The African ur-language

By Razib Khan | April 14, 2011 3:49 pm

Several people have emailed/tweeted at me about the new paper in Science, Phonemic Diversity Supports a Serial Founder Effect Model of Language Expansion from Africa:

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.

Though there are major differences between biological evolution, constrained by relatively regular forms of inheritance, and cultural evolution, which is much more potentially protean, I think that there is great potential for unity of model and process. That is why I read A Replicated Typo (and presumably why several of the contributors to that weblog read the content here). But I generally have zero ability to evaluate the linguistic plausibility of these sorts of hypotheses about the origin and development of languages.

Generally attempts to translate biological models into linguistics seem to be met with skepticism, but Nick Wade in The New York Times has some quotes from linguists who do not seem overly hostile toward the new model. This in particular was kind of funny in my opinion:

“We’re uneasy about mathematical modeling that we don’t understand juxtaposed to philological modeling that we do understand,” Brian D. Joseph, a linguist at Ohio State University, said about the Indo-European tree. But he thinks that linguists may be more willing to accept Dr. Atkinson’s new article because it does not conflict with any established area of linguistic scholarship.“I think we ought to take this seriously, although there are some who will dismiss it out of hand,” Dr. Joseph said.

Sociology of science in action! In any case, I’m waiting to see if anyone at A Replicated Typo will humor me and perhaps touch upon the plausibility of this model. It isn’t as if everything published in Science is really quite as firm as outsiders might assume. This is a huge finding if valid. But extraordinary claims need to be met with caution.

CATEGORIZED UNDER: Human Evolution
  • Charles Nydorf

    The relatively greater phonetic diversity of African languages has long been recognized in by linguists. What is new in the article is a the extrapolation of this pattern into a global cline.

  • http://www.replicatedtypo.com James Winters

    Keep your eyes peeled over the next few days… I actually wrote about a similar topic on my website, performing a statistical analysis on the phoneme inventory size and demographic variables (which you linked to a while ago): here and here.

    I’ve only just skimmed the paper and it looks pretty good… But I don’t think he’s completely ruled out competing hypotheses, namely: that learnability pressures of a speaker population, especially one with lots of adult learners, might select against a large range of phonemic diversity. I’m not saying that founder effects weren’t involved… It’s just there’s also the possibility that selection against diversity (in order to increase learnability for adults i.e. during trade etc and the subsequent pressures on a lingua franca) might inflate the stats somewhat. Anyway, this is just a first stab at some thoughts hovering about in my head… I’ll give a fuller discussion over the next few days (prob cross post over at GNXP classic too).

  • Charles Nydorf

    Uriel Weinreich (1926-1967) wrote explicitly about founder effects in language in a 1964, article ‘Western Traits in Transcarpathian Yiddish” which appeared in the anthology “For Max Weinreich.” In the same year his student, Marvin Herzog, used the concept in his dissertation, “The Yiddish Language in Northern Poland.”

  • Sean Rutledge

    The timing – 70,000 years ago, corresponds to the Toba supervolcano, thought to be a human bottleneck with possibly 2000-15000 world survivors, and prior to the Out of Africa migration. It would not be unusual for them to all speak the same language.

  • John Emerson

    One of the problems: simplification (pidginization and creolization of expanding languages) and complexification (of isolated languages spoken by tight groups) are both historical processes in languages. I checked with a linguist friend and phonemic complexification (gaining new distinctions and increasing the phoneme count) does occur. (I had only known about syntactical and morphemic complexification.)

    Before this piece came out I had come to suspect that language reconstruction further back than a few thousand years may be impossible, because alternating periods of simplification and complexification (plus areal borrowing) would have the effect of churning or stirring, by erasing original features and producing new features not in the original language.

    My hypothesis has not been proven or even tested, AFAIK, though I’m not in touch. (One linguist I talked to, working on the very isolated language of Nivkh / Gilyak, which probably hasn’t been spoken by anyone but native speakers for centuries and is becoming more complex rather than less, thought my idea was interesting).

