Language Survival

Use, Transmission, and Network Dominance

How a Language Survives

Language is more than a vocabulary. It is a shared social system through which people communicate, teach, remember, joke, argue, worship, work, and pass knowledge from one generation to the next. A language survives when enough people continue to use it in ordinary life and, above all, when children acquire it as a living means of communication. If that transmission stops, the language may remain in books, recordings, inscriptions, or ritual use, but it no longer functions as an ordinary inherited language.

Languages change, split, merge, borrow, spread, contract, and sometimes disappear. Their fate is determined by the behavior, opportunities, pressures, institutions, and choices of speakers—not by any drive internal to the language.

Language does not create physical reality, but it can preserve a community’s categories, stories, technical knowledge, humor, memory, and habits of expression. When a language falls silent, reality remains. What may disappear is a distinctive historical record of how one community organized and transmitted its experience.

Language Change Is Cultural, Not Biological

It is tempting to describe expanding and declining languages in Darwinian terms. Languages do change, branch, compete for speakers, and sometimes disappear, but the process is cultural rather than biological. A language has no genes, no reproductive strategy, and no intrinsic drive to survive.

August Schleicher explicitly applied Darwinian ideas to language in 1863, and Darwin later compared changes in words and linguistic forms with natural selection. The comparison remains useful only if its limit is clear: the decisive environment for a language is social. Families, schools, governments, markets, migration, prestige, media, and technology shape whether people continue to use it.

A language with fewer speakers is not inherently inferior, and a large language is not inherently better. A small language may remain vigorous when children acquire it and use it in ordinary life. A much larger language can nevertheless become endangered if its speakers stop transmitting it to the next generation.

How Many Languages Are at Risk?

Current estimates differ because the boundary between a language and a dialect is sometimes uncertain and because vitality can be measured in several ways. UNESCO’s current multilingualism data report about 8,324 languages documented and approximately 7,000 still in use, with around 1,500 at risk of losing their users in the near future and only about 1,000 represented online. The older UNESCO Atlas figure that roughly 40 percent of languages are endangered, and Ethnologue’s similar EGIDS-based share of about 44 percent, remain the conventional range in public discussion. Those percentages depend on how “endangered” is defined.

The exact share matters less than the shape of the distribution. A relatively small number of languages account for most of the world’s population, while thousands of languages are spoken by much smaller communities. Small population alone does not make extinction inevitable, but small communities are more vulnerable when transmission weakens or social domains shift to a dominant language.

The Most Important Question: Are Children Learning It?

UNESCO identifies intergenerational transmission as the most commonly used indicator of language vitality. A language can have thousands, or even hundreds of thousands, of adult speakers and still be endangered if children are no longer acquiring it naturally at home and using it with one another.

This is why raw speaker counts can be misleading. A language with 50,000 speakers across all age groups may be healthier than a language with 500,000 speakers concentrated among older adults. What matters is not only how many people can speak the language today, but whether the language is reproducing itself socially from one generation to the next.

Multilingualism by itself is not a threat. Communities can successfully maintain two or more languages. The danger appears when one language increasingly replaces another in the home, school, workplace, media, religion, government, or digital life until the smaller language is confined to fewer and fewer situations.

A Practical Language Vitality Framework

Language vitality is best understood through several interacting questions. The following framework tracks factors used by UNESCO’s 2003 vitality paper and by later work in the Fishman–EGIDS tradition, while keeping the central issue in view: is the language still being transmitted and used? Joshua Fishman’s reversing-language-shift model adds a further point: restoring home and neighborhood use is harder than adding a school subject, a broadcast service, or an app.

No single factor in the table can carry a language by itself. Official recognition fails if children no longer hear it at home. Raw speaker numbers fail if use shrinks to ceremonies and elders. Apps and keyboards fail if nobody needs the language for ordinary life. A dominant network language such as English can take over school, work, media, and software without the smaller language being a weaker system of thought. That is why headcount misleads. The question is whether people still have places in which they need, and choose, to speak it to one another and to their children.

Why Languages Decline

Language shift usually occurs gradually. A community becomes bilingual, one language becomes more useful or prestigious in important domains, and parents begin favoring that language for their children. The older language may remain emotionally important long after ordinary daily use begins to weaken.

