December 26, 2012

“Mismodelling Indo-European Origins” Talk

Martin Lewis and Asya Pereltsvaig have been critical of the recent paper on Indo-European origins on the GeoCurrents blog, and they recently gave a talk at Stanford on the topic.



Some relevant past posts:

I think that the Indo-European question has been debated for more than two centuries without any clear resolution. Over the next few years, I think that either of two things will occur:

  • A clear unambiguous pattern of expansion mimicking the IE dispersal will appear in ancient DNA, providing the "smoking gun" for one of the different hypotheses.
  • No such Eurasian-wide pattern will emerge, and it will turn out that Indo-Europeanization was effected with minimum dispersal of populations.
I suspect the former will be the case, but it will nonetheless be interesting to see how the different parties coming from archaeology and linguistics will react to the (archaeo)genetic avalanche that will doubtlessly provide us with new information about the prehistoric past.

53 comments:

  1. In most areas of science, this would be regarded as a rogue and extremely desperate talk. It is acceptable that you spend a few minutes criticizing opponents, but largely, you need to provide your own evidence and research.

    Also, some of the criticism is trivial. We know many of the limitations of such models. For example, in genetics it is just now that admixture is being incorporated into tree models. Same here with borrowing.

    You simply cannot criticize a new, rapidly-evolving and improving model just based on its trivial, known shortcomings. Such a thing is ludicrous and paints a truly bad picture of the talk presenters.

    ReplyDelete
  2. @eurologist

    You should read their blog posts (there are dozens of them, with all the details). There is much much more to this.

    Also "one's own evidence and research" is in fact 200 years of sound science that these new models, all of them developed by non-linguists, of all people, want to sweep aside and replace. Let's not shift the burden of proof.

    You don't parachute into a discipline saying you are going to debunk an established paradigm (that's essentially what is happening with these models) and expect a smooth reaction. Making elementary mistakes that betray unfamiliarity with the discipline at hand, such as some of those papers do, won't help, either.

    ReplyDelete
  3. Indo-European expansion was surely a multi-stage process involving multiple waves in the same and sometimes overlapping directions.

    For example, we know to a fair degree of confidence that speakers of the Indo-European Baltic languages were faced with a wave of Slavic speakers (from a related branch of the Indo-European language family) who brought about language shift for most of the Baltic linguistic region and also brought about partial demic replacement/dilution.

    Similarly, we know that in Britain there were separate waves of Celtic, Anglo-Saxon, Danish and Norman migrations, all Indo-European half from the Germanic branch of Indo-European languages, and half from Celto-Italic branches.

    Similarly, we know that an important wave of Indo-European language expansion in Western and Northern Europe (whether or not it was the first such wave) happened fairly close in time to Bronze Age collapse (ca. 1200 BCE), while Indo-Iranian expansion had probably differentiated into early Rig Vedic Sanskrit, the root of the Indo-Aryan languages, by about 2000 BCE, eight centuries earlier.

    It is entirely possible, and indeed likely, that Indo-European expansion was a serial founder effect event in which early waves caused language shift in substrate populations that were incorporated into later waves.

    For example, Pontic populations could have conquered and incorporated Anatolian or Caucuasian or pre-Greek populations who came to speak an Indo-European language as a result of the conquest, before expanding further and diluting the original proto-Indo-European population genetic signature in later waves.

    Statistical models, and in particular Baysean statistical linguistic models, need to incorporate the facts that are already established from other fields such as history and archaeology to the fullest extent possible, and also tend to overestimate the importance of random drift relative to punctuated language change upon language contact or the initial differentiation of a new language.

    ReplyDelete
  4. ~ 4500 BC: invention of the potter's wheel
    This is very close to the date of the separation of Pseudo-Tocharians with the other IE ~4800 BC(date in the Bouckaert work).
    The Hittite language possibly borrowed the word for wheel from pre-Pseudo-Tocharian.

    The Scythian language first recorded in Saqqez(west Iran), so this won't change the results in the program.

    ReplyDelete
  5. First, I agree with Eurologist's comment. It is rather extraordinary to start with ad hominem attacks.

    I don't have a dog in this fight. I think both camps could be right. All it would take is PIE to originate in N. Anatolia and then somehow acculturate the horse rearers of the steppe, who were responsible then for the massive spread.

    Having disclosed my biases, let me say how much of this lecture is way overstated. Around minute 28, she says, "we know PIE can't be so old because it has a word for wheel, and wheels were invented in 3500 B.C." That is just demonstrably false.

    I won't go into the scientific. I'll give a more timely example. A lot of social-networking sites use the word, "circle." Meaning "friends list" or "circle of friends." Thus, cell-phone companies now use the term too. There is that "Friends and Family" commerical from one of the major carriers that says, "call anyone in your circle for free..."