  • Linda Seebach

    Mark Liberman at LanguageLog (blog for and mostly by linguists) has a post
    http://languagelog.ldc.upenn.edu/nll/?p=3088#more-3088 (on another topic, mainly) that mentions this paper and says he will consider it further, so that’s a blog to watch if the topic interests you.

  • http://www.kinshipstudies.org German Dziebel

    It’s good to see another confirmation that linguistic structures are responsive to population size constraints. This new phonetic paper should be read in conjunction with another one: (http://www.plosone.org/article/info:doi/10.1371/journal.pone.0008559), which found that small population size correlates with complex morphology, while large population size with simplified morphology and a shift to lexical coding of syntactic relations. The authors advanced a hypothesis that in small population complex morphology helps language acquisition by children, while simplified linguistic morphology of large populations stems from the fact that large populations have a large number of adults who find this complexity overbearing. Evolution simplifies some structures (often all the way to making them extinct) and elaborates on others. Complexity and simplicity always go hand in hand. This is the case with genetics where intragroup variability is highest in Africa but intergroup variability is highest in the New World and in Melanesia. This is because New World populations have been diverging from each other for a long period of time without much intermixing and population growth, whereas African populations experienced a population growth (from 50K on, according to archaeology and some genetic studies) and population mixing after the original period of isolation and diversification.

    The build up of phonetic inventories in Africa may reflect this population growth. It’s also likely that click phonemes in Khoisan languages are responsible for the cline between South Africa and the rest of the world. Click phonemes exist on top of more common phonemes. As clicks are not attested in other languages outside of Africa (Lardil is Australia is the only exception but it’s a ritual language, hence it doesn’t count), it’s likely that they emerged in Africa after Africa had been peopled or after Khoisans separated from the rest of humanity. In any case, clicks must post-date the emergence of non-African humans. If the excess in African phonemic inventories is due to clicks but clicks emerged in Africa after the separation between Africans and non-Africans, then the cline that Atkinson reported is not indicative of the African origin of speech and modern humans. It’s also unlikely that the serial bottleneck idea (already falsified by the Denisova and Neanderthal “genetic admixture” results) can mechanically translate into the phonemic level.

    It’s also worth pointing out that Africa and Europe are markedly less diverse in terms of grammatical structures (e.g., word order types) and the number of language families (20 in Africa vs. 140 in the Americas). Again, it makes the out of Africa model with a serial bottleneck unlikely.

  • http://washparkprophet.blogspot.com ohwilleke

    IIRC, this Atkinson is the same one who made headlines in 2003 by co-authoring a paper with Gray arguing from lexical diversity (i.e. differences in word roots) that the Indo-European languages were about 8,000 years, supporting the disfavored Anatolian hypothesis.

    But, he somewhat redeemed himself in 2008, by playing with the original model, which had assumed a constant rate of linguistic change, to estimate that a substantial amount of lexical change happens in a punctuated way as languages differentiate themselves (more than 20% of the variation in Indo-European languages) by comparing rates of lexical change in languages that had many branches in a known time period to those that had few. A punctuated equalibrium model brings mathematical estimates of language age from lexical data much closer to those suggested by trying to associate archaeological cultures with languages (e.g. the Kurgan hypothesis).

    A clinal, serial founder effect for phonemes has something to be said for it, but I’m skeptical that one can really massage the data to show that effect and that recent linguistic shift hasn’t made ancient serial founder effects too obscure to notice – although I suppose it depends to some extent on how fine grained a search one makes for a cline. If you are looking a subcontinent sized regions, perhaps the impacts of recent history are more modest, because original serial founder effects would be shared by all regional languages.

