but for landscape factors influencing human movement, such as mountains and roads, values within the language area are not fully informative because we wish to capture movement between language areas. For these variables, we averaged over all grid cells in a ‘neighbourhood’ centred on the language distribution. For full details of the spatial resolution of each variable, see Supplementary Methods .

The variables included in this study necessarily represent current environments, socioeconomic status and contemporary policy settings. Aside from shift variables (Table 1 ), which represent change over time, we cannot directly capture historical processes, such as past educational programmes, historical disease epidemics, warfare or genocide. These are important factors in language endangerment but cannot be easily represented in globally consistent, universally available variables, so investigating the impact of these factors is beyond the scope of this analysis.

Previous analyses of global language endangerment included relatively few potential predictors and did not control for the confounding effects of both spatial proximity and relationships between languages 2 , 4 , 20 , 66 . Languages that cluster in space will share many environmental, social and economic features. Related languages may share not only many linguistic features but also many environmental, social and economic factors and shared historical influences 17 . All analyses rest on the assumption that datapoints are statistically independent of each other, so if we find that the residuals of the model show phylogenetic signal, then phylogenetic non-independence (when datapoints are related by descent) violates the assumption of standard statistical tests and can lead to spurious relationships 70 , 71 . Our method estimates the contribution of relatedness to observed patterns of endangerment, so that if there is little or no influence of relatedness on patterns of endangerment, then the phylogeny will have no effect on the outcomes 17 . A large contribution of phylogeny tells us that languages tend to be more similar to related languages in their endangerment status than they are to randomly selected languages. This does not imply that languages inherit either their endangerment status or threatening processes from their ancestors, but that relatives show patterns of similarity of endangerment 72 . If this is the case, we need to account for this phylogenetic non-independence in our analysis, so that we can identify factors that are significantly associated with endangerment above the association which is expected purely due to their shared relationships (closely related languages having more similar patterns of endangerment).

Failure to account for spatial autocorrelation can lead to false inference of patterns of language endangerment 19 . For example, socioeconomic indicators such as GDP have a strong latitudinal gradient, and so does language diversity and range size, so if range size is associated with endangerment, we would expect a significant correlation between GDP and language endangerment even if there is no direct influence of one on the other 71 . Just as repeatedly sampling two neighbouring areas but counting each observation as a unique datapoint inflates perceived environmental correlations by pseudoreplication 73 , repeatedly sampling related languages with similar cultural traits, linguistic features, historical influences and language ecologies also potentially inflates perceived associations between endangerment and environmental or social factors 19 , 70 . Both of these sources of covariation in the data must be accounted for to find meaningful correlates of language endangerment.

In our analysis, the dependent variable is the level of endangerment, based on EGIDS rankings (Supplementary Table 1 ). We are seeking global correlates of language endangerment, but we are aware that some threatening processes may have greater or lesser impact in different regions (Supplementary Methods 2.1.3 ). Therefore, in addition to the predictors we described above, we included an interaction term between each region and each independent predictor, to account for any region-specific effect of the predictor on endangerment. This interaction term was constructed by taking the product of the predictor and a binary variable recording whether a language belongs to the region. Any interaction term with no variation in the corresponding region was removed. Instead, we included an intercept for each region to account for differences in the average level of language endangerment among regions. In total, we have 51 predictors, 51 by 12 interaction terms, and 12 intercepts in the independent variables (Supplementary Data 3 ).

The basic steps of our statistical analysis are:

applying transformations to the 51 predictors (Supplementary Methods 4.1 , Table 1 ), then calculating their interaction terms;

grouping the 51 predictors according to their pairwise correlation (Supplementary Methods 4.2 ) and grouping interaction terms with their corresponding predictors (Supplementary Data 3 );

dividing the dataset into two, with two-thirds of the languages assigned to a training dataset and one-third to a test dataset. The training dataset was used to select the independent variables (candidate models) to predict current endangerment level (Supplementary Methods 4.3 ) and the test dataset was used to evaluate the fit of these candidate models to predict endangerment level (Supplementary Methods 4.4 );

using the best model, re-estimating the model parameters using all 6,511 languages;

using the predicted change in L1 speaker population, environment and climate to generate future values of variables, then using the best model to predict future endangerment given these predicted future values (Supplementary Methods 5 ).

Because the dependent variable in our analysis (endangerment level) is an ordinal variable, we used ordinal probit regression 74 to model language endangerment status. To satisfy the parallel regression assumption (that an independent variable has the same effect on threat status across all endangerment levels) for the majority of variables, we grouped recorded EGIDS scores into seven levels by combining levels 1–6a into a ‘stable’ level (Supplementary Methods 4.2 and Table 1 ). To account for spatial and phylogenetic autocorrelation, we constructed three matrices. The phylogenetic matrix represents relationships between languages as inferred from a taxonomy, with branch lengths scaled to relative divergence depths 17 (Supplementary Methods 3 ). The distance matrix captures similarity in nearby languages due to shared environment using an exponentially decreasing function of the great-circle distance between the centroids of polygons of two languages. Since distance between centroids may not reflect on-the-ground language contact, we also used a contact matrix which contains 1 if two language polygons overlap (allowing a buffer of 100 km around each polygon), and 0 otherwise. We do not expect this contact matrix to fully capture the degree of ongoing contact between languages, which may be determined by local factors including modes of transport, form of subsistence or connectivity, but we included it to allow for an influence of close association between language distributions as an influence on patterns of endangerment, above and beyond the great-circle distances between the centres of language distributions. The distance, contact and phylogenetic matrices had zero diagonals and each row was normalized to unity. Because each matrix had its own coefficient, if patterns of autocorrelation due to distance, contact or relatedness were not important in shaping the values of variables, then the model would estimate the coefficient to zero and the matrix would not influence the result.

We then fitted an autoregressive ordinal probit model to the data. We modelled the threat status of a language as a linear function of not only the independent variables but also the threat status of other languages whose associations with the language depend on the distance, contact and phylogenetic matrices. The model was fitted to the data using a two-stage least squares approach 74 implemented in a custom R code based on the ‘ordinalNet’ package 75 . We used a weighted sum of all the three matrices to describe autocorrelation among languages 17 . The weight was estimated by maximum likelihood using the ‘L-BFGS-B’ method 76 in the ‘optim’ function in R.

To select the best model to predict endangerment level in our data, we first randomly divided the data into a training dataset (including 2/3 of the languages) and a test dataset (the remaining 1/3 of the languages). Then, we grouped highly correlated independent variables together and applied a stepwise selection procedure to the training dataset (see step 3) to select candidate models (details in Supplementary Methods 4.4 ). The procedure started with a model of a single independent variable that had the highest likelihood to the training dataset, then goes through each group (see step 2) in a random order by adding a variable of the group to the model that significantly and maximally increased model fit, and removing a variable of the group from the model that had the least and non-significant impact on model fit. These steps were repeated until there were no more variables that could be added that increased model fit, or could be subtracted without reducing model fit. This model selection procedure left us with a set of candidate models. Lastly, we measured the predictive power of each model by predicting the threat status of the languages in the test dataset and constructed the best model on the basis of its predictive power.

The best model was constructed by including predictor variables that were selected in over one-third of the candidate models which did not significantly differ in their predictive power from the model with the highest predictive power. We then estimated the coefficients of predictor variables on the complete dataset. We used this best model to predict, for each language, the probability that the language falls in each of the seven endangerment levels (combining 1–6a into one ‘Stable’ level; Supplementary Table 1 ). Using these probabilities, we randomly sampled the endangerment level of each language and counted the number of languages with sampled endangerment level of 2 or above (that is, EGIDS 6b–10) as the number of languages predicted to be endangered, or those in the top two levels (that is EGIDS 9–10) as the number of languages predicted to be Sleeping. This procedure was repeated 1,000 times to generate the probability distribution of the number of languages predicted to be endangered or Sleeping. We found that the expected endangerment level tends to be lower than the reported endangerment level for individual languages (Supplementary Fig. 6 ), but, over all the languages, the model accurately predicted the proportions of languages that are endangered and sleeping (Fig. 3 ).

In some cases, the mismatch between predicted and observed endangerment levels may reflect ‘latent risk’ in endangerment status 27 : languages that have characteristics typical of an endangered language, such as low L1 speaker population size, yet are rated as stable (Extended Data Fig. 1 ). These languages may be expected to come under increasing threat in the future. For example, Yindjibarndi {yij}, a language of the Pilbara region of Australia, has an EGIDS rating of 6a (Stable) but has a small L1 speaker population (310) and is in an area where many languages are endangered or no longer spoken. Our model predicts the expected endangerment level of this language as ‘Critically Endangered’ (EGIDS 8) at present, and without intervention to ensure language vitality, it could potentially be no longer spoken within 80 years. The reported endangerment level and the predicted probability of each language falling in each endangerment level at present, in 40 years and in 80 years are listed in Supplementary Data 4 .