    Now imagine 1000 years from now, "circle" is entirely used to mean, "circle of friends" with a cyberspace (telcom, website) connotation. It could easily happen.

    It would simply be inaccurate to say, "we know the languages could not have gave birth to the cognate 'circle' before 2012 AD because there were no cell phones or computers."

    The same goes for the word "wheel." It is possible, indeed likely, that the original term for wheel was a re-assignment of an existing term, like my circle example above. Building on the circle meaning (except going a different route), early wheel-creators could have taken the word for "sphere" or "legs" or "speed" and re-assigned it to the term for "wheel."

    This happens all the time! Note even the darn lecturer -- in her own lecture -- notes how "knecht" in German (SERVANT) was re-assigned to be "knight" (minor nobility who SERVED as vassal) in English.

    Thus her statement, "the language couldnt have had a word for something that didn't exist yet" is LAUGHABLE! Knights didn't exist before the Middle Ages, yet the languages all had a word for "knight." That's because the word meant "servant" before the Middle Ages!

    ReplyDelete
  6. They're exposing how applications of mathematical / computer statistics may sometimes be severely flawed science; and how it might be distorted in order to support certain preconceived ideologies. (ie Garbage in, garbage out.) Intuitively I knew that, but it's good to hear someone actually seriously challenging it with clear explanations, examples. The origins of Indo-Europeans is too important a topic to ignore, exagerate, romanticize, idealize or treat lightly.

    ReplyDelete
  7. "You simply cannot criticize a new, rapidly-evolving and improving model just based on its trivial, known shortcomings. Such a thing is ludicrous and paints a truly bad picture of the talk presenters."

    I'm afraid your effeminate idea of proper protocol has no bearing on actual science. Gray and Atkinson's "innovation" is insisting that Bayesian phylogenetics with limited and sometimes questionable inputs of data can produce highly accurate and precise readouts of linguistic history that supercede all previous linguistic and archaeological knowledge. Their results may dazzle twits like you and appeal to those who find their results politically or ethnically congenial. But the first question a serious person would ask is how closely Gray and Atkinson's attempts at reconstruction recapitulate recent/known linguistic history. That they frequently fail to do so is extremely germane to the question of how much faith one should put in their deeper reconstructions.

    Statistical models are not magic. Bayesian tree building is not magic. Even with large corpuses of genetic data, the "most likely" tree is often overwhelmingly likely to be wrong. For genetics, where there's an explosion of data with comparatively few human analysts and little or no historical context, such results are useful, being often the best we have until additional data and further refinements of models appear. On the other hand, in linguistics, where on the PIE question relatively many human analysts have been poring over a comparatively limited corpus for many decades, it's up to Gray and Atkinson to demonstrate they have something useful to contribute. Every indication says they do not.

    ReplyDelete
  8. The presenters do not seem to be major league acedemics. One commentor did present them with the orthodox method of critique.
    But from what I could tell the rather basic problems with the research seemed clear. Watching it a second time there were a number of clear methodological critiques put forth. As in psychology and economics at least, using "quantitative" methods even improperly seems to lend a scientific rigor. The legitimate question is how this paper passed Science scrutiny.
    The sad truth is that both Science and the NYTimes have taken to supporting their political agendas in their content

    ReplyDelete
  9. "It is entirely possible, and indeed likely, that Indo-European expansion was a serial founder effect event in which early waves caused language shift in substrate populations that were incorporated into later waves".

    To me that seems avery likely scenario. We will not find 'A clear unambiguous pattern of expansion mimicking the IE dispersal will appear in ancient DNA, providing the "smoking gun" for one of the different hypotheses'. The haplogroups too will have undergone a knock on effect.

    ReplyDelete
  10. The best candidate for a genetic signature of the Indo-European expansion is R1a1a.

    Breaking it up into as many subclades as possible, and aging them correctly, should tell us precisely how the early Indo-Europeans spread out from Eastern Europe.

    Unfortunately, this might take some time at academic level. Recently, Italian scientists claimed they couldn't find any R1a1a in Italy, despite the fact that the haplotypes from their own study showed three subclades - R1a-Z280, R1a-M458 and even R1a-Z93.

    On the other hand, Indian scientists keep insisting that R1a is native to South Asia, even though only one subclade has ever been found there (R1a-Z93). They even managed to turn an R2 sample into R1a* in one of their studies.

    ReplyDelete
  11. """Davidski said...
    The best candidate for a genetic signature of the Indo-European expansion is R1a1a."""

