Humans May Have Spoken 75,000 Languages Just 2,000 Years Ago. Then Empires Rose

Today’s 7,600 languages might seem like a lot, but they're just a fraction that survived a vast, ancient extinction.

The ancient world may have been speaking in tens of thousands of languages. Most went completely silent.

A new study in Science reconstructs the rise and fall of language diversity over the past 12,000 years. Its models suggest that humanity reached a linguistic “golden age” between roughly 1000 B.C.E. and C.E. 1000, when perhaps 20,000 to 75,000 languages coexisted — several times the roughly 7,600 spoken and signed today.

Then diversity plunged abruptly. While it’s tempting to blame colonialism, this was not the prime factor. The researchers argue that the collapse began well before European colonialism, as expanding states and multinational empires spread their languages, institutions, populations and diseases across previously independent communities. Colonial powers did later accelerate that process but may not have started it.

Counting Languages That Left No Fossils

Writing appeared only about 6,000 years ago, and even afterward scribes recorded a tiny, politically skewed sample of human speech. To reach further back, the researchers used ethnographic information from 171 hunter-gatherer and fisher societies whose traditional ways of life had not been profoundly reshaped by farming populations.

Such present-day ethnolinguistic groups contained a median of about 800 people. Combining their size distribution with estimates that 4.4 million to 7 million humans lived 12,000 years ago, the team calculated that the early Holocene probably held about 4,500 to 6,200 languages, though the wider interval ran from 3,300 to 7,800.

That’s about as many languages as today for a global population about 2,000 times smaller.

The researchers then joined that starting point to the present using historical population estimates. They assumed that, over long periods, technologies, agriculture and political organization allowed ever more people to share one language, and generated thousands of possible trajectories.

The introduction of agriculture was perhaps the most disruptive episode in this regard. Earlier work, including a landmark 2003 review of farmers and language expansions, emphasized how food-producing societies displaced or absorbed hunter-gatherer populations. But farming also supported many more people and encouraged permanent, sometimes isolated settlements where speech could diverge. The new study considers that the net effect of farming societies on the development of new languages was positive.

×

Get smarter every day...

Stay ahead with ZME Science and subscribe.

Daily Newsletter
The science you need to know, every weekday.

Weekly Newsletter
A week in science, all in one place. Sends every Sunday.
No spam, ever. Unsubscribe anytime. Review our Privacy Policy.

Thank you! One more thing...

Please check your inbox and confirm your subscription.

Across most models, population growth initially outran linguistic consolidation. Language numbers rose for millennia and peaked around 2,000 years ago, often at roughly 10 times their early-Holocene level.

“You have this slowly brewing linguistic and cultural diversity,” Damián Blasi, the study’s lead author, told Scientific American. “At some point, a few of these [languages] explode, and they wipe out everything that’s around them.”

Then came the age of empires.

When Empires Become Language Machines

The model does not identify a single empire as the culprit. Instead, it places the turning point roughly 1,000 to 3,000 years ago, when large states and multinational empires became increasingly capable of moving soldiers, settlers, traders and administrators across vast territories. The Roman Empire is one familiar example, but the pattern was global.

Expanding powers, whether Rome around the Mediterranean, the Han dynasty across much of China, or later Arab-Islamic states across the Middle East and North Africa, carried dominant languages into newly conquered or absorbed communities. Latin, varieties of Chinese and Arabic spread through government, trade, religion and education, often displacing smaller local languages over generations.

Language loss did not necessarily require mass killing or a formal ban. Conquered or absorbed populations could survive while gradually abandoning their own speech. A state language might become necessary for social advancement, while children increasingly learned it instead of the language spoken by their grandparents. Over generations, a small language could disappear even as its former speakers became part of a larger population.

European colonial expansion after about 1500 CE intensified this process, sometimes through the deliberate suppression of Indigenous languages and religion. But the study argues that the decline had already been underway for many centuries. “It takes much less than a colonial powerhouse to destroy linguistic diversity,” Blasi told Scientific American.

“The languages we see today are the survivors of a massive and highly selective historical bottleneck,” added Russell Gray, a co-author, said in a Yale press release.

The ancient pattern also gives a deeper backdrop to the current crisis. A 2022 global analysis of 6,511 languages projected that, without intervention, at least 1,500 could cease to be spoken by 2100. It classified 37 percent as threatened or worse and found that the number of people who learned a language from birth was the strongest predictor of its survival. But infrastructure and education also mattered. Minority languages surrounded by denser road networks faced greater danger, probably because roads connect previously remote communities to cities, wage labor, centralized government and dominant languages used in commerce.

RelatedPosts

Talking in a foreign language makes our decision more about utility, less about emotion
We’re Starting to Sound Like ChatGPT — And We Don’t Even Realize It
When Did Humans First Speak? New Genetic Clues Point to 135,000 Years Ago
Researchers Claim They Have Cracked the Code of Teotihuacan’s Mysterious Writing System

A Provocative Model, but No Time Machine

The uncertainty in the presented figures is enormous. The paper’s central graph is a broad fan of possible histories rather than a clean curve, and its result depends on treating modern-day ethnolinguistic groups as rough proxies for ancient languages. It also assumes that the global ratio of languages to people declined over the long run.

Lyle Campbell, a University of Hawaii linguist who was not involved, told Scientific American that the team was “grappling with stuff in the remote past that is hard to get a solid purchase on.” The scenario, he said, is “highly speculative.”

There is also the slippery boundary between a language and a dialect. In a landscape of small neighboring communities, speech forms may have blended into chains of partial mutual understanding.

Claire Bowern, a Yale linguist and study co-author, called the estimates “a first hypothesis.” They remind us not only of languages that are dying now, but of how much diversity political expansion had already removed before modern documentation began.

“This finding is important for the many endangered languages of today because political currents don’t have to move in one direction. It’s not inevitable that endangered languages become extinct,” Bowern said.

In a FABBS profile, Blasi called linguistic diversity “a repository of human ingenuity: a record of the many different solutions that human societies have developed over myriads of human generations.”