Clever Study Uses Genetics Trick to Trace Language Back to Its Very Beginning, in Africa

By Valerie Ross | April 15, 2011 2:47 pm

Likely area of language origin, in white, based on:
A) phonemes found in individual languages and
B) phoneme diversity averaged across language families

What’s the News: Southern Africa may be the birthplace of human language, according a new study published yesterday in Science. The study further suggests that language may have arisen only once, with one ancestral language giving rise to all modern tongues, an idea linguists have long debated. This finding parallels the human migrations out of Africa supported by genetic and fossil evidence.

How the Heck:

  • The study’s author, evolutionary psychologist Quentin Atkinson of the University of Auckland in New Zealand, looked at 504 modern languages from around the world.
  • He then tallied the phonemes—the distinct sounds of consonants, vowels and tones—that make up each language. Languages vary widely in how many phonemes they have: Some of the Khoisan languages in Africa (widely known for their click sounds)  have more than a hundred phonemes, while languages spoken in many Pacific islands have far fewer, such as Hawaiian’s 13. (English is somewhere in the middle, with about 45.)
  • To build a model for the origins of language, Atkinson borrowed an idea from population genetics: the founder effect, which says that when a small group branches off from a population, it loses genetic diversity.
  • Performing mathematical analysis to bring together the number of phonemes a language had and its location on the globe, Atkinson found a linguistic founder effect: The farther from Africa, the fewer phonemes a language had—the less diverse its sounds were. This distance from Africa explained 30% of the variation in number of phonemes a language had, and still explained nearly 20% of the variation when modern population size was taken into account (since smaller populations are also linked to a smaller number of phonemes).

What’s the Context:

  • Language is believed to be at least 50,000 years old, and could be 100,000 years old, predating the human migrations out of Africa that began about 50,000 years ago. The development of language in Africa, Atkinson told the Wall Street Journal, “was the catalyst that spurred the human expansion that we all are a product of.”
  • Using the traditional methods of tracking how words evolve, linguists have only been able to trace particular languages back less than 10,000 years, since words change so quickly (and there are no written records going back further). By taking a different approach, Atkinson may have gleaned information about language much further back in time.

Not So Fast:

  • While Atkinson’s idea stems from population genetics, the mechanisms of change in the two fields are very different: A population’s genetic diversity changes over generations, while phonemes can change far more rapidly. The founder effect accounts for a different amount of the variation between groups, as well, explaining up to 85% of the genetic diversity in populations but only about 20% of the phoneme diversity in languages.
  • Many linguists aren’t yet sure if they should believe what they hear. As University of Pennsylvania linguist Donald A. Ringe told the New York Times, “It’s too early to tell if Atkinson’s idea is correct, but if so, it’s one of the most interesting articles in historical linguistics that I’ve seen in a decade.”

Reference: Quentin D. Atkinson. “Phonemic Diversity Supports a Serial Founder Effect Model of Language Expansion from Africa,” Science, April 15, 2011. DOI:10.1126/science.1199295

Image: “Phonemic Diversity Supports a Serial Founder Effect Model of Language Expansion from Africa”

  • Ryan

    How is “language” defined, Valerie? I’d think “language” would be much older than 100,000 years if our ancestors were making tools millions of years ago. I guess it all depends on how you’re defining “language”.

    I’m not sure which publication I read this in, but some monkeys apparently have different calls for different types of threats. The different calls mobilize members of the group to different actions. For instance, the call for a predatory bird causes the monkeys to crouch in the brush while a call for a land-stalking predator causes the monkeys to recede into the trees. Would this not be a rudimentary form of language?

  • lovecraft

    Ryan talks the truth :)

  • John Lerch

    Indeed surely at some point the number of phonemes was zero. So surely the correct model would need to have some middle number of phonemes from which some became more varied and some less. I would guess the geometric mean would be about correct. So the geometric mean of 100 and of 13 os about 35. So it looks like English is the progenitor language. :-). Seriously. I would think it harder to become more varied than less varied. So I’m not sure how one would find such an appropriate mean. But 70 sounds about right.