    The more common view in lingustics has been that languages grow more complex over time, picking up more involved grammatical rules, more phonemic complexity, etc. until creolization, language differentiation or large numbers of adult language learners in the population strip out that complexity and the process starts all over again. Thus, for example, Mexican Spanish dispenses with some of the grammatical nuances found in Iberian Spanish, and the Romance languages generally simplify some of the more elaborate structure found in Latin. Indo-Europeanists say that successor Indo-European languages have only partially overlapping subsets of the original phonemic complexity of that language (presumably lost as the languages expanded and were adopted by non-native speakers), but that would suggest a cline from proto-languages to major modern languages, not a serial founder effect from Africa. Since places closer to Africa experienced the Neolithic revolution, on average, sooner than places more distant from Africa, major language expansions may be younger in the East than in the West.

    The Caucasus mountains region has far more phonemes than the areas around it, for example, which has to be an interruption in any cline from Africa to there and beyond. But, this makes sense if one sees the Caucusian languages as languages in a linguistic refugia that haven’t expanded much since their formation and have thus have time to develop complexity not present in expanding languages.

    The time scale for a serial founder effect from Africa would have to be tens of thousands of years ago, yet shrinkage in the phoneme set in modern languages has occurred much more rapidly. English for example, has reduced its phoneme set in the time period that has elapsed from Old English to contemporary English. For example, in English, as Language Log recently noted, words that start with “wh” (which was originally spelled “hw”) started with an hw-like phoneme that is no longer a part of modern English and several other h-consonant phonemic dipthongs are gone too.

    Another example of a phoneme set that has shrunk substantially in the historical era is Japanese. The palate of sounds represented in the oldest written Japanese documents is substantially broader than the existing palate of founds in spoken Japanese, something that might be a product of the integration of large population of adult language learning Jomon into the Yaoyi linguistic sphere.

    The time scale in which change in phoneme number is observed to take place is all wrong for Atkinson’s hypothesis.

    Also, if Atkinson’s cline hypothesis were true, this would mean that phonemes generally survive linguistic shift. This point is supported by the retention of click sounds in some Bantu languages. It also would imply that West African languages should have richer phoneme sets than English. Yet, if this is true, why aren’t African-American dialects of English, and African-American influenced dialects of English in the American South more phoneme rich than other dialects of English (the most phoneme rich dialects of American English are in New York City).

    There are also counting issues. For example, functionally, tone serves essentially the same purpose as more conventional phoneme, but, the use of tone in language is not always counted as part of the phoneme set. Many languages of Asia make heavy use of tone. So, if one fails to count tone variations as phonemes, one is going to systemically undercount the number of phonemes in Asian languages relative to those closer to Africa.

    Also, there is quite a bit of phoneme detail in many languages that isn’t captured by rudimentary early rounds of linguistic analysis or part of what even educated literate uses of a language realize is part of their repetoire. For example, few people other than dialect specialists are even aware that the number of vowel sounds in the American English language varies in different dialect regions. People who speak vowel poor dialects of American English often can’t even hear the differences between different vowel sounds that are heard as distinct in vowel rich dialects of American English. In a survey of 504 different languages, it is very plausible that the sophistication of the linguistic work declines as one gets further from Africa, for example, for Native American, indigenous Siberian and Melanesian languages. Less sophisticated linguistic work will systemically miss phonemes, but is unlikely to find phonemes that aren’t there. Moreover, if the survey of 504 languages were to use a single number to describe the number of phonemes in the English language it would be wrong. Some dialects of English have different numbers of phonemes than others.

    Bad data necessarily makes conclusions drawn from that data unreliable, no matter how good the statistical analysis of the data by the researcher may be.

  • http://www.kinshipstudies.org German Dziebel

    Interestingly, the best-fit second origin region is South America (Suppl. Mat Fig. S6), which is precisely the area of world highest language diversity measured in terms of independent stocks. Amazonia is the area with the greatest number of language isolates, which are also small populations.