    East European branch 70% R1a1a1b1-Z283 Erzya(Uralic people)
    http://www.familytreedna.com/public/Erzya-Moksha%20DNA%20Project/default.aspx?section=yresults


    Asiatic branch 82.5%(33/40) R1a1a1b2-Z93 Khoton(Turkic people) from northern Mongolia http://s020.radikal.ru/i717/1212/5e/e2ad61db8f36.jpg

    ReplyDelete
  12. You don't parachute into a discipline saying you are going to debunk an established paradigm (that's essentially what is happening with these models) and expect a smooth reaction.

    Teo,

    I have read numerous papers on numerically constructed trees, and haven't seen one that claims its goal is to debunk established paradigms.
    There is a difference between a not-so-smooth reaction and uncivilized acts of desperation caused by wild terror and years of incompetency selling a position and field of study properly. Clearly, what's at stake here is the future of the field, and with that, grant money and faculty positions.

    This is the larger picture that many posters in this thread seem to fail to recognize.

    I'm afraid your effeminate idea of proper protocol has no bearing on actual science. Gray and Atkinson's "innovation" is insisting that Bayesian phylogenetics with limited and sometimes questionable inputs of data can produce highly accurate and precise readouts of linguistic history that supercede all previous linguistic and archaeological knowledge. Their results may dazzle twits like you and appeal to those who find their results politically or ethnically congenial.

    n/a,

    You realize that using such ad hominems not only uncovers your poor argumentation skills and makes readers unlikely to follow what you have to say, but also makes you an easy target to get banned from this site?

    What you state Gray and Atkinson insist upon is not so. Clearly, the current methods are still lacking in many ways, and most practitioners are cognizant of that. However, if emerging methodologies were always blasted like this in their infancy, there would be no innovation in science.

    The wheel argument has been debunked hundreds of times, and concentrating on "mistakes" of the tree model in the past 2,000 years and at the fringe is outright silly: there the neglect of "admixture" is accumulative and the largest.

    ReplyDelete
  13. "All it would take is PIE to originate in N. Anatolia and then somehow acculturate the horse rearers of the steppe, who were responsible then for the massive spread."

    Or a first wave of sheep-herders on foot spreading out in multiple directions from one heimat which - in those regions the first wave settled which later turned out to be the best suited for horses and horse-based herding - later evolved into a horse based culture which expanded in a dramatic second wave.

    i.e. a horse-based IE expansion over the top of an earlier foot-based IE expansion.

    If true or partly true though i'd guess that would imply there ought to be mountain vs valley language dialect distinctions in various places?

    ReplyDelete
  14. ^ Davidski,

    "They even managed to turn an R2 sample into R1a* in one of their studies."

    Could you please share the source of that omission?

    That would be quite a mistake on their part as R2a-M124 can be discerned quite easily from R1a1-M17 on even 12 STR's.

    ReplyDelete
  15. mooreisbetter: "Thus her statement, "the language couldnt have had a word for something that didn't exist yet" is LAUGHABLE! Knights didn't exist before the Middle Ages, yet the languages all had a word for "knight." That's because the word meant "servant" before the Middle Ages! "

    George: "~ 4500 BC: invention of the potter's wheel"

    OK but the thing is that the word for wheel spreaded from ONE original form in PIE (there is apparently a consensus on this from linguists, they are not subsquent loanwords) with the same meaning, so it had to mean "wheel" at the time the PIE separated in several populations that were vectors of IE-ization leading to the several different IE branches.
    Thus the time of the PIE stage had to be carrier of a PIE word for wheel (at least that's by far the most logical explanation) -- but also of words for a metal or whool (both also identifying a specific moment in time, roughly in the same time (chalcolithic would work well with it)).

    That being said, it seems that there were two words for "wheel" (kwekwl-, rot-) in PIE - some language have only one of those roots and some languages have both (like apparently Indo-iranian and germanic), so I guess maybe it could actually be more open than that about the original meaning of one of them, but it's still weird they all ended up meaning only specifically "wheel" despite the diversity of the IE branches (notice how knight and knecht don't mean the same thing!).

    ReplyDelete
  16. "The same goes for the word "wheel." It is possible, indeed likely, that the original term for wheel was a re-assignment of an existing term, like my circle example above.... This happens all the time!"

    Inded it does. There is a similar conundrum in Muskogean around the term for corn. The etymon, cognate across all the languages, pre-dating the splits, is a lot older than the cultigen.

    http://www.peopleofonefire.com/proto-musk/reconstructing_protomuskogean_environments.htm

    And not just Muskogean. Somethng similar applies to corn terminlogy in Uto-Aztecan. The etymon that gave rise to 'tamal' in Nahuatl can refer to mush made out of wild grain in some langugaes. So what was the direction of semantic extension?