UNESCO identifies both external and internal pressures. External pressures can include conquest, displacement, discriminatory policy, schooling in another language, economic marginalization, migration, or mass media dominated by a larger language. Internal pressures often follow: speakers may come to believe that their own language is an obstacle to mobility or that children will be better served by adopting the dominant language.

Urbanization and migration can accelerate this process because families are brought into larger and more linguistically mixed networks. But the outcome is not automatic. Bilingual education, strong community institutions, family transmission, and media can allow a minority language to coexist with a dominant one for generations.

Prestige, Opportunity, Power, and Network Dominance

Languages also carry social signals. Accent, dialect, vocabulary, and register can influence how speakers are perceived in schools, employment, politics, and other institutions. This does not make one dialect linguistically superior to another. It means that societies attach prestige to some varieties and stigma to others, and that prestige can be magnified when a language is tied to powerful economic, political, educational, technological, and cultural networks.

The distinction matters. Social mobility depends on education, networks, discrimination, economic conditions, and the ability to shift registers when circumstances require it. A person can maintain a local or family variety while also acquiring the standard variety used in formal institutions. Prestige attached to a dialect is a social fact, not evidence that one variety is structurally superior to another.

Culture shapes language, while language can reinforce social boundaries created by culture. Prestige, stigma, education, opportunity, and network reach therefore influence language survival even when the languages themselves are structurally capable of expressing the same range of human thought.

American English is the clearest modern example of how political and economic power can become linguistic power. English first acquired worldwide reach through British imperial expansion, trade, and industrialization. During the twentieth century, the United States added economic scale, military and diplomatic reach, universities and scientific publishing, mass entertainment, aviation, computing, and later the Internet. The result was not that English proved linguistically superior. English became attached to institutions and networks that people around the world wanted, or needed, to enter. David Crystal argues that military power can help establish the international position of a language, while economic power is crucial to maintaining and expanding it. Eric Anchimbe similarly emphasizes the postwar influence of American political, commercial, technological, and popular culture on international English.

Once a language reaches this scale, network dominance begins to reinforce itself. A person in Brazil may learn English to communicate with a person in Japan; neither needs to be British or American. Scientists publish in English because other scientists read English. Programmers use English because documentation, software communities, and technical terminology already use it. International businesses adopt it because employees in many countries already know it. Every additional user makes the network more useful to future users.

This creates a powerful feedback loop: power creates reach; reach creates usefulness; usefulness attracts more speakers; more speakers increase the value of the network. At that point a language can retain global importance even as the distribution of geopolitical power changes. Network dominance is therefore different from simple political dominance. It becomes partly self-sustaining.

The same network also carries ideas. An expression coined in American English can move quickly through universities, entertainment, journalism, corporations, technology platforms, and social media, then pass into other languages through translation or borrowing. The language does not determine which ideas are true. It gives ideas access to an unusually large transmission network.

Three advantages are easy to blend and should be kept distinct. The first is historical political-economic power: Britain, then the United States, attached English to trade, science, entertainment, and administration. The second is network externalities: the language is useful because other people already use it. The third is institutional lock-in: aviation, computing, scientific publishing, and now large-scale AI systems were built around English-language corpora, interfaces, and documentation. A language can keep the second and third advantages after the originating power declines.

Network dominance does not prove that minority-language maintenance is futile. Hebrew and Welsh show that a smaller language can hold high-value local domains inside a larger network. The threat is domain capture—the drift of home, school, work, and digital life into the network language—not the mere existence of a world language.

Language Survival in the Digital World

Early computing created real disadvantages for languages whose writing systems were poorly supported by dominant technical standards. ASCII, for example, was designed around a small set of Latin characters and could not represent most of the world’s scripts.

Today the more relevant issue is not whether a language fits ASCII. Unicode was created specifically to encode writing systems far beyond basic Latin, and modern operating systems can support a very large range of scripts. The remaining challenge is practical digital infrastructure: reliable fonts, keyboards, spelling tools, search, speech recognition, machine translation, educational content, and enough online material for speakers to use the language naturally.

UNESCO’s current multilingualism page reports approximately 7,000 languages still in use but only around 1,000 represented online. The digital divide is therefore not merely about alphabet complexity. It is about whether a language has the resources and user base necessary to participate in modern communication. Because digital platforms reward large existing networks, online visibility can further strengthen already dominant languages while making smaller languages harder to discover and use.