    Also, it’s not Khoisan or Nilo-Saharan, but Hmong-Mien, Nakh-Daghestanian, Wakashan and Salishan language families that hold world highest values of consonantal diversity. Macro-Ge tops the bill on vowel diversity. Oto-Manguean exceeds Khoisan and Nilo-Saharan on tonal diversity, although it’s the paucity of tones that drags total diversity values down for Amerindian languages. Tai-Kadai, Sino-Tibetan and Hmong-Mien have the highest tone diversity in the world.

    Finally, Atkinson makes a claim that “Simulations of language evolution also predict that rates of change will be faster in small populations, slowing as populations grow (S32)” (Suppl Mat p. 9). They cite Nettle, but this thesis was later disproved by Soren Wichmann (email.eva.mpg.de/~wichmann/population_upload.pdf). Atkinson also caveats “Note that faster rates of change in small populations do not entail greater within- population phonemic diversity. Whilst new distinctions can spread to fixation more quickly in small populations, existing distinctions are more easily lost.” What this means is that if a population is small, it may be old but the rate of phoneme extinction will be much higher than that of a large population. This again invalidates the out of Africa serial bottleneck interpretation advanced by Atkinson. Even if the stats are right it only means that the New World and Oceania were simply worse than Africa in retaining new phonemes.

    If language family mean is a good proxy for this family’s proto-language values, then it’s hard to see how African languages can have the highest phonemic diversity values. I think they arrived at this conclusion by cherry-picking individual languages that often do show the highest values across consonantal, tonal and vowel categories in Africa.

  • Matt

    Stuff:

    If this is true, interesting that it could be unobserved for such a long time. Phonemic diversity wouldn’t seem to decline by a subsetting method like genetic diversity does from Out-of-Africa, so I assume the model would be that subsetting happens but the chain shifts cause alterations (to entirely new phonemes which probably were not part of the last subset) which cause it not to look like clean subsetting. Maybe this is what caused this pattern to be missed for so long?Exceptions like the highly phonemically internally diverse Caucasian languages (if I remember correctly) might also confuse the pattern (as we’d be talking about a tendency, not a rule).

    The figures in the paper also seem to show a second peak in phonemic diversity at 25000 from Africa, with pretty equal phonemic diversity per language to Africa – I wonder what kind of region are we talking about there? Depending on what this region actually represents, maybe historical population density might be a better proxy than distance from Africa…

    Theoretically, it’s tough for me to understand why phonemic splits (which happen) wouldn’t tend to either at least maintain or increase phonemic diversity rather than reduce it – it would seem like there are advantages to having more phonemes, even if we accept that Heinlein Speedtalk is kind of absurd, i.e. this isn’t neutral diversity. I guess the counterargument would be that phonemic loss is much easier to compensate for through other linguistic strategies than I’m assuming it is, possibly particularly as our linguistic capabilities evolve, so phonemic loss doesn’t really matter, at least to the point where there is a strong tendency not to replenish phonemic diversity.

    Also, for a theory like this, I guess it would be interesting to test this via comparisons with understood language families that we have records of in historical time and fairly good reconstructions of, although the sample size of this of course would be pretty small. I.e. do Indo-European languages (or Sino-Tibetan languages or Afro-Asiatic languages) low down in the family tree tend to display less internal phonemic diversity than earlier/higher up languages? That would argue in favor of this paper, while the contrary would argue against it.

  • http://www.riverellan.blogspot.com Tom Bri

    An interesting question is what, if any, is the practical maximum # of phonemes a language can have. It would seem to me that if this argument is correct, and African languages exhibit more phonemes, it must be because these languages have had a longer time to acquire them. Out of Africa dates can be argued, but let’s just say 100,000 years. This implies that the age of modern language ability must be considerably in excess of this time period, or the daughter languages would have had time to redevelop phonemes and catch up to African diversity levels. If there is a practical maximum of phonemes.