Future prediction

We used the best model of current language endangerment status to predict possible future changes in endangerment status for our global database of languages. Current EGIDS levels give us information on intergenerational transmission, so we can use that information to model declining L1 speaker population: if a language is currently only spoken by adults and not transmitted to children, then, without revitalization, there will be no more L1 speakers once the current speakers die. EGIDS also indicates which languages are declining in L1 speaker population so we can model the probable decline in numbers in 40 years (2060) and 80 years (2100; Supplementary Methods 5.2.1 ). These models predict possible patterns of language loss in the absence of revitalization programmes that might increase the number of L1 speakers, by assuming that without intervention to improve language transmission and vitality, endangered languages will undergo demographic shift that changes endangerment level, as described in Supplementary Methods 5.1 and Table 7 . These predictions are conservative in the sense that they assume that languages that are not currently endangered will remain stable into the future. We emphasize that this procedure is specifically modelling the shift in number of first language (L1) speakers of a language, not the population they belong to. A population may thrive and its ethnic identity remain strong even if speakers shift to a different language. To model the L1 speaker population size, we need to consider generational transmission of the language (that is, are children learning it as their first language?), rather than the number of people in the population that they belong to.

For example, if a language with an EGIDS level of 6b (Threatened) is predicted to be Endangered (EGIDS level 7) in the future on the basis of having no child L1 speakers, we adjust the probability distribution of the endangerment level predicted by the model for the language at that timepoint by shifting the probability distribution one level up, setting the probability that the language has an endangerment level lower than Endangered to zero, and renormalizing the probability distribution. We then randomly sample the endangerment level of each language, and count the number of languages overlapping each hex grid that are Endangered or Sleeping. This procedure is repeated 1,000 times to get the probability distribution of the number of languages predicted to be endangered or sleeping in each hex grid. We plot the combined predictions on a map, showing both the expected value of the number of languages per grid that are endangered or sleeping in 40 and 80 years, and also the proportion of languages per grid that are Threatened, Endangered or Sleeping. In the Supplementary Information , we demonstrate how this predictive model can be extended to incorporate future values of predictor variables, such as changing climate or land use.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

All variables analysed are provided in Supplementary Data . These variables are derived from a range of sources, as cited in the text and in Table 1 (most of these data are freely available but some are under license).

Code availability

Code for data preparation is available at https://github.com/rdinnager/language_endangerment . Code for running the analysis is available at https://github.com/huaxia1985/LanguageEndangerment . The custom R code includes functions that modify functions in the ‘ordinalNet ’ R package to correct for autocorrelation in ordinal probit regression.

Change history

03 february 2022.

A Correction to this paper has been published: https://doi.org/10.1038/s41559-022-01684-4

Rehg, K. L. & Campbell, L. The Oxford Handbook of Endangered Languages (Oxford Univ. Press, 2018).

Romaine, S. in Language and Poverty (eds Harbert, W. et al.) Ch. 8 (Multilingual Matters, 2009).

Sallabank, J. & Austin, P. The Cambridge Handbook of Endangered Languages (Cambridge Univ. Press, 2011).

Sutherland, W. J. Parallel extinction risk and global distribution of languages and species. Nature 423 , 276–279 (2003).

Article   CAS   Google Scholar  

Eberhard, D. M., Simons, G. F. & Fennig, C. D. Ethnologue: Languages of the World 22nd edn (SIL International, 2019); https://www.ethnologue.com/

Moseley, C. Atlas of the World’s Languages in Danger (UNESCO Publishing, 2010); http://www.unesco.org/culture/en/endangeredlanguages/atlas

Catalogue of Endangered Languages (University of Hawaii at Manoa, 2020); http://www.endangeredlanguages.com

Campbell, L. & Okura, E. in Cataloguing the World’s Endangered Languages 1st edn (eds Campbell, L. & Belew, A.) 79–84 (Routledge, 2018).

The IUCN Red List of Threatened Species Version 2019-2 (IUCN, 2019); http://www.iucnredlist.org

Romaine, S. in The Routledge Handbook of Ecolinguistics (eds Fill, A. F. & Penz, H.) Ch. 3 (Routledge, 2017).

Crystal, D. Language Death (Cambridge Univ. Press, 2000).

Simons, G. F. Two centuries of spreading language loss. Proc. Linguist. Soc. Am. 4 , 27–38 (2019).

Article   Google Scholar  

Krauss, M. The world’s languages in crisis. Language 68 , 4–10 (1992).

Brondizio, E. S., Settele, J., Díaz, S. & Ngo, H. T. (eds) Global Assessment Report on Biodiversity and Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES, 2019).

Bowern, C. Language vitality: theorizing language loss, shift, and reclamation (Response to Mufwene). Language 93 , e243–e253 (2017).

Mufwene, S. S. Language vitality: The weak theoretical underpinnings of what can be an exciting research area. Language 93 , e202–e223 (2017).

Hua, X., Greenhill, S. J., Cardillo, M., Schneemann, H. & Bromham, L. The ecological drivers of variation in global language diversity. Nat. Commun. 10 , 2047 (2019).

Grenoble, L. A. & Whaley, L. J. in Endangered Languages (eds Grenoble, L. A. & Whaley, L. J.) 22–54 (Cambridge Univ. Press, 1998).

Cardillo, M., Bromham, L. & Greenhill, S. J. Links between language diversity and species richness can be confounded by spatial autocorrelation. Proc. R. Soc. B 282 , 20142986 (2015).

Amano, T. et al. Global distribution and drivers of language extinction risk. Proc. R. Soc. B 281 , 20141574 (2014).

Loh, J. & Harmon, D. Biocultural Diversity: Threatened Species, Endangered Languages (WWF, 2014).

Fishman, J. A. Reversing Language Shift: Theoretical and Empirical Foundations of Assistance to Threatened Languages Vol. 76 (Multilingual Matters, 1991).

Lewis, M. P. & Simons, G. F. Assessing endangerment: expanding Fishman’s GIDS. Rev. Roum. Linguist. 55 , 103–120 (2010).

Google Scholar  

Hinton, L. in The Green Book of Language Revitalization in Practice (eds Hinton, L. & Hale, K.) 413–417 (Brill, 2001).

Hobson, J. R. Re-awakening Languages: Theory and Practice in the Revitalisation of Australia’s Indigenous Languages (Sydney Univ. Press, 2010).

Di Marco, M. et al. A novel approach for global mammal extinction risk reduction. Conserv. Lett. 5 , 134–141 (2012).

Cardillo, M., Mace, G. M., Gittleman, J. L. & Purvis, A. Latent extinction risk and the future battlegrounds of mammal conservation. Proc. Natl Acad. Sci. USA 103 , 4157–4161 (2006).

Bolam, F. C. et al. How many bird and mammal extinctions has recent conservation action prevented? Conserv. Lett. 14 , e12762 (2020).

Balmford, A. Extinction filters and current resilience: the significance of past selection pressures for conservation biology. Trends Ecol. Evol. 11 , 193–196 (1996).

Brenzinger, M. Language Death: Factual and Theoretical Explorations with Special Reference to East Africa (Mouton de Gruyter, 1992).

Aikhenvald, A. Y. in Language Endangerment and Language Maintenance: An Active Approach (eds Bradley, D. & Bradley, M.) 24–33 (Taylor & Francis, 2002).

Aikhenvald, A. Y. in Lectures on Endangered Languages: 5. Endangered Languages of the Pacific Rim (eds Sakiyama, O. & Endo, F.) 97–142 (ELPR, 2004).

van Driem, G. in Language Diversity Endangered (ed. Brenzinger, M.) Ch. 14 (Mouton de Gruyter, 2007).

Muysken, P. in Historicity and Variation in Creole Studies (eds Highfield, A. & Valdman, A.) 52–78 (Karoma, 1981).

Gal, S. Language Shift: Social Determinants of Linguistic Change in Bilingual Austria (Academic Press, 1979).

Holmquist, J. Social correlates of a linguistic variable: a study in a Spanish village. Lang. Soc. 14 , 191–203 (1985).

Dobrin, L. M. in Endangered Languages: Beliefs and Ideologies in Language Documentation and Revitalization (eds Austin, P. K. & Sallabank, J.) Ch. 7 (British Academy, 2014).

Sasse, H.-J. in Language Death: Factual and Theoretical Explorations with Special Reference to East Africa (ed Brenzinger M.) 7–30 (Mouton de Gruyter, 1992).

Wang, Y. & Phillion, J. Minority language policy and practice in China: the need for multicultural education. Int. J. Multicult. Educ. 11 , 1–14 (2009).

McCarty, T. L. in Language Policies in Education: Critical Issues (ed. Tollefson, J. W.) 285–307 (2002).