    ReplyDelete
  17. After many years of cutting up this problem from a lot of angles my guess is that the SPREAD of IE languages was primarily a matter of economics. Domestication of plants and animals, the horse, the wheel, metal mining and working -- all of these spread through markets and trade routes -- both know-how and the actual products -- all depended on the exchange of information. Common languages are essential for this kind of dispersal. The ORIGIN of IE languages therefore might have been incidental. I witnessed a Finn and a Greek speaking to each other in English about matters that had nothing to do with English. There is a tendency for languages to diverge over time and that creates the need for a common language. PIE (the proto-language) may have been no more than the language of middle men who were mediating trade and know-how between different people who spoke different languages. Eventually these people may have been by-passed but their language stayed common.

    ReplyDelete
  18. About my former post, a little oversight:

    "it seems that there were two words for "wheel" (kwekwl-, rot-) in PIE - some languages have only one of those roots and some languages have both (like apparently Indo-iranian and Germanic)"

    Oops. And Tocharian as well.


    @ Unknown: "There is a tendency for languages to diverge over time and that creates the need for a common language"

    Lingua francas exist without replacing the local languages, it just brings loanwords in it. Your Finn will still speak Finnish and your Greek will still speak Greek.

    ReplyDelete
  19. The language phylogeny paper they criticize was a very important paper. But no matter how good we should not holding it up as an object of worship that cannot be criticized. It SHOULD be criticized. Otherwise it just becomes religion.

    I personally found some of their arguments compelling, no matter how obvious their personal agenda was. The effect of borrowing on Romanian is similar to the effect of genetic admixture for example. The impact of conservation on Sardinian also was interesting.

    The model DOES NOT fit some of the known facts, but that is reasonable as the model is a simplification of what even the authors know to be a reality of surges and climatic effects.

    Frankly I suspect the authors would be grateful for criticism which allows them to refine the model.

    ReplyDelete
  20. It is possible that the Potter's wheel, is known from the beginning of "Early Neolithic Pottery" 6500 BC

    How did they manage to create this spherical pottery, in early Neolithic pottery?

    ReplyDelete
  21. George,

    The Uralic speaking Erzya and Moksha carry a subset of R1a diversity seen in Central Europe.

    Are you suggesting that Uralics spread from Central Europe to Asia?

    ReplyDelete
  22. "What you state Gray and Atkinson insist upon is not so. Clearly, the current methods are still lacking in many ways, and most practitioners are cognizant of that. However, if emerging methodologies were always blasted like this in their infancy, there would be no innovation in science."

    Whatever shortcomings they may acknowledge, they claim to be able to discriminate between different models of IE origins, while having failed to demonstrate their techniques have anything like the accuracy or precision needed to do so.

    Again: applying similar techniques to genetic data, one can often expect at best a rough approximation of the truth. Estimates of coalescence times may come with very wide confidence intervals. Looking at language rather than DNA, one should expect more error (including sources of error not accounted for in the model) and cruder approximations of reality. If historical linguistics had never happened, Gray and Atkinson might be seen as a crude start. Since historical linguistics did happen, it's again up to G and A to show their techniques produce results that are at all informative within the context of PIE origins "debate".

    Your insistence that Gray and Atkinson are being unfairly "blasted" and need protection remains disgustingly effeminate. Reminder: they're the ones, publishing in Science and getting uncritical NYT writeups, who claim they have the ability to arbitrate the PIE question (coincidentally, in favor of the specious / anti-empiricist / politically popular answer).

    The issue is not with computational linguistics in general (though exactly what such approaches ultimately will contribute remains an open question). The issue is with Gray and Atkinson's overselling the usefulness of their particular models.


    "You realize that using such ad hominems not only uncovers your poor argumentation skills and makes readers unlikely to follow what you have to say, but also makes you an easy target to get banned from this site?"

    You realize this is a comment section of a web log and your opinions are not important, particularly when they're backed up by minimal understanding. You're a fucking twit, and this will be obvious to most people reading your comments, even if they're too polite or have considered you too insignificant to tell you this. Your contributions are on about the same level as those of "Annie Mouse". Read some of her comments. This is how you come off to other people.

    ReplyDelete
  23. """"""Davidski said...
    George,
    The Uralic speaking Erzya and Moksha carry a subset of R1a diversity seen in Central Europe."""""
    Which of the nine R1a1a Erzya haplotypes are close to Central European branches?

    Erzya->Mordva->Mordhvar->Androphagi
    Androphagi in Herodotus times originated to the west of the Northern part of the river Don.

    ReplyDelete
  24. Wow n/a what did I do? LOL

    Do I smell a troll?