  • http://simon.net.nz Simon Greenhill

    @ohwilleke – yes, it’s the same Atkinson as the 2003 Indo-European origins research. You are wrong, however, in stating that the method required a constant rate of change. It allowed rates to change between languages and over time. It’s also been replicated by about five papers. There’s a great blog post on this here

    Simon

  • http://www.kinshipstudies.org German Dziebel

    @Simon Greenhill

    “It’s also been replicated by about five papers. There’s a great blog post on this here”

    With 80 comments most of which are critical of the method and results from vastly different perspectives.

  • http://simon.net.nz Simon Greenhill

    …and most of those are from people with utterly unorthodox “perspectives” not supported by any evidence at all – like yourself. The rest are from people who haven’t bothered to read any of the articles, and complain about irrelevancies (“rates of change aren’t constant” for example)

  • http://www.kinshipstudies.org German Dziebel

    “…and most of those are from people with utterly unorthodox “perspectives” not supported by any evidence at all – like yourself.”

    It’s yours truly who works with facts and evidence, you apparently have only numbers. But numbers don’t count. Gray, Atkinson & Greenhill 2011 lumped together a set of extant languages with a set of extinct languages. There’s an infinite number of possibly extinct languages that are as divergent and as IE as Hittite and Tocharian but weren’t recovered by archaeologists. One of them may have been spoken in the Pontic steppe where most archaeologists locate the homeland of Indo-European. Your basal level of IE is made up of 2 extinct language clusters – Hittite and Tocharian – which are as widely apart from each other geographically as Turkey and China. You can’t replicate Anatolian homeland with the closed set of extant languages, just because there’re no extant languages spoken in Anatolia, but then you base your homeland conclusion only on extinct languages of Anatolia. You just have too many logical flaws in your analysis (disjunctive set, circularity) to make anybody without an upfront partiality for numbers to believe you.

  • onur

    i

  • Onur

    Gray, Atkinson & Greenhill 2011 lumped together a set of extant languages with a set of extinct languages.

    They just used all the available languages – whether extant or extinct – for their work, and it is better than using only the extant ones, as the extant ones are more limited than the extant and extinct ones combined.

  • http://www.kinshipstudies.org German Dziebel

    “They just used all the available languages – whether extant or extinct – for their work, and it is better than using only the extant ones, as the extant ones are more limited than the extant and extinct ones combined.”

    I get that part, Onur. But this is where a flaw came in: you can’t rely on extinct languages to draw a radical conclusion on the homeland of a language family. It’s an open, infinite set. Especially if there are only 2 elements in this set and doubly so if the outcome is a judgment on geographic center of dispersal but those 2 elements occur in such widely separated geographic areas as Turkey and China.

    The decision on the composition of a dataset should take into account the objectives of the research and the theoretical environment in which the research is taking place. The competing theory, namely the Kurgan theory, is based on archaeology into which linguistic data is retrofitted because all the original Pontic IE languages didn’t survive. Gray, Atkinson and Greenhill 2011 do nothing to undermine the Kurgan theory because they base their conclusion on half (Anatolian) of the known set of extinct languages.

  • Onur

    German, by your logic, using only the extant languages to draw conclusions on the IE homeland should be equally or even more invalid than using both the extant and extinct languages for that aim and conclusions should instead be based on archaeology (how much the Kurgan theory is supported by archaeology is another issue).

  • http://www.kinshipstudies.org German Dziebel

    “German, by your logic, using only the extant languages to draw conclusions on the IE homeland should be equally or even more invalid than using both the extant and extinct languages and conclusions should instead be based on archaeology (how much the Kurgan theory is supported by archaeology is another issue).”

    By any logic, Onur, if your basal level only has 2 extinct languages and they show a wide geographic spread, which is not found in the extant languages, it’s hung jury, and all judgments on the geographic center of dispersal of this family on the basis of a set of extinct and extant languages are futile. They won’t satisfy a necessary condition for a testable theory and they won’t provide a sufficient alternative to the existing theory based on archaeology because archaeology can detect a culture whose language left no direct descendants.