Wiese, A.-M. & Garcia, E. E. The Bilingual Education Act: language minority students and equal educational opportunity. Biling. Res. J. 22 , 1–18 (1998).

Bromham, L., Hua, X., Algy, C. & Meakins, F. Language endangerment: a multidimensional analysis of risk factors. J. Lang. Evol. 5 , 75–91 (2020).

Gao, X. & Ren, W. Controversies of bilingual education in China. Int. J. Biling. Educ. Biling. 22 , 267–273 (2019).

Dimmendaal, G. J. in Investigating Obsolescence: Studies in Language Contraction and Death (ed. Dorian N. C.) 13-32 (Cambridge Univ. Press, 1989).

Brenzinger, M. in Language Diversity Endangered (ed. Brenzinger, M.) IX–XVII (Mouton de Gruyter, 2007).

Kuussaari, M. et al. Extinction debt: a challenge for biodiversity conservation. Trends Ecol. Evol. 24 , 564–571 (2009).

Tilman, D., May, R. M., Lehman, C. L. & Nowak, M. A. Habitat destruction and the extinction debt. Nature 371 , 65–66 (1994).

Meijer, J. R., Huijbregts, M. A., Schotten, K. C. & Schipper, A. M. Global patterns of current and future road infrastructure. Environ. Res. Lett. 13 , 064006 (2018).

Laurance, W. F. & Balmford, A. A global map for road building. Nature 495 , 308–309 (2013).

Newbold, T. et al. Global effects of land use on local terrestrial biodiversity. Nature 520 , 45–50 (2015).

Crawford, J. Language politics in the U.S.A.: the paradox of bilingual education. Soc. Justice 25 , 50–69 (1998).

Hallett, D., Chandler, M. J. & Lalonde, C. E. Aboriginal language knowledge and youth suicide. Cogn. Dev. 22 , 392–399 (2007).

Taff, A. et al. in The Oxford Handbook of Endangered Languages (eds Rehg, K. & Campbell, L.) 862–883 (Oxford Univ. Press, 2018).

Dinku, Y. et al. Language Use is Connected to Indicators of Wellbeing: Evidence from the National Aboriginal and Torres Strait Islander Social Survey 2014/15 . CAEPR Working Paper no. 132/2019 (CAEPR, 2020); https://doi.org/10.25911/5ddb9fd6394e8

Essegbey, J., Henderson, B. & McLaughlin, F. Language Documentation and Endangerment in Africa (John Benjamins, 2015).

Davis, J. L. Language affiliation and ethnolinguistic identity in Chickasaw language revitalization. Lang. Commun. 47 , 100–111 (2016).

Clyne, M. in Maintenance and Loss of Minority Languages (eds Fase, W. et al.) 17–36 (John Benjamins, 1992).

Cardillo, M. et al. The predictability of extinction: biological and external correlates of decline in mammals. Proc. R. Soc. B 275 , 1441–1448 (2008).

Evans, N. Dying Words: Endangered Languages and What They Have to Tell Us Vol. 22 (John Wiley & Sons, 2011).

Ndhlovu, F. in Language Planning and Policy: Ideologies, Ethnicities, and Semiotic Spaces of Power (eds Abdelhay, A. et al.) 133–151 (Cambridge Scholars, 2020).

Hammarström, H., Forkel, R. & Haspelmath, M. Glottolog 4.1. http://glottolog.org (Max Planck Institute for the Science of Human History, 2019).

Lewis, M. P., Simons, G. F. & Fennig, C. D. Ethnologue: Languages of the World 17th edn http://www.ethnologue.com (SIL International, 2013).

King, K. A., Schilling-Estes, N., Lou, J. J., Fogle, F. & Soukup, B. Sustaining Linguistic Diversity: Endangered and Minority Languages and Language Varieties (Georgetown Univ. Press, 2008).

Lee, N. H. & van Way, J. Assessing levels of endangerment in the Catalogue of Endangered Languages (ELCat) using the Language Endangerment Index (LEI). Lang. Soc. 45 , 271–292 (2016).

Language Vitality and Endangerment: International Expert Meeting on UNESCO Programme Safeguarding of Endangered Languages (UNESCO, 2003).

Tershy, B. R., Shen, K.-W., Newton, K. M., Holmes, N. D. & Croll, D. A. The importance of islands for the protection of biological and linguistic diversity. BioScience 65 , 592–597 (2015).

Igboanusi, H. Is Igbo an endangered language? Multilingua 25 , 443–452 (2006).

Ravindranath, M. & Cohn, A. C. Can a language with millions of speakers be endangered? J. Southeast Asian Linguist. Soc. 7 , 64–75 (2014).

Venter, O. et al. Sixteen years of change in the global terrestrial human footprint and implications for biodiversity conservation. Nat. Commun. 7 , 12558 (2016).

Bromham, L., Hua, X., Cardillo, M., Schneemann, H. & Greenhill, S. J. Parasites and politics: why cross-cultural studies must control for relatedness, proximity and covariation. R. Soc. Open Sci. 5 , 181100 (2018).

Bromham, L., Skeels, A., Schneemann, H., Dinnage, R. & Hua, X. There is little evidence that spicy food in hot countries is an adaptation to reducing infection risk. Nat. Hum. Behav. https://doi.org/10.1038/s41562-020-01039-8 (2021).

Purvis, A., Cardillo, M., Grenyer, R. & Collen, B. in Phylogeny and Conservatio n (eds Purvis, A. et al.) 295–316 (Cambridge Univ. Press, 2005).

Hurlbert, S. H. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54 , 187–211 (1984).

Dow, M. M. Network autocorrelation regression with binary and ordinal dependent variables: Galton’s problem. Cross Cult. Res. 42 , 394–419 (2008).

Wurm, M. J., Rathouz, P. J. & Hanlon, B. M. Regularized ordinal regression and the ordinalNet R package. Preprint at https://arxiv.org/abs/1706.05003 (2017).

Byrd, R. H., Lu, P., Nocedal, J. & Zhu, C. A limited memory algorithm for bound constrained optimization. SIAM J. Sci. Comput. 16 , 1190–1208 (1995).

Barro, R. L. & Lee, J.-W. A new data set of educational attainment in the world, 1950–2010. J. Dev. Econ. 104 , 184–198 (2013).

Leclerc, J. L’aménagement linguistique dans le monde http://www.axl.cefan.ulaval.ca/monde/index_alphabetique.htm (2019).

Solt, F. The Standardized World Income Inequality Database, Version 8 https://doi.org/10.7910/DVN/LM4OWF (2019).

Global Agro-ecological Zones (GAEZ v3.0) (FAO, IIASA, 2010).

Download references

Author information

These authors contributed equally: Lindell Bromham, Xia Hua.

Authors and Affiliations

Macroevolution and Macroecology, Research School of Biology, Australian National University, Canberra, Australian Capital Territory, Australia

Lindell Bromham, Russell Dinnage, Andrew Ritchie, Marcel Cardillo & Xia Hua

ARC Centre of Excellence for the Dynamics of Language, Australian National University, Canberra, Australian Capital Territory, Australia

Hedvig Skirgård & Simon Greenhill

Department of Linguistic and Cultural Evolution, Max Planck Institute for the Science of Human History, Jena, Germany

ARC Centre of Excellence for the Dynamics of Languages, School of Languages and Cultures, University of Queensland, Brisbane, Queensland, Australia

Felicity Meakins

Mathematical Sciences Institute, Australian National University, Canberra, Australian Capital Territory, Australia

You can also search for this author in PubMed   Google Scholar

Contributions

L.B., R.D., H.S., A.R., M.C., F.M., S.G. and X.H. conceived and designed the experiments; X.H. analysed the data; L.B., R.D., H.S., A.R. and M.C. contributed materials/analysis tools; L.B. wrote the paper.

Corresponding author

Correspondence to Lindell Bromham .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Peer review information Nature Ecology and Evolution thanks Ruth Oliver, Salikoko Mufwene, Claire Bowern and Hannah Wauchope for their contribution to the peer review of this work. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended data fig. 1 residual in the best model for language endangerment level..

Residuals of the model predicting number of endangered languages ( a ) and Sleeping languages ( b ), calculated, for each hex grid, as the predicted number of languages with distribution in the hex grid and with (A) predicted endangerment level above ‘Stable’ (corresponding to EGIDS 6b-10) and (B) predicted to be no longer spoken (ie EGIDS 9-10) minus the number of languages with distribution in the hex grid and with reported EGIDS from 6b-10 (A) and from 9-10 (B). The predicted number of languages in each category is calculated by using the best model to estimate the probability distribution of endangerment level for each language with distribution in the hex grid, sampling from the probability distribution the endangerment level of each language, repeating the sampling 1000 times, and averaging the number of languages with sampled endangerment level of endangered or Sleeping over the 1000 times. A negative value (blue) indicates that the model estimates fewer endangered or Sleeping languages than the reported EGIDS score from Ethnologue (e17/e16). A positive value (red) indicates the model estimating a greater number of endangered or Sleeping languages than observed. In some cases, this could indicate higher ‘latent risk’, for languages that have many of the predictors of high endangerment but are currently rated as stable or at a lower level of endangerment. Dark grey areas do not have data for all the independent variables in the best model for language endangerment level. Language distribution data from WLMS 16 worldgeodatasets.com .