    I was not aware the "fucking" was accepted language on this blog?

    Personal attacks are the last refuge of those who believe they are losing an argument. Do you think you are losing the argument and need to intimidate the opposition into silence? Is everyone who disagrees with you a twit? I dont know who you are, but Eurologist has posted often enough to establish his credentials.

    Please argue the facts and not the person. You may disagree with me it appears, but I have an external well-paid intellectual life that provides my affirmation. :)

    ReplyDelete
  25. George

    Androphagi? Man eater? Either some Southern neighbour did not like them or they are just the scary fairy tale cannibals for the edge of the map. I can't see how you identify them with any of the extant ethnic groups.

    Of course your point is that R1a1a is not exclusively IE. That may be true, but most of it's spread in Europe can be connected to IE still.

    (Also even Herodotus lived a way after most of this this happened, even if we take seriously this man-eater = Erzya thing.)

    ReplyDelete
  26. ""Slumbery said..."""
    Well, compare the name of the red haired Vudini(Budini) who were to the north of the Scythians and tribe Vudmurt(part of Udmurts, who are also red haired).
    It is very possible that Vud(ini)=Vud-Murt(Murt=Man)
    And Andro-Phagi=Mord-Hvar=Mordva.

    R1a1a is almost absent among Italic people, Greeks from Italy, Armenians, Ossetians, some Kurds.
    While the Slavic language came together with Christians from the south, see the Christian isogloss "γ" from the south in the Eastern Slavic and in CzechoSlovakian languages .
    Slavs=Those who Glorify (the God), Christian ethnonim.

    ReplyDelete
  27. http://f11.ifotki.info/org/295ebad86bf2400f7a34a6765af8ff574f6726131240808.jpg

    ReplyDelete
  28. I don't have time to moderate, so everyone stop talking about each other and talk about the topic, please.

    ReplyDelete
  29. Annie Mouse,

    I'm sure you're above average in intelligence in an absolute sense, but you're incredibly dim-witted relative to how you apparently rate your abilities. Your self-satisfaction and the overblown importance you accord your semi-informed blathering on things you don't understand is grating. It makes no difference if on any particular topic you happen to approach a point of view I agree with or not. Develop an appropriate level of humility.

    unfortunate pseudonym,

    Bizarre post. Also: "ad hominem"! That I find your desire to shut down debate in the name of "science" disgustingly effeminate and consider you a twit are simply facts, which I felt like relaying. I don't think anyone here was under the impression I was trying to "intimidate" you. Meanwhile, I'm not seeing an argument from you, aside from your belief that Gray and Atkinson should be free to pronounce as they wish on science/history without facing criticism. You've failed to respond to any of my actual points.

    ReplyDelete
  30. @Annie

    quote:"I suspect the authors would be grateful for criticism which allows them to refine the model."


    Alas, I'm not so sure. Atkinson did reply personally to Asya and Martin in the Geocurrents blog, and basically evaded every point raised by them.

    The response to his comment, with a link to it, can be found here: http://geocurrents.info/cultural-geography/linguistic-geography/response-to-quentin-d-atkinson

    ReplyDelete
  31. You've failed to respond to any of my actual points.

    n/a,
    Given the way you presented yourself here, it did not dawn on me you were actually seeking a serious discussion.

    The way I see it, outside your insults, you have primarily repeated known, trivial criticism, made several incorrect points, and otherwise presented straw men:


    Trivial points:

    ...the first question a serious person would ask is how closely Gray and Atkinson's attempts at reconstruction recapitulate recent/known linguistic history.

    They address much of that themselves. But, as I stated when the paper first came out, and also above, yes - there are criticisms. However, when you get to the past 2,000 years and the fringe, errors accumulate based on the imperfections of the input and model (some of which may or may not be resolvable). More importantly, you are looking at the portion (time line) between the vast majority of the data input and today - which simply is not the goal of the paper, and as such, its accuracy cannot be judged based on that (but the lack of quality of input during that time frame can be). The goal of the paper is working in the opposite direction. I and likely most everyone else agrees that the input in the past ~3,000 years can and should be much improved, e.g., with solidly reconstructed language and known localizations and timings (Celtic, proto-Germanic, proto-Slavic).

    it's up to Gray and Atkinson to demonstrate they have something useful to contribute.

    Yes.


    False Points:

    Gray and Atkinson's "innovation" is insisting that Bayesian phylogenetics with limited and sometimes questionable inputs of data can produce highly accurate and precise readouts of linguistic history that supercede all previous linguistic and archaeological knowledge.