  • matt

    As I understand it, most modern linguists require the following attributes in order to consider something as a language:

    – Semanticity

  • BOb

    I think there’s a very strong likelihood that language is far older than 100,000 years. There is evidence of linguistic adaptations as far back as Homo Heidelbergensis…. there’s no way that their attenuated hearing was there simply to hear grunts. Similarly, the Neanderthals’ possession of a modern hyoid but particularly FOXP2 – a gene heavily pressured to be conservative in our ape cousins – is highly unlikely to be a coincidence.

  • Jeannette Jaquish

    From the article:
    The farther from Africa, the fewer phonemes a language had—the less diverse its sounds were.

    Is this true? African languages have more phonemes – more sounds – than more modern languages farther from Africa? Wouldn’t primitive man hundreds of thousands of years ago have a smaller vocabulary and therefore fewer sounds?

    The facts of this article are hard for me to believe but I guess they wouldn’t say it if it wasn’t true.

    Or was earlier language more symbolic with a different sound for each meaning, while later languages worked with combinations of (fewer) sounds to make a word-meaning.

  • Lila Sovietskaya

    The study made has great validity. Because humanity descended form a very small group of humans, it is almost certain that they had a language that diversified in the Babel of languages that we have now. The old story of Babel, shows that some people long ago, intuitively understood that all languages were once one language

    Define ‘Language’ Are whale songs a language? They communicate something. Are beeves movements a language? They communicate something. One funny thing in many science fiction movies, the aliens speak English. The truth is that languages can have sound, or movements, or radio signals, or lights. Sounds emitted by non-humans would use their sound emitting organs in an atmosphere with different composition and density. It is very unlikely that they will sound anything like human languages and probably we cannot reproduce their sounds or they ours. Inter-species communication will likely take place with the help of AI (Artificial Intelligence) translators. It would be great to meet aliens who can make human sounds and have a language that we could learn to use, like the Klingon language. Alien grammar, syntax, manner of speech and cultural or local references to flora and fauna, will make the job very hard.

  • Matt B.

    I don’t buy this at all. The number of phonemes in a language changes far too rapidly to have any relation to prehistoric migration. And it goes in both directions–Hebrew has lost and Latin has gained in the last 2000 years. I expect the exact results would differ using data from a different era. The map seems to imply that Russian has fewer phonemes than Spanish despite the phonemicness of palatalization in Russian. And the borders of the shades of red don’t correspond to the ranges of language families, such as Slavonic, Germanic and Romance, or even Indo-European.

  • Daniel

    Intriguing, but this sentence gave me pause: “This distance from Africa explained 30% of the variation in number of phonemes a language had”

    Doesn’t this suggest the founder effect failed to explain the majority of the variation we observe?

    Perhaps I’m missing something, but in my book, if a theory explains a third of what I observe, I’d reject that theory.

    Plus, I find it highly, highly unlikely that language evolved only once. But I’m no expert, that’s just my armchair science intuition talking.

  • Francesco Andreoli

    First of all, I apologize for my bad English, which is not my mother language, but I hope I can still make myself clear.
    This theory does not hold water for several reasons: as already pointed out by someone, the number of exceptions is so big, e.g. the paucity of phonemes of Spanish compared to Portuguese, although there are at the same distance from Africa, or the great number of phonemes in some North American native languages, to seriously undermine the theory. Indeed, by studying history, we realize that Spanish has lost phonemes because of a sort of law of the lowest common denominator between variants of the same language. That could be the case for many “phoneme-poor” modern languages.
    There are a number of mechanisms by which the number of phonemes increases. As an example, I have read that in some Indian (in the Indian subcontinent) villages, there are two different ethnic groups who speak completely different languages. The inhabitants are fully bilingual when dealing with people speaking the different language. In some villages, each language has borrowed heavily for the other. Suppose, for example, that there were originally 50 phonemes in each language, with only 20 in common. Suppose that after a few hundreds of years the original languages were lost. Therefore we would have a community who speak a single language, with 20 +30 +30 = 80 phonemes. Suppose something like this has happened in the lost past, 20 or 50 thousands years ago, in several different places in the world, we would only see the resulting “phoneme-rich” languages scattered here and there, which is actually the case.
    Furthermore, it has been found by other studies that the number of phonemes is proportional to the number of speakers of a language. Big populations have a much greater variety of phonemes than small tribes. As the human population has been quite small until the onset of agriculture, we can be sure that the number of phonemes spoken, say 100 thousand years ago, was quite small, and the increase is a rather recent phenomenon.
    But the theory it is a blatant misunderstanding of reality for a very basic reason. Europeans, Asians etc. do not descend from “Africans”, whatever the number of phonemes “Africans” could speak.
    They are descended from a small group of perhaps a hundred individuals who crossed the Red Sea and spread to Asia and Europe etc. from a well-defined East African region, probably in the area of Ethiopia.
    Therefore, what matters is only the language they spoke there some 90.000 years ago, not the ones they would speak (then or now) for example in Mali or Uganda or in the Kalahari.

  • Charlie Craig

    I’ve always thought you could trace migrations of people and genes by following the origins of language. Where ever you find a language in a new land you also find genetic replication as well. Your not going to have a group of people just start speaking another groups language without having any crossing of genes between the groups it never happens like that and it never will.

  • Charlie Craig

    theres another factor, Neanderthals and Denisovans. We know now as a matter of fact that when homo sapiens migrated out of africa they mated and integrated with other humanoid cultures such as the Neanderthals in the north and Denisovans in the east. We know that Neanderthals had language, religion, music(including the first flutes), art, spears and other advanced weaponry, Their brains were larger than ours were on average and the fossil record shows this, Their bodies were also larger and they were adapted to the cold climates of europe and russia/siberia and some of those adaptations included the control of fire which was required for survival, lighter skin and hair to blend in with the environment which was a result of generations of low exposure to UV light and a reduced need to produce melanin in the skin and hair. This explains why all-non africans have lighter skin and why all non-africans have around 6% Neanderthal DNA. The Denisovans were just discovered so no genome project has been started on their species yet but i’d be willing to bet that people from east asian have a some Denisovan DNA. Theres also a good chance that homo floreiensis(the hobbit) has left some of its genes in the asian population which would explain they’re smaller stature, the hobbit people of indonesia were quite intelligent as well and in light of a recent archaeological find we know they were making massive monuments in indonesia for thousands of years(20,000 to be exact) the megalithic site called Gunung Padang in indonesia was just tested last year after being ignored for over one hundred years by the scientific community, carbondating proves the inner chambers at Gungung Padang are 20,000 years old and built using obsidian tools. Some of the tools of the artifacts of the hobbit people date back to 94,000 years ago. Indonesia underwent a cataclysm around 12,500 years ago when the ice age ended and sea levels went up by several meters which submerged most of the continent that was indonesia and which extended from korea all the way down to Australia and the also former continent of zealand (the highlands of which survive today as New Zealand)

    • Charlie Craig

      It’s also important to note that the Hobbit people of indonesia died out around 12,000 years ago which is great evidence that the rising sea levels destroyed the coastal regions (and created the chain of islands that make up indo,micro and melanesia, and the Philippines) where they were most likely to live as all humanoids do to this day. It’s also important to note that in almost every tribal religious tradition, higher ground was sacred and monuments were often constructed on top of hills and mountains. which would explain how an entire population could be wiped out and have some of their structures remain. Some sacred ground was forbidden such as the island of delos in greece. another interesting note is that in Plato’s account of Atlantis which he learned from Egyptian priests tells of a great civilization that was swallowed by the sea 9,500 years before(plato wrote this around 2,500 years ago) which relates to the exact time of the extinction of the hobbit and the and of the ice age which changed all the coastal regions in the world and submerged several land masses such as the land bridges between alaska and siberia, Canada and greenland greenland and iceland iceland and ireland ireland and england england and mainland europe, india and sri-lanka(but not completely there is still a bridge there that almost looks artificial and in hindu mythology it was constructed by a monkey army under the control of Vishnu who’s lover was taken to sri-lanka against her will and imprisoned on the island) There is evidence of civilization in india dating back 7,000 years and prototypes of hindu gods are found in Mehrgarh. Just recently in Turkey the Gobbekli tepe monument was confirmed to be 11,500 years old and it was part of a large settlement called nevali cori(the settlement in the area never really ended so it’s almost safe to say that the city of Urfa is the oldest surviving city in all of human history.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!


80beats is DISCOVER's news aggregator, weaving together the choicest tidbits from the best articles covering the day's most compelling topics.

See More

Collapse bottom bar