Extended Data Fig. 2 Current and future predicted distribution of endangered languages.

The current patterns of language endangerment are plotted as absolute number of languages with a reported EGIDS score of 6b-10 with distribution in each hex grid. a ) the number of languages with observed EGIDS from 6b to 10 at present. b ) the predicted number of languages with EGIDS from 6b to 10 in 40 years minus the predicted number of languages with EGIDS from 6b to 10 at present. c ) the predicted number of languages with EGIDS from 6b to 10 in 80 years minus the predicted number of languages with EGIDS from 6b to 10 in 40 years. The predicted number of languages is calculated in the same way as Supplementary Fig. 7 . Dark grey areas have no data for independent variables in the best model for language endangerment level. Language distribution data from WLMS 16 worldgeodatasets.com .

Extended Data Fig. 3 Current and future predicted proportion of endangered languages.

a ) the proportion of languages with observed EGIDS from 6b to 10 at present. b ) the predicted proportion of languages with EGIDS from 6b to 10 in 40 years minus the predicted proportion of languages with EGIDS from 6b to 10 at present. c ) the predicted proportion of languages with EGIDS from 6b to 10 in 80 years minus the predicted proportion of languages with EGIDS from 6b to 10 in 40 years. The predicted proportion of languages is calculated as the predicted number of languages divided by the total number of languages with distribution in each hex grid, where the predicted number of languages is calculated in the same way as Fig. 7. Dark grey areas have no data for independent variables in the best model for language endangerment level. Language distribution data from WLMS 16 worldgeodatasets.com .

Extended Data Fig. 4 Current and future predicted number of languages no longer spoken.

a ) the number of languages with observed EGIDS from 9 to 10 at present. b ) the predicted number of languages with EGIDS from 9 to 10 in 40 years minus the predicted number of languages with EGIDS from 9 to 10 at present. c ) the predicted number of languages with EGIDS from 9 to 10 in 80 years minus the predicted number of languages with EGIDS from 9 to 10 in 40 years. The predicted number of languages is calculated in the same way as Fig. 7. Dark grey areas have no data for independent variables in the best model for language endangerment level. Language distribution data from WLMS 16 worldgeodatasets.com .

Extended Data Fig. 5 Current and future predicted proportion of languages no longer spoken.

The proportion of Sleeping languages with distribution in each hex grid. a ) the proportion of languages with observed EGIDS from 9 to 10 at present. b ) the predicted proportion of languages with EGIDS from 9 to 10 in 40 years minus the predicted proportion of languages with EGIDS from 9 to 10 at present. c ) the predicted proportion of languages with EGIDS from 9 to 10 in 80 years minus the predicted proportion of languages with EGIDS from 9 to 10 in 40 years. The predicted proportion of languages is calculated as the predicted number of languages divided by the total number of languages with distribution in each hex grid, where the predicted number of languages is calculated in the same way as Fig. 7. Dark grey areas have no data for independent variables in the best model for language endangerment level. Language distribution data from WLMS 16 worldgeodatasets.com .

Supplementary information

Supplementary information.

Supplementary Information.

Peer Review Information

Supplementary data.

Supplementary Data 1–4.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Bromham, L., Dinnage, R., Skirgård, H. et al. Global predictors of language endangerment and the future of linguistic diversity. Nat Ecol Evol 6 , 163–173 (2022). https://doi.org/10.1038/s41559-021-01604-y

Download citation

Received : 06 May 2021

Accepted : 27 October 2021

Published : 16 December 2021

Issue Date : February 2022

DOI : https://doi.org/10.1038/s41559-021-01604-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Le système meta ai donne un coup de pouce aux langues menacées - à condition que les humains ne soient pas oubliés.

Nature Africa (2024)

Meta’s AI system is a boost to endangered languages — as long as humans aren’t forgotten

Nature (2024)

Islands are engines of language diversity

• Lindell Bromham
• Keaghan J. Yaxley
• Marcel Cardillo

Nature Ecology & Evolution (2024)

Critical transitions in the Amazon forest system

• Bernardo M. Flores
• Encarni Montoya
• Marina Hirota

Descriptive ethnobotanical studies are needed for the rescue operation of documenting traditional knowledge

• Łukasz Łuczaj

Journal of Ethnobiology and Ethnomedicine (2023)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Researching Language Loss and Revitalization

• Living reference work entry
• First Online: 01 January 2016
• Cite this living reference work entry

• Leena Huss 5  

Part of the book series: Encyclopedia of Language and Education ((ELE))

646 Accesses

Language loss refers to a societal or individual loss in the use or in the ability to use a language, implying that another language is replacing it. Revitalization, in turn, is commonly understood as giving new life and vigor to a language that has been decreasing in use and is today a rapidly growing field of study. Both fields are highly multidisciplinary, drawing from linguistics, sociology, education, psychology, anthropology, political science, and other disciplines.

Since the 1990s, the research interest in endangered languages and consciousness of the need to contribute to their survival have grown among researchers, and numerous studies have been undertaken to present what has been done to curb language decline and to explain why some languages survive and others do not. Researchers have also tried to pinpoint the most relevant factors and the ways in which they interact. Still, to establish language revitalization more firmly as an independent field of study, more research and theorization are needed.

Many revitalization efforts are connected with ethnic revival movements as revitalization of the language is often seen as a crucial part of the overall ethnic revival. As a reaction to former forced assimilation and oppression, revitalization movements are often seen as ways to healing, redress, and empowerment. Therefore, a growing part of revitalization research is today being done by, or in close collaboration with, researchers and other members coming from the language communities themselves.

The chapter deals with research approaches in the field of language loss and revitalization, as well as challenges faced by scholars in this area.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Institutional subscriptions

Similar content being viewed by others

Ryukyuan Language Reclamation: Individual Struggle and Social Change

Language Endangerment: Diversity and Specificities of Native American Languages of Oklahoma

Aikio, M. (1988). Saamelaiset kielenvaihdon kierteessä. Kielisosiologinen tutkimus viiden saamelaiskylän kielenvaihdosta 1910–1980 [The Saami in the process of language shift. A sociolinguistic study on language shift in five Sámi villages during 1910–1980] . Helsinki: Suomalaisen Kirjallisuuden Seura.

Google Scholar  

Amery, R. (2000). Warrabarna Kaurna! Reclaiming an Australian language . Lisse: Swets and Zeitlinger.

Crawford, J. (2000). At war with diversity: U.S. language policy in an age of anxiety . Clevedon: Multilingual Matters.

Cruz, E., & Woodbury, A. C. (2014). Collaboration in the context of teaching, scholarship, and language revitalization: Experience from the Chatino. Language Documentation, 8 , 262–286.

Crystal, D. (2000). Language death . Cambridge: Cambridge University Press.

Book   Google Scholar  

Dorian, N. (1981). Language death: The life cycle of a Scottish Gaelic dialect . Philadelphia: University of Pennsylvania Press.

Dorian, N. (1993). Discussion note, a response to Ladefoged’s other view of endangered languages. Language, 69 (3), 575–579.

Article   Google Scholar  

Dorian, N. (1998). Western language ideologies and small-language prospects. In L. A. Grenoble & L. Whaley (Eds.), Endangered languages, language loss and community response (pp. 3–21). Cambridge: Cambridge University Press.

Chapter   Google Scholar  

Dorian, N. C. (2014). Small-language fates and prospects. Lesson of persistence and change from endangered languages [Collected Essays] . Boston: Brill.

Edwards, J. (1991). Multilingualism . Oxford: Routledge.

Edwards, J. (2002). Forlorn hope? In L. Wei, J.-. M. Dewaele, & A. Housen (Eds.), Opportunities and challenges of bilingualism . Berlin: Mouton de Gruyter.

Eisenlohr, P. (2004). Language revitalization and new technologies: Cultures of electronic mediation and the refiguring of communities. Annual Review of Anthropology, 33 , 21–45.

Fillmore, L. W. (1991). When learning a second language means losing the first. Early Childhood Research Quarterly, 6 , 323–346.

Fishman, J. (1966). Language loyalty in the United States; the maintenance and perpetuation of non-English mother tongues by American ethnic and religious groups . The Hague: Mouton.

Fishman, J. (1991). Reversing language shift: Theoretical and empirical foundations of assistance to threatened languages . Clevedon: Multilingual Matters Ltd.

Fishman, J., & García, O. (Eds.). (2011). Handbook of language and ethnic identity. The success-failure continuum and ethnic identity efforts (Vol. 2). New York: Oxford University Press.