    Incorrect. They have stated numerous, serious limitations openly and often (e.g., no early Celtic input --> they miss Celtic in much of W and SW Europe). Also, they are actually making use of "previous linguistic and archaeological knowledge" in the model. And their result of an Anatolian origin is completely mainstream within the 3-5 or more different models of IE origin - it does not supersede anything. They even state that Kurgan expansion may have played a role.

    For genetics, where there's an explosion of data with comparatively few human analysts and little or no historical context, such results are useful, being often the best we have until additional data and further refinements of models appear.

    ...applying similar techniques to genetic data, one can often expect at best a rough approximation of the truth....


    That is a gross understatement of the achievements and accuracy of Bayesian statistical computations in genetics.

    ReplyDelete
  32. continued:

    Straw men:

    I'm afraid your effeminate [sic] idea of proper protocol has no bearing on actual science.

    Of course not. But proper protocol insures that the actual science is taken seriously, that the scientists are taken seriously, that they may have a shot at a career -- and provides an opportunity that the community may support both the science and the scientists. That is the way the real world and funding work. The presenters of this talk are doing a disservice to themselves, to their argument, and to the community.

    Your insistence that Gray and Atkinson are being unfairly "blasted" and need protection remains disgustingly effeminate [sic].

    Of course they should not be protected from criticism - I have criticized them myself.

    your belief that Gray and Atkinson should be free to pronounce as they wish on science/history without facing criticism.

    Again, not so - I have criticized them myself, and of course they should face criticism - just like anyone else.

    it's up to Gray and Atkinson to demonstrate they have something useful to contribute.

    Of course - no one is arguing against that.

    Whatever shortcomings they may acknowledge, they claim to be able to discriminate between different models of IE origins, while having failed to demonstrate their techniques have anything like the accuracy or precision needed to do so.

    In their paper, the authors provide the results of numerous different model calculations. All of them, with a variety of differing assumptions and model parameters, lead to Anatolia as the origin. They make no claim in the paper that they are differentiating between different IE origin models - they just were doing the best ("state-of-the-art") that they could do at the time, which is better than what anyone else has done to date, and show the results.

    Am I convinced? No. I think for almost all of Europe, the most likely origin of IE is in the densely populated, rich cities of the western Pontic that had 1,000 years of contact with the much less numerous nomads of the steppes. I don't know how that relates to Anatolia or further East, other than that I could very well imagine a trans-Pontic Sprachbund early on (which could also explain the long hiatus before further branching). Are there sensationalist science stories in the media? You bet there are.

    That I find your desire to shut down debate in the name of "science" disgustingly effeminate [sic] and consider you a twit [sic] are simply facts

    I have no desire to shut down debate, quite the opposite - I am obviously and clearly trying to enable an actual debate and to guide those misguided about the process. And, yours as anyone's opinions is not fact, even before proven wrong.

    ReplyDelete
  33. Given the lack of indication or willingness of n/a to participate in an actual intellectual discourse, this may be my last response to this recent internet noise:

    @ n/a:

    There is no denying your verbiage "has method." Nevertheless:

    "your effeminate idea," "twits like you," "disgustingly effeminate," "You're a fucking twit" are not mainstream intellectual or argumentative responses. Perhaps you suffer from Tourette syndrom, in which case I suggest you seek medical assistance.

    Further, exclamations like:

    your opinions are not important

    Your contributions are on about the same level as those of "Annie Mouse". Read some of her comments. This is how you come off to other people.

    disgustingly effeminate

    consider you a twit

    ...have similarly no intellectual merit, and further raise the suspicion that it is only the lack of the aforementioned that makes you steep this low. Finally:

    Also: "ad hominem"!

    The accepted plural for the newly-evolved English noun is ad hominems - as much as it pains me.

    I don't think anyone here was under the impression I was trying to "intimidate" you.

    And I know for certain the majority was - based on the feedback I got. And my estimate is that you are perfectly aware of the psychological reasons for your need of excessive cursing (e.g., anger, frustration) and your rhetorical reasons for oozing insults (facing cognitive dissonance; attempts at intimidation out of self-exasperation).

    ReplyDelete
  34. George
    "And Andro-Phagi=Mord-Hvar=Mordva."

    I find this etymology rather forced and unlikely, also unnecessary for the point, because even if it were true, it still would not add anything to the primary question about R1a1a in IE or non IE.

    The Erzya population can be quite old there, but we do not know when they got R1a1a and even if they had it (quite possibly) from very old times, they did not conquer Europe, IE did.

    ReplyDelete
  35. The video represents the strong allegiance that the historical linguistics community has to the whole “Steppes” theory.

    I won’t go into all that’s causing this rather emotional loyalty, but part of it is simply self-defense.