  • Onur

    Both theories are based on archaeology; the difference is that the Anatolian theory also has the language family tree in its support… and no, you can’t make me believe that a language family tree with only extant languages is more valid than a language family tree with both extant and recorded extinct languages.

  • http://www.kinshipstudies.org German Dziebel

    “you can’t make me believe that a language family tree with only extant languages is more valid than a language family tree with both extant and recorded extinct languages.”

    It’s not more valid. It’s just two different trees. And the extinct languages tree is not robust.

    “the Anatolian theory also has the language family tree in its support”

    The extinct languages tree supports an Anatolian homeland just as much as it does a West China homeland.

    “Both theories are based on archaeology, the difference is that the Anatolian theory also has the language family tree in its support”

    The Kurgan theory can postulate an extinct Kurgan language and be square. We can be absolutely sure it existed. Or you think these archaeological cultures that are older than Hittite had no language?

  • Onur

    Or you think these archaeological cultures that are older than Hittite had no language?

    They certainly had languages, but not necessarily IE ones.

    The extinct languages tree supports an Anatolian homeland just as much as it does a West China homeland.

    As Dienekes stated, parsimony favors the former.

    It’s not more valid. It’s just two different trees. And the extinct languages tree is not robust.

    Why is that so?

  • http://www.kinshipstudies.org German Dziebel

    “As Dienekes stated, parsimony favors the former.”

    Parsimony has nothing to do with this situation. Dienekes admits an ascertainment bias but calls it “parsimony.”

    “They certainly had languages, but not necessarily the IE ones.”

    The archaeological support for the Anatolian theory that you mentioned earlier comes from dates far preceding the attested dates for Hittite. It means those archaeological cultures that document the emergence of agriculture in Anatolia may not have been IE-speaking either. Again this is not a valid argument.

    “Why is that so?”

    Because there are only two elements in this set – Tocharian and Hittite. And they are widely separated geographically.

  • Onur

    Parsimony has nothing to do with this situation. Dienekes admits an ascertainment bias but calls it “parsimony.”

    I see the language family tree as a supplementary, so “favors” is far from “proves”.

    The archaeological support for the Anatolian theory that you mentioned earlier comes from dates far preceding the attested dates for Hittite. It means those archaeological cultures that document the emergence of agriculture in Anatolia may not have been IE-speaking either. Again this is not a valid argument.

    Validity is in that the archaeological support for the Kurgan theory is as much suspect as the archaeological support for the Anatolian theory. That is why I refrain from sticking to either of the theories.

    Because there are only two elements in this set – Tocharian and Hittite. And they are widely separated geographically.

    But enough time had passed since the separation of the common ancestors of the speakers of the either languages to allow for such a big geographical separation when Tocharian was recorded (1st millennium CE).

  • http://www.kinshipstudies.org German Dziebel

    “Validity is in that the archaeological support for the Kurgan theory is as much suspect as the archaeological support for the Anatolian theory. That is why I refrain from sticking to either of the theories.”

    So do I. I’m more interested in the Tocharian/Y-DNA R1a link right now. I’m also doing lots of etymological work in IE.

    “But enough time had passed since the separation of the common ancestors of the speakers of the either languages to allow for such a big geographical separation when Tocharian was recorded (1st millennium CE).”

    Sure, and some languages went extinct without a trace.

  • Onur

    I’m also doing lots of etymological work in IE.

    You can share with us your findings whenever you want. I would be glad to learn them.

  • http://www.kinshipstudies.org German Dziebel

    I will. Thank you for your interest, Onur.

  • Fogbraider

    @ohwilleke

    “Less sophisticated linguistic work will systemically miss phonemes, but is unlikely to find phonemes that aren’t there.” I would disagree – trained phoneticians can hear a myriad fine distinctions. This is the empirical data. Phonemes, however, are a bit like species in biology – the product of a process of analysis and classification. Take, for instance, the sound /t/ in English. This sounds slightly different in different contexts, e.g. before a vowel, between vowels, or word-finally – these allophones of /t/ are acoustically different.