Gal, S. (1979). Language shift: Social determinants of linguistic change in bilingual Austria . New York: Academic.

Granadillo, T., & Orcutt-Gachiri, H. A. (2011). Introduction. In T. Granadillo & H. A. Orcutt-Gachiri (Eds.), Ethnographic contributions to the study of endangered languages (pp. 1–12). Arizona: University of Arizona Press.

Grenoble, L. A., & Whaley, L. J. (Eds.). (1998). Endangered languages: Current issues and future prospects . Cambridge: Cambridge University Press.

Haugen, E. (1953). The Norwegian language in America: A study in bilingual behavior I–II . Philadelphia: University of Pennsylvania Press.

Henderson, J. (2013). Language documentation and community interests. In M. C. Jones & S. Ogilvie (Eds.), Keeping languages alive. Documentation, pedagogy, and revitalization (pp. 56–68). Cambridge: Cambridge University Press.

Hinton, L. (Ed.). (2013). Bringing our languages home. Revitalization for families . Berkeley: Heyday Books.

Hinton, L., & Hale, K. (Eds.). (2001). The green book of language revitalization in practice . San Diego: Academic.

Holton, G. (2011). The role of information technology in supporting minority and endangered languages. In P. K. Austin & J. Sallabank (Eds.), The Cambridge handbook of endangered languages . Cambridge: Cambridge University Press.

Hornberger, N. (Ed.). (2007). Can schools save indigenous languages? Policy and practice on four continents . Hampshire: Palgrave Macmillan.

Huss, L., & Lindgren, A.-R. (2011). Introduction: Defining language emancipation. In L. Huss & S. Nolan (Eds.), The many faces of language emancipation [Special Issue]. International Journal of the Sociology of Language , 209 , 1–15.

Huss, L., Camilleri Grima, A., & King, K. (2003). Transcending monolingualism. Linguistic revitalization in education . Lisse: Swets and Zeitlinger.

Jaakkola, M. (1973). Språkgränsen: en studie i tvåspråkighetens sociologi [Language border: A study in the sociology of bilingualism] . Stockholm: Aldus/Bonnier.

King, K. (2001). Language revitalization processes and prospects: Quichua in the Ecuadorian Andes . Clevedon: Multilingual Matters Ltd.

King, K. A., & Hermes, M. (2014). Why is it so hard?: Ideologies of endangerment, passive language learning approaches, and Ojibwe in the United States. Journal of Language, Identity & Education, 13 (4), 269–282.

Krauss, M. (1992). The world’s languages in crisis. Language, 68 (1), 4–10.

Kulick, D. (1992). Language shift and cultural reproduction: Socialization, self, and syncretism in a Papua New Guinea Village . Cambridge: Cambridge University Press.

Lambert, W. E. (1974). Culture and language as factors in learning and education. In F. E. Aboud & R. D. Mead (Eds.), Cultural factors in learning and education . 5th Western Washington Symposium on Learning, Bellingham.

Makoni, S., & Pennycook, S. (Eds.). (2007). Disinventing and reconstituting languages . Clevedon: Multilingual Matters.

May, S. (2014). Contesting metronormativity; exploring indigenous language dynamism across the urban-rural divide. Journal of Language Identity & Education, 13 (4), 229–235.

McCarty, T. L. (2002). A place to be Navajo: Rough rock and the struggle for self-determination in indigenous schooling . Mahwah: Erlbaum.

Norberg, M. (1996). Sprachwechselprozess in der Niederlausitz. Soziolinguistische Fallstudie der deutsch - sorbischen Gemeinde Drachhausen/Hochoza [Language shift process in Niederlausitz: A sociolinguistic case study on the German - Sorbian community Drachhausen/Hochoza] (Doctoral dissertation). Retrieved from ProQuest. (C503412).

Olthuis, M.-L., Kivelä, S., & Skutnabb-Kangas, T. (2013). Revitalising indigenous languages: How to recreate a lost generation . Bristol: Multilingual Matters.

Reyhner, J. (1999). Introduction: Some basics of indigenous language revitalization. In J. Reyhner, G. Cantoni, R. N. St. Clair, & E. P. Yazzie (Eds.), Revitalizing indigenous languages . Flagstaff: Northern Arizona University.

Roche, G., Ban+de mkhar, Bkra shis bzang po, G.yu lha, Snying dkar skyid, Tshe ring rnam gyal, Zla ba sgrol ma, Charles Kevin Stuart (2010). Participatory culture documentation on the Tibetan Plateau. In I. Gunn & M. Turin (Eds.), Language documentation and description: Vol. 8. Special issue on oral literature and language endangerment (pp. 140–158). London: SOAS.

Sicoli, M. (2011). Agency and ideology in language shift and language maintenance. In T. Granadillo & H. A. O. Gachiri (Eds.), Ethnographic contributions to the study of endangered languages , (pp. 161–176). Tucson: University of Arizona Press.

Skutnabb-Kangas, T. (2000). Linguistic genocide in education or worldwide diversity and human rights? Mahwah: Erlbaum.

Stiles, D. B. (1997). Four successful indigenous language programs. In J. Reyhner (Ed.), Teaching indigenous languages (pp. 148–262). Flagstaff: Northern Arizona University.

Stordahl, V. (1996). Same i den moderne verden. Endring og kontinuitet i et samisk lokalsamfun [To be Sámi in the modern world: Change and continuity in a Sámi community] (Unpublished doctoral dissertation). University of Tromsø, Norway.

Todal, J. (2002). “-jos fal gáhttet gollegielat”: vitalisering av samisk språk i Noreg på 1990-talet [“-if only you safeguard your golden language”: Vitalization of the Sámi language in Norway in the 1990s . Tromsø: University of Tromsø.

Weinreich, U. (1953). Languages in contact . The Hague: Mouton.

Download references

Author information

Authors and affiliations.

The Hugo Valentin Centre, Uppsala University, SE-751 20, Uppsala, Sweden

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Leena Huss .

Editor information

Editors and affiliations.

Second Languages and Cultures Program, University of Minnesota, Minneapolis, USA

Kendall King

University of Minnesota, Minneapolis, Minnesota, USA

Faculty of Education, University of Auckland, School of Critic, Auckland, New Zealand

Stephen May

Rights and permissions

Reprints and permissions

Copyright information

© 2016 Springer International Publishing Switzerland

About this entry

Cite this entry.

Huss, L. (2016). Researching Language Loss and Revitalization. In: King, K., Lai, YJ., May, S. (eds) Research Methods in Language and Education. Encyclopedia of Language and Education. Springer, Cham. https://doi.org/10.1007/978-3-319-02329-8_7-1

Download citation

DOI : https://doi.org/10.1007/978-3-319-02329-8_7-1

Received : 15 September 2015

Accepted : 08 August 2016

Published : 05 October 2016

Publisher Name : Springer, Cham

Online ISBN : 978-3-319-02329-8

eBook Packages : Springer Reference Education Reference Module Humanities and Social Sciences Reference Module Education

• Publish with us

Policies and ethics

• Find a journal
• Track your research

• Share full article

Can You Lose Your Native Tongue?

After moving abroad, I found my English slowly eroding. It turns out our first languages aren’t as embedded as we think.

Credit... Artwork by PABLO DELCÁN

Supported by

By Madeleine Schwartz

Madeleine Schwartz is a writer and editor who grew up speaking English and French. She has been living in Paris since 2020.

• May 14, 2024

It happened the first time over dinner. I was saying something to my husband, who grew up in Paris where we live, and suddenly couldn’t get the word out. The culprit was the “r.” For the previous few months, I had been trying to perfect the French “r.” My failure to do so was the last marker of my Americanness, and I could only do it if I concentrated, moving the sound backward in my mouth and exhaling at the same time. Now I was saying something in English — “reheat” or “rehash” — and the “r” was refusing to come forward. The word felt like a piece of dough stuck in my throat.

Listen to this article, read by Soneela Nankani

Other changes began to push into my speech. I realized that when my husband spoke to me in English, I would answer him in French. My mother called, and I heard myself speaking with a French accent. Drafts of my articles were returned with an unusual number of comments from editors. Then I told a friend about a spill at the grocery store, which — the words “conveyor belt” vanishing midsentence — took place on a “supermarket treadmill.” Even back home in New York, I found my mouth puckered into the fish lips that allow for the particularly French sounds of “u,” rather than broadened into the long “ay” sounds that punctuate English.

My mother is American, and my father is French; they split up when I was about 3 months old. I grew up speaking one language exclusively with one half of my family in New York and the other language with the other in France. It’s a standard of academic literature on bilingual people that different languages bring out different aspects of the self. But these were not two different personalities but two separate lives. In one version, I was living with my mom on the Upper West Side and walking up Columbus Avenue to get to school. In the other, I was foraging for mushrooms in Alsatian forests or writing plays with my cousins and later three half-siblings, who at the time didn’t understand a word of English. The experience of either language was entirely distinct, as if I had been given two scripts with mirroring supportive casts. In each a parent, grandparents, aunts and uncles; in each, a language, a home, a Madeleine.