Artificial intelligence can deepen that advantage. In the WildChat corpus of public ChatGPT interactions, English accounted for about 53 percent of turns. At the same time, OpenAI reported in 2026 that users who predominantly use a language other than English now represent more than half of active consumer users. Those findings are not contradictory: adoption is rapidly becoming multilingual, while the supply of high-quality training data and technical resources remains uneven across languages. Recent NLP research continues to describe English as exceptionally data-rich and many other languages as data-constrained. A language can have full Unicode support and still be harder to search, dictate, translate, or support with AI systems at comparable quality. That gap can add to the practical pull of English in school and work.

Three Paths: Revival, Maintenance, and Revitalization

Hebrew, Welsh, and Cherokee illustrate three very different paths by which a language can persist or return to wider use. Their histories show that survival cannot be reduced to population size. The useful comparison is the bottleneck each case still faces.

Hebrew is especially important because it demonstrates that “dead” and “living” are not always absolute categories. Hebrew continued for centuries in religious, literary, and scholarly functions before becoming the everyday spoken language of a new native-speaking population. Scholars therefore describe its modern history as a revival or emergence into ordinary speech rather than the creation of a language from nothing. Revival did not mean invention from zero, and it did not mean dictionaries alone. It meant new native-speaking children plus institutions that made Hebrew the default public language.

Welsh illustrates a different pattern: continuity rather than resurrection. The Welsh Government’s Annual Population Survey estimated about 843,700 people aged three or older could speak Welsh in the year ending March 2026, or 27.3 percent of that population, while warning that survey uncertainty has increased. That APS figure should not be confused with the 2021 Census count of 538,300 speakers, or 17.8 percent; the Welsh Government treats the Census as the stock measure and cautions against using the APS to track the million-speaker target. The important lesson is that even a language with extensive policy support must still be measured through actual use and transmission, not through official recognition or school enrollment alone.

Cherokee illustrates urgent revitalization. Recent estimates vary. Cherokee Nation’s public Language Department page gives a rounded estimate of about 2,000 first-language speakers, while Cherokee Nation reporting in late 2025 and Apple reporting in May 2026 put the number of fluent speakers below 1,500, most of them older adults. The difference itself illustrates how quickly endangered-language counts can change and how definitions vary. The Nation maintains immersion schools, master-apprentice programs, language technology, fonts, keyboards, and online resources. The objective is not merely to archive the language, but to create new proficient speakers who will use it with children.

Can a Language Come Back?

Yes, but “revival” can mean several different things. At one end, documentation preserves recordings, grammar, vocabulary, and texts after ordinary transmission has largely ceased. At another, adults learn the language as a second language. Stronger revitalization creates fluent young speakers and restores use in homes, schools, media, ceremonies, workplaces, and public institutions.

The hardest step is usually not producing a dictionary or an app. It is rebuilding a social environment in which people regularly need and want to use the language with one another. Joshua Fishman emphasized in Reversing Language Shift and later essays that durable maintenance ultimately depends on restoring intergenerational transmission and the social institutions that support it.

This is why technological preservation and social revival should not be confused. A perfectly documented language can still be socially extinct. Conversely, even modest digital tools can be valuable when they strengthen an active community of speakers.

What Is Lost When a Language Disappears?

When the last fluent speakers of a language die without transmitting it, the loss is not simply a reduction in the number of labels available for the same world. Languages preserve oral histories, ecological knowledge, kinship systems, humor, metaphor, songs, ritual formulas, and distinctive patterns of expression. Some of that knowledge can be translated. Some cannot be translated without substantial loss of context.

Reality is not created by vocabulary. Languages provide different historical archives of how human communities have classified and communicated their experience of that reality.

Language preservation therefore has an option value. Documentation allows future generations to recover knowledge that would otherwise become inaccessible. Revitalization goes further by preserving not only information about a language but the community practice of speaking it.

Does Every Language Have to Survive?

Languages have always changed and disappeared. Not every case of language shift is the result of coercion, and preserving linguistic diversity does not require freezing communities in the past. Speakers may voluntarily adopt another language because it expands education, trade, mobility, or participation in a wider society.

The central question is agency. Language shift looks very different when a community freely becomes bilingual or adopts another language than when children are punished for speaking their own language, schooling excludes it, or speakers are pressured to abandon it because of stigma or discrimination.

A reasonable objective is therefore not to command that every language must remain unchanged forever. It is to preserve the possibility of transmission wherever communities value it, remove unnecessary barriers to its use, document languages at risk, and allow speakers themselves to decide what role their language should play in their future.

References and Further Reading

1. UNESCO Ad Hoc Expert Group on Endangered Languages, Language Vitality and Endangerment (2003). The classic nine-factor framework, emphasizing intergenerational transmission and domains of use. https://ich.unesco.org/doc/src/00120-EN.pdf
2. UNESCO, Multilingualism and Linguistic Diversity. Current figures on languages documented and in use, legal recognition, digital representation, and near-term risk. https://www.unesco.org/en/multilingualism-linguistic-diversity
3. UNESCO, International Decade of Indigenous Languages 2022-2032. https://www.unesco.org/en/decades/indigenous-languages/about/idil2022-2032
4. Charles Darwin, The Descent of Man (1871), discussion of language change and the extinction of languages. https://darwin-online.org.uk/content/contentblock?itemID=F937.1&basepage=1&viewtype=side
5. August Schleicher, Die Darwinsche Theorie und die Sprachwissenschaft (1863). https://darwin-online.org.uk/converted/pdf/1863_Darwinsche_Schleicher_A1250.pdf
6. Yael Reshef, “The Role of Children in the Revival of Hebrew,” in No Small Matter (Oxford University Press, 2021). https://academic.oup.com/book/39421/chapter-abstract/339139871
7. Welsh Government, Welsh language data from the Annual Population Survey: April 2025 to March 2026. https://www.gov.wales/welsh-language-data-annual-population-survey-april-2025-march-2026-html
8. Cherokee Nation Language Department, current language programs and speaker estimates; see also recent Nation and local reporting on first-language speaker counts. https://language.cherokee.gov/
9. Joshua A. Fishman, Reversing Language Shift (Multilingual Matters, 1991), and “If Threatened Languages Can Be Saved, Then Can Dead Languages Be Revived?” Current Issues in Language Planning 2 (2001): 222–230. https://www.tandfonline.com/doi/abs/10.1080/14664200108668024
10. David Crystal, English as a Global Language, 2nd ed. (Cambridge University Press, 2003). On British expansion, American power, and the rise of global English. https://doi.org/10.1017/CBO9780511486999
11. Eric A. Anchimbe, “World Englishes and the American Tongue,” English Today 22, no. 4 (2006): 3-9. On American power, culture, and the international spread of American English. https://doi.org/10.1017/S0266078406004020
12. Ethnologue, language-status and EGIDS summaries. Useful for the conventional finding that roughly two in five living languages are endangered under a graded transmission scale. https://www.ethnologue.com/
13. OpenAI, How ChatGPT adoption has expanded / Signals individual data (2026). Consumer-user language mix; non-English-primary users now a majority of active consumer users. https://openai.com/index/how-chatgpt-adoption-has-expanded/
14. Wenting Zhao et al., “WildChat: 1M ChatGPT Interaction Logs in the Wild” (2024). Public multilingual prompt logs in which English is about 53 percent of turns. https://arxiv.org/abs/2405.01470
15. Skyler Seto et al., “Training Bilingual LMs with Data Constraints in the Targeted Language,” Findings of ACL 2025. On English as a data-rich language and constraints facing lower-resource languages. https://aclanthology.org/2025.findings-acl.977/
16. Apple, “Cherokee language learners bridge generations with iPad and Mac” (May 28, 2026). Reports fewer than 1,500 fluent Cherokee speakers. https://www.apple.com/newsroom/2026/05/cherokee-language-learners-bridge-generations-with-ipad-and-mac/

Editorial note: language-vitality statistics are estimates. Definitions of “speaker,” “native speaker,” “endangered,” and even “language” vary among sources, so exact totals should be treated cautiously. Digital and AI usage figures likewise depend on whether one counts users, messages, or tokens, and on whether the sample is consumer chat, enterprise use, or developer traffic.