    The main criterion for isolating phonemes is that they distinguish one word from another, e.g. /t/ and /d/ are different phonemes in English because we can find minimal pairs like ‘toe’ and ‘doe’ where the difference between words depends solely on the difference between these two sounds.

    There are many possible complications. For instance, a language might have a sound that only occurs at the beginning of words and another that never occurs at the beginning of words, so the minimal pair test can’t be applied. Whether they are treated as allophones of the same phoneme will be a subjective judgement based on how acoustically similar they are. (This might be modified by a knowledge of the language’s history and dialects, e.g. the glottal stop form of /t/ is pretty distant acoustically from the usual alveolar /t/, but we know how the sounds correspond between dialects.)

    Sounds slide into each other, and it is again a subjective judgement whether we treat clusters as individual phonemes or as sequences of phonemes, e.g. English /st/ is treated as two phonemes, but the sound spelled ‘tch’ as one.

    For languages that have been given an alphabetic written form, these judgements have already been made, generally in a maximally parsimonious way. So we would expect alphabetically written languages to be analysed into a smaller number of phonemes.

  • BoNo

    “So do I. I’m more interested in the Tocharian/Y-DNA R1a link right now. I’m also doing lots of etymological work in IE.”

    Please eloborate – when appropriate!

    “The archaeological support for the Anatolian theory that you mentioned earlier comes from dates far preceding the attested dates for Hittite. It means those archaeological cultures that document the emergence of agriculture in Anatolia may not have been IE-speaking either. Again this is not a valid argument.”

    The Hittites may have arrived with the agriculture to Anatolia. They spoke IE.
    A major part of agriculturalization – such as domestication of animals – should definitly not be ascribed to the futile Crescent. The first animals of domistication were wolf, goat and sheep – which is a matter of arctic animals. The oldest cows and oxes are found in Denmark, where they dwelled 9500 years ago. The only demografic groups in the world where 90% can digest cowmilk are placed in the fertile Crescent from southern Holland – via Denmark and Skane – to Pomerania and Wizla in Polen. The further you come from this core-area the less lactose-persistance you will find. Anatolia hardly reach 50 % and are no more milk-drinkers than the Arabs, the Kenyans, Etiopians and Maroccoans.

    What is interesting about this repport is the span of 7.1000 – 9.800 year as the appearance of the IE language – along with the first palefaced Eurasians, the population of northern Euraisa, the domestication of plants and animals, the initiations of inter-regional production, travel and trade and the introduction of pastoralism, slash-burn-farming, cow-farming and various other forms of agriculture. Megalithic structures seems to have the same period of appearance – soon followed by monumental architecture and the rise of the old civilizations – in which the language was developed into ‘families’ – that would follow the demographic maps that differ between etnic groups such as Chineese, Indian, Persian, Arabian, Greek and Egyptian.

    Perhaps we can come to some kind of basis for the comparision of languages if the range and variations of sounds used in various languages could be met and compared. Which language have the clear sounds (vocals) and how many do they have? Then, who have the widest range of con-sonances?!

    The answers should be in some kind of alignmeent with the phonetic alphabets that do exist already. Unless there is such a reference-grid it will be futile to draw comparisions between languages in terms of defining their ages.

  • http://simon.net.nz Simon Greenhill

    @German – we only have “numbers”??! What do you have then, weak conjecture and rampant speculation? Your discussion of extinct vs. extant languages being an “open” “infinite” set shows that you clearly don’t understand the methods we use.

NEW ON DISCOVER
OPEN
CITIZEN SCIENCE
ADVERTISEMENT

Discover's Newsletter

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

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at http://www.razib.com

ADVERTISEMENT

See More

ADVERTISEMENT

RSS Razib’s Pinboard

Edifying books

Collapse bottom bar
+

Login to your Account

X
E-mail address:
Password:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »