Where was human language born?
Anthropologists and geneticists alike believe that modern humans originated in Africa. But what about human language? Did it originate in Africa too? Was it “born” only once? And can we be more precise about where in Africa might it have happened?
Quentin D. Atkinson from the University of Auckland in New Zealand thinks he has the answers to these questions. His research has just been published in an article in the journal Science and has been discussed in the New York Times by Nicholas Wade.
Atkinson, a biologist by training, has applied mathematical methods used in genetics to linguistic data from 504 languages around the world, and has found a simple but striking pattern: A language area uses fewer phonemes the farther that early humans had to travel from Africa to reach it.
Since the largest phoneme inventories are found in click-using languages of Africa (some of which have more than 100 phonemes), Atkinson places the origin of human language in southwestern Africa (see the lightest-shaded area in the picture below). Hawaiian, toward the far end of the human migration route out of Africa, has only 13. English has about 45 phonemes (depending on the dialect).
What do linguists think about this type of research? Martin Haspelmath, a linguist at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany said that “…this paper comes closest to convincing me that this type of research is possible”. Another linguist, Donald A. Ringe of the University of Pennsylvania, is inspired by Atkinson’s research too: “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”. But others remain less than convinced. For example, Brian D. Joseph, an expert in historical linguistics at Ohio State University explains: “We’re uneasy about mathematical modeling that we don’t understand juxtaposed to philological modeling that we do understand”. And it’s not just about Atkinson’s methods. His results don’t make sense.
Take, for instance, his idea that human language arose in the area of the largest phonemic inventories. It is true that some of the languages with the richest phoneme inventories are those that use click sounds. However, a quick examination of the WALS map reveals that languages with rich (consonant) phoneme inventories — marked by dark red dots — include not only African languages like !Xóõ and Ju|’hoan, but also non-click languages of the Caucasus (e.g., Lezgin and Kabardian), some Papuan languages and even — in direct contradiction to Atkinson — some languages in South America, the farthest end of the human migration route out of Africa (e.g., Jaqaru and Araona). Also, quite a few languages with very small (consonant) phoneme inventories are located in western Africa, as revealed by the dark blue dots on this map.
Furthermore, taking the presence of click sounds (and consequently larger phoneme inventory) as a sign of the language’s closeness to the birthplace of human language does not seem fair: it is well-known among linguists that click sounds are indigenous to Khoisan languages, while those among Bantu languages that have clicks as well must have borrowed them from their Khoisan neighbors. One piece of evidence to support the borrowing theory of click sounds in Bantu is that Bantu languages that do have click sounds at all employ only a small subset of possible clicks. In particular, only dental, alveolar and lateral clicks are found in Bantu languages, while Khoisan languages also feature bilabial and palatal clicks.
Another problem with Atkinson’s findings is that anthropologists and geneticists typically place the origins of modern humans in eastern Africa:
While it is not impossible that human language originated in southwestern Africa later than the first anatomically modern humans arose in eastern Africa, most geneticists and evolutionary theorists believe that anatomically modern humans with lightly built skeletons could not have existed without being behaviorally modern too, that is without having that secret weapon — language.
What Atkinson’s research seems to miss is that current geographical distribution of languages is not always indicative of their original distribution. For example, it is quite well-proven that Khoisan languages originated further north than their present-day area of distribution, which resulted from Bantu migrations. The expansion of the Bantus pretty much limited the distribution of the original click-languages, the Khoisan, to the Kalahari desert. Clearly, in order to reveal where the first human language was born, one needs to factor in the known migration patterns.
The history of the Khoisan people and their languages reveals yet another contradiction in Atkinson’s argument. He assumes that the number of phonemes in a language increases with the number of people who speak it. This may work for certain languages, regions and times, but it cannot be a general principle. If it were true, why would Khoisan languages, which constantly lose speakers throughout history, have some of the richest (consonant) phoneme inventories?
In addition to the methodological problems discussed above, Atkinson draws a conclusion that does not make sense to me. He claims that
“the number of phonemes used in a global sample of 504 languages […] fits a serial founder–effect model of expansion from an inferred origin in Africa. This result […] points to parallel mechanisms shaping genetic and linguistic diversity and supports an African origin of modern human languages.”
The decrease in genetic diversity as one moves further and further away from the birthplace of modern humans in Africa is “predicted by a serial founder effect in which successive population bottlenecks during range expansion progressively reduce diversity”. It is clear why this happens: each small group of founders splitting off and migrating away from the parent population carries with it only a subset of the genes from the parent population’s gene pool:
But why would it be that a founder group splitting off from the parent population would take with it only a subset of phonemes in the original language’s inventory? Take vowels, for example. If the original languages has a five-vowel system (such as in the picture below), is it necessary for a founder group to walk away with only three of those five vowels?
Moreover, we can predict from well-described phonological changes, such as the Great Vowel Shift in English, that if a founder group dared to walk away from the rest of the tribe with an unbalanced vowel system, for example, a three-vowel system of /e/, /o/ and /a/, it will quickly reintroduce the missing high vowels (/i/ and /u/) to maximally utilize the vowel space.
Overall, I feel a great deal of scepticism about Atkinson’s research and its findings. And it’s not just the application of mathematical methods, which I may or may not understand. It’s that application of mathematical methods may spit out results that don’t jibe with what we linguists know about human languages.