Where and when did languages emerge? The answer

In Science, a new paper by Quentin D. Atkinson “Phonemic Diversity Supports aSerial Founder Effect Model of Language Expansion from Africa” is generating a lot of well-deserved interest (see here, here, or here for instance).

Abstract:

Human genetic and phenotypic diversity declines with distance from Africa, as predicted by a serial founder effect in which successive population bottlenecks during range expansion progressively reduce diversity, underpinning support for an African origin of modern humans. Recent work suggests that a similar founder effect may operate on human culture and language. Here I show that the number of phonemes used in a global sample of 504 languages is also clinal and fits a serial founder – effect model of expansion from an inferred origin in Africa. This result, which is not explained by more recent demographic history, local language diversity, or statistical non-independence within language families, points to parallel mechanisms shaping genetic and linguistic diversity and supports an African origin of modern human languages.

Comments Disabled

  • comment-avatar
    José-Luis Guijarro 18 April 2011 (18:37)

    Interesting paper, indeed. I was unable to read the original, but its abstract gives us an idea of its possible theoretical implications. That the phonological characteristics of human languages seem to have evolved from a first spoken language in Africa helps to strengthen the idea that this sort of tool was put together in only one place, and was subsequently exported to the rest of the human kin. So much for the multiple geographic origins of spoken languages! However, as I said in some other thread, if human spoken languages originated by symbiosis of two precedent faculties, the paper (or, rather, its abstract) does not really have anything to say about those two (precedent) faculties, namely, the communicative faculty (which had probably many things in common with that of other species’) and the language faculty (understanding language here as Sperber & Wilson do in their [i]Relevance[/i] book: as a tool for mentally formalising, storing, and manipulating representations). From reading the abstract it is uncertain that the authors wanted to explain the origin of that human tool in all of its complexity (the possible ways to organize its layers of structure, its embedding potentiality, the kind of link between some of its elements and conceptual representations, etc.) which, in my mind, are a lot more crucial for describing language than concentrating on the way we are apt to build up our linguistic sounds we use when talking. That is: (1) We are still at the hypothetical level when figuring out how human language originated in the first place, be it by evolution from primitive communicative acts, be it, as I strongly believe, by symbiotic union of our mental linguistic tool and our communicative faculty. (2) If the symbiotic account is favoured, it is interesting to have some evidence on how the first human being who got the idea of blending those two faculties (a [i]good trick[/i], in Dennett’s words, if there ever was one!) seems to have been a common ancestor in Africa –for the way he built up the phonological apparatus to behave linguistically has been shown to have evolved from there by the authors of the paper. To repeat yet again: so much for the multiple origin idea and the [i]exclusive[/i] cultural characteristics of human languages!

  • comment-avatar
    Fernando Carvalho 19 April 2011 (19:32)

    That’s my feeling, at least in principle. . Though I only had access to the paper’s abstract so far, it seems highly misleading to think of a (genetic) founder effect applying to populations of genes and to populations of linguistic entities (say, phonemes). . A genetic founder effect depends on the fact that given a polymorphic population, not all variants are present in any single individual, and if by chance (that is, in a manner independent of the genetic identity of individuals) some individuals split from the main population, a new group may be formed with changed relative frequencies for the variants (by chance, some variants may be not represented in the splitting population, or a single variant may be fixed, i.e., attaining 100% frequency). It is, in a sense, a population-level counterpart of statistical non-representative sampling (the “new” population taken as a sample of the earlier one). Now, with phonemes, as assumed in the best of our models, [i]all[/i] individuals in the population have all the variant “genes”. (reference here is to individual persons of course, members of a human population). I fail to see how it logically follows (as it does in the case of evolutionary biology for the definition of the founder effect) that a founder effect promoted by some population bottleneck, would give rise to such founding effects characterized by extant reduced diversity. . The crucial point here is that the relation between population bottlenecks and reduction of “phoneme diversity” seems to be assumed by Atkinson to be mediated by the results of the Hay and Bauer 2007 study pointing to a positive correlation between phoneme inventory size and population (number of speakers) size. As far as I can tell, as with any statistical induction over non-experimentally produced samples of research unities, the Hay and Bauer result may be simply a spurious correlation, and it stands, in may opinion, as a shaky foundation for such wide-ranging claims (e.g., that the “structure” of tranmission of both languages and genes would be similar; ‘structure’ here in the sense of Boyd & Richerson 2004). . As another main point, it seems to me that another reason for the likelihood that many of these statistical effects may be spurious or artifactual is that only a tiny fraction of what gets transmitted in language actually finds its way in the data. Morphological, syntactic and lots of other phonological and phonetic representations and processes all get transmitted in the course of language’s histories,not simple phoneme inventories. Indeed, as far as we know, phonemes may be quite epiphenomenal, and phonological features (sub-phonemic units) are really the level of analysis at which most of the action goes. Further, as linguistic theory shows us, the behavior of phonemes (or features) as “information bearers” in language structuring and its functioning is conditioned and co-dependent on several other aspects of the grammatical structure, including syllabic and prosodic organization, morphological structure and even the structuring of its allophonic range.Phonemes have been chosen, probably, because they are easily treatable mathematically, counting metrics and some amout of discrete and combinatorial math doing all the job. My suspicion concerning such works would be easened if more representative indexes of grammatical structuring were devised, and though it looks like a daunting task, linguists working on typology have tried hard to develop such encompassing indexes (cf. e.g. Haspelmath’s 2009 study on morphological agglutination). . I guess that’s it for now. I’ll need now to read the entire paper carefully to properly qualify my take on this work. References: Boyd & Richerson (2004) Not By Genes Alone. University of Chicago Press. Haspelmath, M. (2009) “An Empirical Test of the Agglutination Hypothesis” In: Scalise, Magni & Bisetto (eds.) Universals of Language Today. Dordrecht: Springer. Hay & Bauer (2007) “Phoneme Inventory Size and Population Size” Language 83 (2): 388-400.