I moved to Paris in October 2020, on the heels of my 30th birthday. This was both a rational decision and something of a Covid-spurred dare. I had been working as a journalist and editor for several years, specializing in European politics, and had reported across Germany and Spain in those languages. I had never professionally used French, in which I was technically fluent. It seemed like a good idea to try.

When I arrived in France, however, I realized my fluency had its limitations: I hadn’t spoken French with adults who didn’t share my DNA. The cultural historian Thomas Laqueur, who grew up speaking German at home in West Virginia, had a similar experience, as the linguist Julie Sedivy notes in “Memory Speaks,” her book about language loss and relearning her childhood Czech. Sedivy cites an essay of Laqueur’s in which he describes the first time he learned that German was not, in fact, a secret family language. He and his brother had been arguing over a Popsicle in front of the grocery store near his house:

A lady came up to us and said, in German, that she would give us a nickel so that we could each have a treat of our own. I don’t remember buying a second Popsicle, but I do remember being very excited at finding someone else of our linguistic species. I rushed home with the big news.

My own introduction to speaking French as an adult was less joyous. After reaching out to sources for a different article for this magazine with little success, I showed the unanswered emails to a friend. She gently informed me that I had been yelling at everyone I hoped to interview.

We are having trouble retrieving the article content.

Please enable JavaScript in your browser settings.

Thank you for your patience while we verify access. If you are in Reader mode please exit and  log into  your Times account, or  subscribe  for all of The Times.

Thank you for your patience while we verify access.

Already a subscriber?  Log in .

Want all of The Times?  Subscribe .

Advertisement

Many of the world's most remote languages are in danger of disappearing. Here, neighbors in the Altai mountains in China craft a new pair of skis. The range connects Russia, China, Mongolia, and Kazakhstan, making the threatened Altai language an unusual blend of dialects.

The Race to Save the World's Disappearing Languages

Every two weeks a language dies. Wikitongues wants to save them.

On a residential block at the border between Brooklyn and Queens, Gottscheer Hall appears like a mirage from 1945.

Blue awnings advertise the space for weddings and events. Inside, an entryway is covered with the saccharin smiles of “Miss Gottschee” contestants from decades past. “Back then you had to know the language to compete,” says 92-year-old Alfred Belay, pointing out his daughter’s beaming face from the 1980s. Nowadays, there are years with only a single contestant in the pageant.

Belay has been coming to Gottscheer Hall since he arrived in America more than 60 years ago. Then, the neighborhood was filled with refugees from Gottschee, a settlement that once occupied the highlands of modern-day Slovenia. Now, he’s one of a few thousand remaining speakers of its language, Gottscheerisch. Every Christmas he leads a service in his 600-year-old native language that few understand.

“Imagine if someone who plays music suddenly can’t use their fingers,” he says. “We’re still alive but can only remember these things.”

Belay and his sister, 83-year-old Martha Hutter, have agreed to let 26-year-old Daniel Bogre Udell film them having a conversation. They walk past the dark wood bar of Gottscheer Hall serving pretzels and sausages, and they climb the stairs to an empty banquet room. Bogre Udell sets up his camera and the siblings begin to banter in their inscrutable Germanic mother tongue.

Slip of the Tongue

Hearing such a rare language spoken on a residential block of Queens is not unusual for Bogre Udell, the co-founder of a nonprofit called Wikitongues . There are some 800 languages spoken within the 10-mile radius of New York City, which is more than 10 percent of the world’s estimated 7,099 languages. Since he has decided to record all of them, the melting-pot metropolis is a natural launching point.

Bogre Udell, who speaks four languages, met Frederico Andrade, who speaks five, at the Parsons New School in New York City. In 2014, they launched an ambitious project to make the first public archive of every language in the world. They’ve already documented more than 350 languages, which they are tracking online , and plan to hit 1,000 in the coming years.

“When humanity loses a language, we also lose the potential for greater diversity in art, music, literature, and oral traditions,” says Bogre Udell. “Would Cervantes have written the same stories had he been forced to write in a language other than Spanish? Would the music of Beyoncé be the same in a language other than English?”

Between 1950 and 2010, 230 languages went extinct, according to the UNESCO Atlas of the World’s Languages in Danger . Today, a third of the world’s languages have fewer than 1,000 speakers left. Every two weeks a language dies with its last speaker, 50 to 90 percent of them are predicted to disappear by the next century.

( Read about what happens when a language dies .)

In rare cases, political will and a thorough written record can resurrect a lost language. Hebrew was extinct from the fourth century BC to the 1800s, and Catalan only bloomed during a government transition in the 1970s. In 2001, more than 40 years after the last native speaker died, the language of Oklahoma’s Miami tribe started being learned by students at Miami University in Ohio. The internet has connected rare language speakers with each other and with researchers. Even texting has helped formalize languages that don’t have a set writing system.

An ethnic German from the region of Gottschee poses for a portrait in 1936. After World War II, the settlement was disolved and thousands of its inhabitants left for America. Today, few still speak the language.

Knowing they wouldn’t be able to record, or even locate, the majority of these languages themselves, Wikitongues has enlisted a network of volunteers in 40 countries to film native speakers talking in the past, present, and future tenses of their mother tongue. To get a range of tones and emotions, they’re asked to reminisce about childhood, talk about romance, and discuss their hopes and goals.

One volunteer in the South Pacific islands of Vanuatu recorded a language that had never before been studied by linguists. Another tracked down a speaker of Ainu , a rare indigenous language in Japan that is an “ isolate ,” meaning it bears no relation to any other known language.

Wikitongues isn’t the only initiative working to document rare languages. National Geographic Society’s Enduring Voices project supported the Living Tongues Institute for Endangered Languages in their effort to build Talking Dictionaries comprised of definitions, audio files, and images. Someone looking to learn Tuvan, a Turkic language spoken in Siberia, can download the app to their phone.

You May Also Like

The Māori saved their language from extinction. Here’s how.

Peru's Quechua rappers have the world taking notice

Hawaii’s Native language nearly vanished—this is the fight to bring it back

Starting this year, Wikitongue’s collections will be stored at the American Folklife Center through a partnership with the Library of Congress. But their goals stretch past documentation—the founders also plan to provide a way to learn languages long after they’ve gone extinct. An app they’re building called Poly allows people to create language dictionaries using text, audio, and video.

Missing Words

Priceless documentation opportunities disappear regularly. Not long ago, one of the last two speakers of a Saami language dialect in the Russian steppes died right before his recording session with Wikitongues. Some 500 languages could slip through their grasp in the next five years, they estimate.

Political persecution, a lack of preservation, and globalization are to blame for the dwindling language diversity. For much of the 20th century, governments across the world have imposed language on indigenous people, often through coercion. Some 100 aboriginal languages in Australia have disappeared since European settlers arrived. A half-century after China annexed Tibet, dozens of distinct dialects with unique alphabets are on the verge of extinction. Studies have shown that suppressing language impairs everything from health to school performance.

This forced suppression, however, is no longer the biggest threat facing our linguistic ecosystem. “Most languages die today not because of abject and outright persecution—though this does happen on occasion—but rather because they are made unviable,” says Andrade. Factors like climate change and urbanization force linguistically diverse rural and coastal communities to migrate and assimilate to new communities with new languages.

“This form of language loss is a cancer, not a gunshot.”

A group of indigenous Ainu people in Japan sit together for a photograph in the 1880s. Linguists have been unable to find any other language in the world that resembles Ainu.

For Future Generations

In Gottscheer Hall, Belay and Hutter transform as they chatter for Daniel Bogre Udell’s video camera. At one point Hutter breaks into song. In Gottscheerisch, they recall growing up in a single bedroom home where they spoke Gottscheerisch—German was used for school and church.

In 1941, Gottschee was annexed by the Italians and its residents were sent to resettlement camps. Four years later, the Gottscheer Relief Association opened its doors to the thousands of immigrants arriving in New York. By the time Belay and Hutter arrived, in the 1950s, the neighborhood was so full of immigrants that Hutter was barely able to practice her English.

The newcomers spoke Gottscheerisch to each other and raised their kids with English. Now, 60 years later, Belay has started speaking to his kids in Gottscheerisch for the first time, but the language is on the brink of extinction.

As a street language, Gottscheerisch was rarely written down. It could only be learned by ear until 1994, when Hutter published a five-year effort collecting definitions for 1,400 words: the first English-Gottscheerisch dictionary.

“The old Gottscheers were convinced that nobody can learn Gottscheerisch, so they didn’t try to teach it,” Hutter recalls. “But any language can be learned, so I thought, ‘This old language is going to die and they won’t know anything.’”