    Historical linguistics has been at family tree-building for longer than biology. But particularly modern computer DNA analysis of descent in organisms has been overwhelming in its success. And it has been very difficult for linguists to do the same thing with their data, which they’ve collected for very many years.

    The most important reason for this failure is that linguists have no real way to connect language change (sound changes in “related” words, cognates) to time in pre-history (that is, before writing.)

    How long, for example, did it take for the hypothetical Proto-European language to turn into German? If you go by historical rates of sound changes in written German (say, medieval to modern), what mathematical rate of change do you get? And how does that apply to the gap between PIE and historical German?

    Some linguists will say languages change at different rates and no statistically valid rate of change can be given. This, of course, is unacceptable without demonstration. To my knowledge, no historical linguist has ever even attempted to establish the rate of sound changes in recorded Indo-European languages. That would be a prerequisite to making such a statement.

    So, linguists are left to making links between linguistic “cognates” and extra-linguistic archaeological data. Part of the definition of “cognate” words is that they have the same meaning. So, of course, the original Proto-European word for the wheel HAD to mean wheel, or else it would not a cognate. In this circular definition, words don’t change meaning in the 3000 years between that 4500 BC date and the first appearance of the word in writing. And all subsequent uses of the “wheel word” for something circular or round are metaphorical.

    There’s the further argument that the wheel is part of a kit of wagon words. Needless to say none of these words can be justified as being neologisms. The same thing can be done with the English words for a computer and its parts and not one of the words is a neologism.

    The word “computer” itself, for example, appeared in English more than 400 years ago and probably well before that in French.

    Historical linguistics has gotten itself into a fine mess with this Indo-European origins thing, no matter how much they feverishly promote it. And they don’t seem to be able find a rational way out of it.

    ReplyDelete
  36. I think the main problems with understanding the origins of Indo-European tribes are these:

    1- Indo-Europeans' genetic structures reveal patterns displaying ~multiple origins, chronologically rooted. ie, ~Some of the more ancient components of their genomes, particularly some of their mtDNA, may indeed have originated in the Caucasus or Anatolia -- but certainly not all of it. A relatively large percentage of their yDNA is Asian or African. There also seem to be certain genetic markers entirely peculiar to Indo-Europeans. It is problematic, especially in view of the fact that Genetics still hasn't adequately answered certain basic questions, such as whether or not the 'Out of Africa' theory can actually be proven correct.

    2- Their multiple origins is reflected in their broad, complex range of languages and dialects. I do believe however, that knowledge about the histories of those languages should provide important clues to the puzzle.

    3- It is further reflected in their social and political histories, much of which has been documented fairly well. Somehow the three disciplines need to be integrated, in order to produce a more complete picture.

    (Languages, genetics, political / social histories... And I'm sure there are other disciplines, like archeology and so forth, which might contribute to solving the problem. The point being that no single aspect of the subject can adequately shed enough light on it to present a clear enough picture.)

    ReplyDelete
  37. @Unknown

    Quote- “strong allegiance that the historical linguistics community has to the whole “Steppes” theory”

    That's not correct, for two reasons. First, many mainstream linguists have worked and still work with other PIE Urheimat models (Balkans, Bactria, Baltic, Armenia) – although, after 200 years of research, the Steppes model does seem the most likely one in light of the accumulated evidence. And the “new” evidence put forth by this paper is just not convincing enough to change that.

    A more important point, however, is that the main disagreement here is NOT about specific conclusions of the model, but about its methodology. So much so that serious linguists who reject the Steppes model never get the same level of backlash.

    I agree with eurologist n one thing: what we are witnessing is effectively a fight about the future of the field – issues such as methods and interdisciplinarity – and not about PIE, so your assessment that this somewhat parochial issue is what actually matters is inaccurate, I think.

    Historical linguistics as we know it will survive just fine without a Steppes model for PIE.

    On the other hand, a move toward more quantitative methods would be quite significant.

    Though, TBH, this is already happening anyway, only gradually, so the real issue here is that Atkinson et al. have gone a bit too far and thrown the baby out with the bathwater.


    Quote- “To my knowledge, no historical linguist has ever even attempted to establish the rate of sound changes...”

    It's not so much the rate of sound changes that really matters in this case, but rather the rate of lexical replacement. And, yes, attempts at determining that have been going on since the 1950s, and the misgivings of linguists today derive precisely from the failure of those models.

    ReplyDelete
  38. Technically, “Indo-Europeans” are anyone who speaks an Indo-European language.

    The notion that there was a “tribe” of Indo-Europeans or more properly Proto-Indo-Europeans comes only from reconstruction of a hypothetical language by linguists using the comparative method. This reconstructed language comes from comparing words in the various Indo-European languages and assuming sound changes based on “systematic correspondence” that go back to an original parent language.

