Did Human Language Begin in Southern Africa? A Contested 2011 Study

In April 2011, University of Auckland psychologist Quentin Atkinson published a study in Science arguing that human language, like the human species itself, traces back to a single point of origin in sub-Saharan Africa. His method drew on phonemes, the distinct units of sound, vowels, consonants, and tones, that distinguish words from one another. Analyzing phoneme inventories across 504 languages worldwide using the World Atlas of Language Structures, Atkinson found that African languages, particularly those in the southern part of the continent, tend to carry the largest phoneme counts, with diversity declining steadily the farther a language’s location sits from Africa, bottoming out in recently settled regions like South America and the Pacific islands.

The pattern, Atkinson argued, mirrors a well-established finding in human population genetics: genetic diversity also declines with distance from Africa, a signature of what geneticists call a serial founder effect, in which each wave of migrants carries only a subset of the previous population’s variation, progressively narrowing diversity the farther a lineage travels from its origin. If language expansion followed the same repeated bottlenecking pattern as human migration itself, phoneme diversity should trace the same geographic decline, and in his dataset, it did. The finding made international headlines as the first quantitative evidence for a single African origin of spoken language, a question linguists had previously considered largely unanswerable given how quickly languages themselves change and how little written evidence exists from tens of thousands of years ago.

The claim did not go unchallenged. Subsequent published critiques argue that Atkinson’s result depends heavily on the specific dataset and statistical choices he used rather than reflecting a robust underlying signal: researchers who reran comparable analyses using a different phoneme database found not one African origin point but two separate ones, one in eastern Africa and another entirely outside Africa, in the Caucasus, and found that adding additional geographic factors to the model shifted the best-fit origin to New Guinea instead. Other linguists have gone further, arguing the entire premise, that phoneme counts behave like neutral genetic markers passed down with predictable loss over migration, rests on an unproven analogy between how genes and how spoken sounds actually change over generations, systems governed by very different processes.

What survives that debate is not a debunked idea so much as a genuinely open one. Southern Africa’s Khoisan languages, spoken by peoples including the San, do carry unusually large sound inventories, including the region’s distinctive click consonants, a real and striking linguistic fact regardless of how the origin debate resolves. But the specific claim that this diversity constitutes proof of where human language itself began remains, more than a decade after Atkinson’s paper, a contested hypothesis actively argued over in the linguistics literature rather than an established scientific fact, and popular retellings that state it as settled are getting ahead of where the field actually stands.