“We did the same thing,” Belay interjects. “Our kids could have learned it.”

“There is a time in the future when families won’t speak it,” says Hutter. “When they’ll say, ‘Our family spoke— what was it ?’”

Related Topics

• ENDANGERED LANGUAGES

Hebrew wasn’t spoken for 2,000 years. Here’s how it was revived.

Why it’s never too late to learn a language as an adult

How scientists are piecing together a sperm whale ‘alphabet’

North America’s Native nations reassert their sovereignty: ‘We are here’

How did Viagra become ‘Viagra’? Through this fascinating process

• Environment
• Paid Content

History & Culture

• History & Culture
• Destination Guide
• Terms of Use
• Privacy Policy
• Your US State Privacy Rights
• Children's Online Privacy Policy
• Interest-Based Ads
• About Nielsen Measurement
• Do Not Sell or Share My Personal Information
• Nat Geo Home
• Attend a Live Event
• Book a Trip
• Inspire Your Kids
• Shop Nat Geo
• Visit the D.C. Museum
• Learn About Our Impact
• Support Our Mission
• Advertise With Us
• Customer Service
• Renew Subscription
• Manage Your Subscription
• Work at Nat Geo
• Sign Up for Our Newsletters
• Contribute to Protect the Planet

Copyright © 1996-2015 National Geographic Society Copyright © 2015-2024 National Geographic Partners, LLC. All rights reserved

New Times, New Thinking.

• The Weekend Essay

In search of a language of loss

Is there a correct way to mourn? When my mother died, I scoured the literature of grief for answers

By Johanna Thomas-Corr

The night before my mother died, she threw a cocktail party. It was her 73rd birthday and she knew she was dying. The 20 or so friends who gathered in her flat knew she was dying, too. But Mum didn’t want to talk about it. She insisted on manoeuvring her stiff, shrunken arms into a sparkly jumper and backcombing her hair the way it was when she was a teenager. She wanted to drink Gin & Its – or more precisely, she wanted us to drink Gin & Its, since all she could manage were tiny sips of mango juice.

By lunchtime the next day, all the guests had trickled away. Mum’s best friend, my brother and his wife had left Bristol for London. My husband had also left to take care of our two young children. “I can come back with the boys,” he said, to which I replied, a little too tersely: “One-year-olds and syringes do not mix.” In truth, I was scared. I had never seen anyone die. Now I was alone with my mother, an oxygen machine and a fridge full of leftover party food.

I had given birth only the summer before, and the experience of watching her face death had much in common: my mother’s emphasis on breathing, her look of helpless pleading and then supreme focus. She was in the zone. She was barely able to speak but her face flashed with past, more energetic selves: the school running champion winning the county race; the English teacher holding court; the political firebrand addressing the rally.

We were lucky: we had managed to get Mum out of hospital – apart from everything else, it meant that she could retain her near-religious faith in the NHS – but we had no professional care in place. Just friends. Julia, a former nurse, came by to help me administer morphine. My friend Sarah, a doctor, drove over with a suitcase of my clothes so I could stay the night. My mother, sitting on the sofa, one leg tucked protectively underneath her, looked desperately relieved to see her. We stepped out of the room to discuss how to move Mum into her bedroom – and in these two or three minutes, she died. Lung failure, heart failure.

We toasted her with prosecco and some macaroons she had been given as a birthday present. It was the best way to remember her, Julia and Sarah agreed. It was also a sedative for the shock of losing your mother on a cold December night. Then the three of us lifted Mum’s tiny, contorted body from the sofa to her bedroom. Julia’s jumper snagged on a door handle. “Hang on! Stop! We need to reverse.” They didn’t want to be the first to laugh so it was down to me to get the giggles. “Mum would have found this hilarious.”

The Saturday Read

Morning call.

Would have. She had been dead less than an hour. I now realised that the last thing she would have seen were her still-wrapped birthday presents, piled up around the fireplace.

“The layers of loss make life feel papery thin,” writes the novelist Chimamanda Ngozi Adichie in Notes on Grief (2021), her spare, lyrical outpouring of sorrow. Adichie’s 88-year-old father, James Nwoye Adichie – Nigeria’s first professor of statistics – died suddenly when she was thousands of miles away in the US and couldn’t get back. Two aunts died shortly afterwards, prompting a “sensation of eternal dissolving”. For Adichie, the physicality of her own grief also recalled the experience of birth, of pushing out a child.

I was given Adichie’s book by a friend of Sarah’s who had lost her mother to Covid-19 ; she had heard that I had lost my mother and grandma within a few months of each other. This is the kind of activity that goes on among the bereaved, I discovered, the passing of books, poems, lines. I suppose it’s because we are searching desperately for language to describe the peculiar transition from being to not being – metaphors and similes that borrow imagery from something generally understood (a tidal wave, a cliff edge, a dark tunnel) to help explain something bewildering. And it was books that provided the bond between me and Mum, who had been an English teacher, and didn’t feel the need for many other possessions.

[See also: In this time of conflict, I recall my late mother’s Christian wisdom and kindness ]

In her memoir about the death of her father Lost & Found (2022), the journalist Kathryn Schulz uses a wonderfully vivid metaphor to explain how sorrow is assailed by something more complex: her grief, she writes, manifested as anxiety, irritability and distraction; her sadness was more “furtive” and “vulnerable”, like “a small neutral nation on a bellicose continent whose borders were constantly overrun by more aggressive emotions”. Like me, she longed to experience a purer grief, yearning for “moments when my sorrow ran through me like a river at night, dark and clear, untainted by any more insidious emotion”. Yet, as she acknowledges, “such things aren’t responsive to our wishes”.

After my mother’s death, it was Adichie’s depiction of loss as “an erosion” that resonated. “Grief is a cruel kind of education,” she writes. “You learn how much grief is about language and the grasping for language.” I kept reaching for my own figures of speech, only for them to writhe out of my hands. It was as if a building had been demolished, leaving a gap in the skyline. It was a tape that had… just stopped. Writing about her was easy: she was so distinctive. But writing about my relationship with her – this was a slippery business.

My mum, Annie Thomas, wasn’t famous, but she was well-known to the thousands of children she had taught, and to seemingly everyone involved in left-wing politics in Bristol. Born to working-class parents in Gosport in 1948 – the same year as her beloved NHS – she was a Labour councillor when I was growing up, known as “Red Annie” in the local press, and thrown out of the party for disobedience in the early 1990s. She remained fiercely committed to feminist and socialist causes – one of my childhood holidays took place at a Socialist Workers’ conference in Skegness – and eventually returned to the fold as chair of the Bristol West Constituency Labour Party when Jeremy Corbyn became leader. She was, as many of her friends said, a fighter, a powerhouse, “the best MP we never had”.

You didn’t want to get into an argument with my mum. She was courageous and uncompromising. My childhood memories involve picket lines of all kinds, Free Nelson Mandela T-shirts, Mum appearing on the local news in a shock of peroxide, declaring she was prepared to go to jail over the poll tax. She did all this while bringing up me and my brother alone, on a teacher’s salary, having divorced my father when I was three. One of the final things I said to her, minutes before she died, was, “Mum, you are so brave.”

But she could be tough on me. I was her eldest child, her only daughter. If there was a misplaced word or an inflection that didn’t quite align with her mood, there would be a flare-up. She often thought I was sneering at her, and I felt she was trying to control me. In The Cost of Living (2018), the second part of her “living memoir”, Deborah Levy writes about the death of her mother, a woman who was (like mine) “braver in her life than I have ever been” but who (also like mine) “gave me a hard time, beyond the call of a dutiful daughter”. Levy quotes Marguerite Duras saying she believes “the mother represents madness. Our mothers always remain the strangest, craziest people we’ve ever met.” But, Levy adds: “Now I can see that I did not want to let her be herself, for better or worse.”

That was true of me, too. Mum was forthright, rebellious, embarrassing. She once smuggled us onto a train from Prague to Paris without paying and made us hide from the inspectors. She was a conscientious objector to all domestic tasks, and one Christmas fell out with a friend who bought her a recipe book. At parties, she got up on chairs to dance and when she went to the theatre, which she loved, she would buy the cheapest standing tickets – before making a dash for an unoccupied seat in the stalls after the interval.

And yet she developed an extensive network of devoted friends. She was single for the last 35 years of her life but one of the least lonely people I knew, having dispersed any need for companionship, affection, humour, intimacy and adventure among dozens of friends. “Life was always exciting with Annie,” one wrote to me after she died. My husband adored her. To my children, she was the most radiant and magical grandmother. But what was she to me? Levy describes her mother’s body as her first “landmark”, a disorienting thing to lose. “Most childhood memories were twinned with her presence on earth.” To me, my mother was beautiful, fierce, familiar. But how much easier to remember her as she was to others.

Even the undertaker, it turned out, had known her. “You’ll want to do something that includes all her friends in politics,” she said. And I did feel this responsibility deeply. When two of her comrades refused to attend her funeral, preferring to commemorate her in their own way, I fretted to one of her older friends that I was misremembering her somehow. “No, Jo,” she replied. “She was your mum.”

[See also: Nick Cave: I don’t think art should be in the hands of the virtuous ]

She repeated this phrase again and again until the strange, incantatory rhythm broke down my defences. Someone gently led her away when I started crying. It took me back to being a child in the dark at night, listening for the sound of my mother’s car returning home from a Labour Party meeting. Waiting for her to open the garage door beneath our house, greet the babysitter, put the kettle on. How long would it take her to come upstairs? Would she go to my bedroom first? She was your mum. She was your mum.

In Clover Stroud’s The Red of My Blood (2022), about the death of her sister, Nell Gifford, the leader of a circus troupe, the author wonders what is an “appropriate” way to remember the deceased. She admits to being transfixed by her 46-year-old sister’s “golden” dead body. She wonders if taking photos of it was “extremely weird”, yet was happy to have that image on her phone whenever she felt strong enough to look at it.

I did the same, partly for my brother (who then told me he never wanted to see it), partly for myself. A month later, when my best friend’s father died in India, she sent a photo of him wrapped in cotton lying in a refrigerated casket on his porch. I sent back the image of my mum covered in a patchwork quilt made by a friend, next to the roses Sarah placed beside her. We had both loved each other’s parents.

That night, I held Mum for as long as I could bear. Too often, I had flinched from her, anticipating some kind of criticism, but now I focused on her hands, the only part of her body that still felt familiar. We used to hold hands while watching the news. We held hands when I was in labour with my first son. It was a relief that she had gone after so many years of pain, a relief she hadn’t died in a hospital corridor, a relief that I didn’t have to preside over another spiralling situation.

In Ian McEwan’s novel, Lessons (2022) , his protagonist Roland expresses a similar desire to hasten the passing of his mother. He struggles to remain present at the deathbed. Watching her stooped figure in the care home, he supposes “he would not see her alive again but after ten minutes that did not stop him wanting to leave. On the contrary. As he saw it she was already dead and he was already grieving but could not do it in her presence.” Lessons was perhaps the first novel published after my mum’s death that I knew she would have read and discussed with me – not least because it followed a man also born in 1948 to a working-class family. It was both cathartic and agonising to read, a reminder that the baby boomers weren’t, as we had once joked, immortal.

Five years before, when my mother was seriously ill, I had read Emily Berry’s poetry collection, Stranger, Baby (2017). Berry’s mother died when she was a child, and one line haunted me: “A mother’s death lasts a lot of years”. When I realised my mother would never get better, and that she would always remain inscrutable on the subject of her own death, I began privately rehearsing the moment I would lose her – often in the shower where my family couldn’t hear me crying. I would repeat to myself: “My mother is dying, my mother is dying.” I began reading more about death because I could see it coming.

And yet I’m not sure how useful this anticipatory grief is. For some time after Mum died, I still woke up thinking: is that it? All the richness and noise and struggle and force of personality. She must have had some kind of plan? And even if she has gone, surely this elaborate enterprise she had got up and running – all these people, all these beautiful and strange connections – isn’t just going to pack up and go home?

I felt a pressure to preserve her world, to maintain all the conversations, to hire some sort of brilliant stage manager to “keep the ball rolling” as Joseph Conrad had it in his novel Chance . Unlike my best friend’s father, an artist and writer, she left no material legacy. I have her books, some newspaper cuttings and a cup we had made with her face on it. Every now and then, when I go to reheat my coffee, I am faintly aware that I am pulling my mother out of the microwave. Would she approve? Am I doing this right?

In Sheila Heti’s novel Motherhood (2018) her narrator reflects on the inevitable entanglements with our mothers and grandmothers, and advises women not to struggle against them. “How far beyond your own mother do you hope to get?… Let the pattern which is the repeating, which was your mother, and her mother before her, live it a little bit differently this time. A life is just a proposition you ask yourself by living it, Could a life be lived like this, too?”

By far the weirdest aspect of grieving is how much I have come to like images of myself, simply because they remind me of her. I used to be so self-conscious – and the vain part of me is horrified by how much grief has aged me (Schulz calls it: “all at once, one very large step into oblivion”). But I rather like the fact I now look a bit like my mother did. I find I am not fighting it.

Johanna Thomas-Corr is literary editor of the Sunday Times

[See also: Kenneth Roth: “My biggest concern is for academic freedom ]

Content from our partners

Energy Security in the Balance: Net zero and reliable supply

More Than a Landlord: The future of social housing

Clean power 2030? Mission accepted

Sally Rooney’s adventures in style

In London’s warehouse district, the fauxhemian thrives

Keir Starmer’s union problem

This article appears in the 25 Jan 2023 issue of the New Statesman, Why Germany doesn’t do it better

You are using an outdated browser. Please upgrade your browser to improve your experience.

DJ Moore addresses his body language in Bears’ loss

Aug 12, 2023; Chicago, Illinois, USA; Chicago Bears wide receiver DJ Moore (2) leaves the field at halftime at Soldier Field. Mandatory Credit: David Banks-USA TODAY Sports

Chicago Bears wide receiver DJ Moore is doing some cleanup after his negative body language was caught on camera during the team’s loss Sunday to the Houston Texans.

Moore looked very frustrated throughout much of the Bears’ 19-13 loss in Houston, which many felt was a sign of frustration with quarterback Caleb Williams and the passing game. A sullen-looking Moore was captured on the bench several times, and he also sparked speculation by seemingly barking something in the direction of backup quarterback Tyson Bagent during the game.

D.J. Moore was not having it all last night. All moments I caught combined. pic.twitter.com/mcId7tBDtu — ✶ Ⓜ️𝕒𝕣𝕔𝕦𝕤 ▶️ ✶ (@_MarcusD3_) September 16, 2024

Moore said Wednesday that he was frustrated, but not at Williams or anyone specific. He said he did some “self-evaluation” after seeing the video and regretted letting his frustrations show as much as they did.

“I shouldn’t have shown as much,” Moore said, via Courtney Cronin of ESPN . “But it’s a part of the game. Like I said, we were one play away from the game changing. And we just couldn’t connect, nobody on offense could connect with the one play or get the one play started to get us on track and go up. It’s football. Everybody is going to have their ups and downs and frustrations.”

Moore added that his demeanor on the bench was not actually different from what it usually is.

“I usually sit on the bench like that anyway, so I don’t know what that’s about,” Moore said.

Chicago’s frustrating night on offense opened the door to this kind of speculation. Fairly or not, some also fixated on Moore because he initially called for the Bears to keep Justin Fields as their quarterback and do something else with the No. 1 pick. That generated a narrative that Moore did not want Williams to begin with, though the wide receiver seemed to come around during the offseason.

Moore caught six passes for 53 yards in the defeat and was targeted 10 times. His numbers were hurt by the fact that the Bears really struggled to protect Williams during the game, which prevented the rookie quarterback from making the most of his weapons.

Article Tags

Subscribe and Listen to the Podcast!

Related Articles

Cowboys’ blowout loss led to hilarious mistake by FEMA

The Dallas Cowboys’ Week 2 loss to the New Orleans Saints was so bad, it prompted a federal government agency to accidentally declare a major disaster. The Saints blasted the Cowboys 44-19 in Dallas on Sunday in a shocking result…

ULM defensive coordinator has confident take about facing Arch Manning

Texas quarterback Arch Manning will get his first collegiate start on Saturday against Louisiana-Monroe, but the defensive coordinator tasked with stopping him does not appear to be losing sleep over it. ULM defensive coordinator Earnest Hill praised Manning’s ability on…

Justin Jefferson makes guarantee about his Week 3 status

Minnesota Vikings wide receiver Justin Jefferson gave the team a scare with his early exit in Week 2, but he is bullish about his Week 3 status. Jefferson essentially guaranteed on Thursday that he would be suiting up for Week…

Could Bryce Young ask Panthers to trade him?

Bryce Young was benched by the Carolina Panthers, leaving his future with the team uncertain. While the Panthers keep saying they will keep him, there are some questions regarding how Young feels about the organization right now. ESPN’s David Newton…

Another 49ers star could miss Week 3 due to injury

The San Francisco 49ers are at risk of being without another star offensive player for Week 3. Tight end George Kittle is experiencing hamstring tightness and did not practice Thursday, according to Adam Schefter of ESPN. That practice absence puts…

Haason Reddick breaks social media silence with cryptic post

Haason Reddick is still away from the New York Jets as they prepare for their third game of the season, but the star pass-rusher has broken his lengthy silence on social media. Reddick posted on his X account on Thursday…