← Endangered language

Version 3 (current)

Nov 13, 2026. Version history

An endangered language is a language at risk of falling out of use, usually because its speakers are no longer passing it on to their children and are shifting to another, more widely spoken language.[1][3]

Experts writing for UNESCO in 2003 defined the term plainly: a language is endangered when it is on a path toward extinction, and it is in danger when its speakers stop using it, use it in fewer and fewer areas of life, and stop passing it from one generation to the next.[1] Estimates of how many languages are affected depend on how they are counted, but a 2021 global statistical study found that 37 percent of the 6,511 spoken languages it examined were threatened or worse.[3]

What makes a language endangered

A language rarely disappears because all of its speakers die. More often, people keep living but change the language they speak. The authors of a 2021 study in Nature Ecology & Evolution stressed that language loss often occurs when people shift from using their heritage language to a different language, rather than through a fall in the number of people in a community, and that not all small languages are endangered.[3] Linguists call this process language shift.

The UNESCO experts' report describes both outside and inside pressures. Endangerment, it says, can result from military, economic, religious, cultural, or educational subjugation, or from a community's own negative attitude towards its language, though such internal pressures often have their source in external ones.[1] The report observes that many Indigenous peoples, associating their disadvantaged social position with their culture, have come to believe that their languages are not worth keeping, and give them up in the hope of escaping discrimination, earning a living or improving their social standing.[1] It adds that people in this position have often been offered an either-or choice between keeping their mother tongue and identity or getting on in life, whereas, in the experts' view, maintaining and using both languages gives better chances still.[1]

Earlier, in 1992, the linguist Michael Krauss had listed the circumstances that had led to language loss as ranging from outright genocide, social, economic or habitat destruction, displacement and "demographic submersion" to language suppression through forced assimilation or assimilatory education.[6] To these he added electronic media, especially television, which he called "cultural nerve gas".[6]

Size matters too. About 97 percent of the world's people speak about 4 percent of its languages, the same report notes, citing the anthropologist H. Russell Bernard, while about 96 percent of the world's languages are spoken by about 3 percent of the world's people.[1] A small community is more exposed to disease, war or natural disaster, and may merge with a neighbouring group and lose its language in the process.[1] Even so, numbers alone are not decisive: even languages with many thousands of speakers, the report says, are no longer being learned by children.[1] Krauss gave two examples: Breton, with perhaps a million speakers within living memory but by then very few child speakers, and Navajo, with well over 100,000 speakers a generation earlier but an uncertain future.[6]

Writing in 2013 about language loss in the digital age, the researcher András Kornai summarized three signs that a language is approaching death well before its last speaker dies: loss of function, as other languages take over whole areas such as commerce; loss of prestige, seen most clearly in the attitudes of the young; and loss of competence, seen in "semi-speakers" who understand the older generation but use a much simplified version of the grammar.[15]

History of the concept

Linguists had studied language shift and "language death" long before the subject reached a wide audience; Lewis and Simons describe an extensive literature on its causes, processes, symptoms and results.[2] A widely cited starting point for the modern concern was a symposium on endangered languages at the annual meeting of the Linguistic Society of America in January 1991, whose papers were published together in the society's journal Language in March 1992.[6][7]

The first essay in that collection, "The world's languages in crisis", was by Michael Krauss of the Alaska Native Language Center at the University of Alaska Fairbanks.[6] Krauss called a language that children no longer learn as their mother tongue "moribund", since unless the course was dramatically reversed it was already doomed; a language still being learned by children, but likely to stop being learned during the coming century, merely "endangered"; and the rest "safe".[6] He noted that, unlike a biological species, a language is potentially revivable under certain conditions, as the case of Hebrew showed.[6]

His figures were stark. In Alaska only 2 of the 20 Native languages were still being learned by children; across the United States and Canada he counted 149 of 187 Native languages, or 80 percent, as moribund; and in Australia he put 90 percent of the roughly 250 Aboriginal languages still spoken in that category.[6] He judged the situation in Central and South America to be much better, with about 17 percent of Meso-American and 27 percent of South American Indigenous languages likely to be moribund.[6] Taking 6,000 as a round figure for the world's languages, and counting as "safe" only those with official state support or at least about 100,000 speakers, he arrived at perhaps 600 safe languages, and noted that the median language had only 5,000 or 6,000 speakers.[6] He concluded that, at the rate things were going, the coming century would see "either the death or the doom of 90% of mankind's languages".[6] Krauss contrasted this with the 7.4 percent of mammal species and 2.7 percent of bird species then listed as endangered or threatened, and asked why linguists were so much quieter about the threat to linguistic diversity than biologists were about biological diversity.[6]

Writing two decades later, Gary Simons and M. Paul Lewis of SIL International called Krauss's essay "the great linguistic call to arms" and noted that its warning about the possible loss of 90 percent of languages had been repeated again and again.[7] Not every linguist agreed. In a later 1992 issue of Language, Peter Ladefoged argued, as H. Russell Bernard summarized the exchange, that in places such as Tanzania, where the authorities encouraged Swahili as a means of building the nation, it would not be responsible for linguists to work for the preservation of local languages, and he described a speaker of Dahalo who was proud that his teenage sons did not speak it.[8] The linguist Nancy Dorian replied in 1993 that much language death takes place in democracies, where economic pressures rather than political ones discourage parents from passing on their language, and where working for language diversity is not a dangerous political act.[8]

UNESCO had earlier run a project called The Red Book of Languages in Danger of Disappearing, to gather information on endangered languages and encourage research on them; the 2003 expert group, of which Krauss was a member, criticized it for lacking the goal of working with communities themselves on maintaining and revitalizing their languages.[1] In October 2001 UNESCO's General Conference had unanimously adopted the Universal Declaration on Cultural Diversity, which recognized a relationship between biodiversity, cultural diversity and linguistic diversity.[1]

Measuring endangerment

Because no single figure captures a language's health, linguists use graded scales. Three prominent ones were brought together in 2010.[2]

Fishman's GIDS. The sociolinguist Joshua Fishman's Graded Intergenerational Disruption Scale, set out in his 1991 book Reversing Language Shift, ranks a language in eight levels according to how far its transmission between generations has been disrupted.[2] In 2010 the linguists M. Paul Lewis and Gary F. Simons called it the seminal and best-known evaluative framework of language endangerment of the previous two decades.[2]

UNESCO's nine factors. The 2003 UNESCO framework, Language Vitality and Endangerment, was prepared by an ad hoc expert group and submitted to an international expert meeting in Paris on 10 to 12 March 2003.[1] It uses six factors to assess a language's vitality, two to assess attitudes towards it and one for the urgency of documenting it.[1] The six vitality factors are [intergenerational transmission, the absolute number of speakers, the proportion of speakers in the total population, trends in the areas of life where the language is used, its response to new areas and media, and the materials available for education and literacy](ich.unesco.org/doc/src/00120-EN.pdf).[1] The attitude factors cover government and institutional policy, including official status, and the attitudes of community members towards their own language; the last factor is the amount and quality of documentation.[1] The report insists that none of the factors should be used alone.[1] It also warns against simply adding up the grades, and suggests the factors be examined according to the purpose of an assessment, whether a community's own self-assessment or an outside evaluation by other bodies.[1]

Its first factor, transmission between generations, gives the scale that is most often quoted:

  • Safe: spoken by all generations, with no sign of threat.[1]
  • Unsafe (sometimes called vulnerable): most but not all children speak it as their first language, perhaps only in some settings such as the home.[1][2]
  • Definitively endangered: children no longer learn it at home, so the youngest speakers are of the parents' generation.[1]
  • Severely endangered: spoken only by grandparents and older people.[1]
  • Critically endangered: the youngest speakers are great-grandparents, who may remember only part of the language and have no one to speak it with.[1]
  • Extinct: no one can speak or remember the language.[1]

The report adds that even "safe" does not guarantee a language's future, because speakers may stop passing it on at any time.[1] It also stresses that multilingualism is a fact of life in most of the world and does not in itself lead to language loss; what matters is that the language keeps a meaningful role in culturally important areas of life.[1]

The Expanded GIDS. In 2010 Lewis and Simons, later co-editors of Ethnologue, the catalogue of the world's languages published by SIL International, aligned Fishman's scale, UNESCO's and Ethnologue's own categories into a single scale of 13 levels, the EGIDS.[2][7] Its upper levels run from international use (level 0) through national, regional, trade, educational and written use (level 5), and its lower levels mirror UNESCO's: "Threatened" (level 6b) corresponds to UNESCO's "vulnerable", and "Shifting", "Moribund" and "Nearly Extinct" to UNESCO's definitely, severely and critically endangered.[2] Unlike Fishman's original, the EGIDS can place a language that no longer has any speakers.[2] A language's level is found by answering five key questions, about its identity function, its use as a language of wider communication, whether it is being passed on to children, how literacy in it is acquired, and which is the youngest generation of proficient speakers.[2] For the 17th edition of Ethnologue, published in 2013, its editors used the scale to estimate a level for every one of the 7,480 living and extinct languages the catalogue tracked, the first time the catalogue had rated the vitality of every language.[7]

"Extinct", "dormant" and "sleeping"

How to describe a language that no one speaks as a first language has become a question in its own right. Lewis and Simons recorded that Ethnologue received a volume of correspondence from members of ethnic groups who objected to the label of "extinct" for their heritage languages.[2] Its 16th edition therefore introduced the category "Dormant" for a language that is no longer spoken but still serves as a marker of identity for a living community, keeping "Extinct" for a language with which no one still identifies.[2] In the EGIDS, a dormant language (level 9) is one in which no one has more than symbolic proficiency; Lewis and Simons noted that revitalization may already be under way or contemplated, and that documentation of a language approaching extinction can make it possible to recover at least that level later if the community wishes.[2] Bromham and her co-authors used the word "Sleeping" for languages with no first-language speakers, the term preferred by many speakers of endangered languages.[3]

For languages coming back into use, Lewis and Simons proposed separate labels for the same levels, reflecting movement upwards rather than downwards: "Rediscovered" (level 9), when adults take up the language again for symbolic and identity purposes; "Reintroduced" and "Reawakened" (levels 8b and 8a), as adults bring it back into everyday interaction and children begin to acquire it in community and some home settings; and "Revitalized" and "Re-established" (levels 7 and 6b), as a second and then a third generation of children acquire it at home.[2]

How many languages are at risk

Counts differ because sources list different numbers of languages and draw the line between "threatened" and "safe" in different places.

  • Krauss, 1992. Krauss estimated that only about 10 percent of the world's languages were safe in the long term and that as many as half might already be moribund.[6][7]
  • UNESCO, 2003. The expert group wrote that at least 50 percent of the world's more than 6,000 languages were losing speakers, and estimated that in most regions about 90 percent of the languages may be replaced by dominant languages by the end of the 21st century.[1]
  • Ethnologue, 2009. Lewis and Simons reported that the edition then current, published in 2009, was the first in the catalogue's history of more than 50 years in which the number of identified living languages went down: 91 languages were recorded for the first time as having no known remaining speakers.[2] Of the 6,909 living languages then listed, 457 were classed as Nearly Extinct, and 3,524, just over half, had fewer than 10,000 speakers.[2]
  • Ethnologue, 2013. Rating every language for the 17th edition, Simons and Lewis found that 63 percent were safely maintained in everyday spoken use or stronger, 32 percent were in some stage of loss or shift, and 5 percent of the languages in use in 1950 were completely extinct.[7] Overall, 19 percent of the world's living languages were no longer being learned by children.[7]
  • Kornai, 2013. A study of the languages present online began from the consensus that some 2,500 of the approximately 7,000 languages then spoken were considered endangered, and argued that this figure greatly understated the danger in the digital sphere.[15]
  • Bromham and others, 2021. Their study of 6,511 spoken languages, published online on 16 December 2021, put 37 percent of them in the category threatened or worse, and 13 percent of those were no longer spoken.[3] They described the approximately 7,000 documented languages as nearly half endangered, and noted that the most pessimistic predictions in the literature suggested 90 percent would be lost within a century.[3] They also cited estimates that languages were being lost at a rate equivalent to one every one to three months.[3]
  • United Nations. As of November 2026, the UN's Department of Economic and Social Affairs states that optimistic estimates suggest at least 50 percent of today's spoken languages will be extinct or seriously endangered by 2100, while more pessimistic estimates put the figure at 90 to 95 percent, most of them Indigenous languages.[4] The same page suggests that humanity may have only 300 to 600 spoken languages left unthreatened by the end of the century.[4]

Where and why: the 2021 global study

The study led by Lindell Bromham of the Australian National University analysed 6,511 spoken languages against 51 predictor variables, covering population, documentation, legal recognition, education policy, socioeconomic indicators and environmental features.[3] Its main findings were these:

  • Number of speakers was the strongest single predictor of endangerment.[3]
  • Contact with other languages was not in itself a driver of loss, contrary to common perception; languages in direct contact with a greater number of other local languages had lower average endangerment levels.[3]
  • Road density was associated with greater endangerment, which the authors suggested may reflect the population movement that roads encourage.[3]
  • Schooling. Higher average years of schooling was associated with greater endangerment, which the authors called evidence that formal education can contribute to loss of language diversity, particularly where bilingual education is not supported.[3] They also found that having a minority education policy was not, globally, associated with lower threat, possibly because such policies are carried out in very different ways.[3] As an example they cited the United States Bilingual Education Act of 1968, which was designed mainly to use a child's heritage language as a bridge to English rather than to help the child keep the first language.[3]

The areas with the most endangered languages in absolute numbers were New Guinea, Central America, the Himalayas and Yunnan, and the region between Central and West Africa, which the authors attributed largely to how many languages those areas hold.[3] The areas with the highest proportion of their languages endangered included Australia, North China, Siberia, North Africa and Arabia, North America and parts of South America, and the greatest loss to date had been in Australia, South America and the United States.[3]

Looking ahead, the model predicted that without intervention language loss would at least triple within 40 years, amounting to at least one language lost per month, and that by the end of the century at least 1,500 languages would cease to be spoken.[3] The greatest absolute losses were predicted for the west coast of North America, Central America, the Amazon rainforest, West Africa, the north coast of New Guinea and northern Australia, and the greatest proportional losses for the Arctic, the interior plains of North America, temperate southern Chile and the Sahara.[3] The authors stressed that these predictions were "not death knells" but possible outcomes in the absence of investment.[3] They gave the example of Alutiiq in Alaska, whose community had begun a revitalization programme that might counter the trend their model predicted.[3]

Regional patterns and colonization

The 2013 Ethnologue analysis found sharply different outcomes in differe