    No matter how accurate that reconstruction is, it can tell us very little about who spoke the language and where.

    Some scholars have used the comparative method to conjecture about the religious beliefs of these hypothetical “Proto-Indo-Europeans.” But just as Christianity crossed borders and languages, there is really no way to know if specific religious or cultural matters were somehow attached to these early speakers or were part of religious movements that spread on their own before or after that language was spoken.

    The method of reconstruction means that Proto-Indo-European, the reconstructed language, is treated as an isolate. There are no postulated sister languages to PIE. There is no English to PIE’s Dutch. There is no Portuguese to PIE’s Spanish.

    But we do know that Proto-Indo-European had to come from somewhere. It was not invented or fall from the sky full-blown. So there is no context because whatever sister languages or mother languages existed have apparently disappeared beyond trace.

    So though the language of these Proto-Indo-European speakers may have been spoken in many different contexts. If the language began to spread at some axis of roads, waterways, trade routes -- at some depot of horse ranchers and cattle traders and wagoneers and warriors for hire -- some town out of the American West or the Far East -- then maybe the Proto-Indo-European language was spoken by many different types of people, different genetics, different belief -- BEFORE it spread.

    The fact is that the model of steppes horsemen has never really resulted in a permanent spread of a dominant language. Mongolian did not spread due to the raids of the Mongols. What the Mongols did do is contribute to the spread of Chinese and Arabic.
The same for the Huns. If we include the Maygars and the Turks, they created only pockets today surrounded by unrelated languages.

    The success story in eastern Europe is the Slavic languages. Given that most historians don’t even identify these speakers before 500 AD, the enigma of the spread of those languages might shed more light on the spread of Indo-European languages than any steppes horsemen.

    The same might be said of the spread of American English to schools in China, Japan, Russia and Finland.

    For all we know, the people who originally spoke Proto-Indo-European died out long ago, leaving no genetic trace.

    A diversity of genetics and cultures might have been the main characteristic of Indo-European as soon as it began to spread.

    And that might have been its real strength.

    ReplyDelete
  39. @n/a
    Since you're being wo brash atm, tell me what you think of me and my contributions?

    ReplyDelete
  40. Unknown,
    "How long, for example, did it take for the hypothetical Proto-European language to turn into German? If you go by historical rates of sound changes in written German (say, medieval to modern), what mathematical rate of change do you get? And how does that apply to the gap between PIE and historical German?"

    You have chosen the standard example that discredts slavish trust in this mathematicla model. Englsih and German experienced sound changes at diffenrnet rates, and Icelandi c exprienced almost noe, in the last 1,000 years.

    "Part of the definition of “cognate” words is that they have the same meaning."

    This is simplistic to the point of falsehood.

    You appear to be criticizing a discipline oyu do not understand. Are you a physicist by chance? They have a reputation in linguistics for doing this all the time.

    ReplyDelete
  41. Dienekes said-
    ''I suspect the former will be the case, but it will nonetheless be interesting to see how the different parties coming from archaeology and linguistics will react to the (archaeo)genetic avalanche that will doubtlessly provide us with new information about the prehistoric past.''
    Dienekes and all the other folks let me tell you that unless a comprehensive aDNA map of the dark ages (around 2000b.c) emerges including the areas of South-Central Asia to the Steppes and from Anatolia to the south central Europe with central asia, the riddle will not be solved!.
    stay well and happy 2013.

    ReplyDelete
  42. """"""Slumbery said... The Erzya population can be quite old there, but we do not know when they got R1a1a and even if they had it (quite possibly) from very old times, they did not conquer Europe, IE did. """"""""
    R1a1a in Europe conquered only north eastern part, Huns(Hungars) conquered this part. And Hungarian language is there.
    Ptolemy 2nd century. Between the Basternae(Germanic) and the Rhoxolani(Azov sea) are the Chuni(Huns)
    http://penelope.uchicago.edu/Thayer/E/Gazetteer/Periods/Roman/_Texts/Ptolemy/3/5*.html

    Huns
    http://upload.wikimedia.org/wikipedia/commons/4/43/Huns_empire.png

    R1a1a map+ Huns and White Huns(White Chions, Hephthalites)
    http://s016.radikal.ru/i335/1104/49/4a87a0e0d2a3.jpg

    In Asia
    White Huns(White Chions, Hephthalites)
    http://s48.radikal.ru/i121/1104/b8/0ad90ef0d935.jpg

    + Chionites(Xionites, Huna) + Red Chions(Red Huns,Kidarites) + Black Chions(Black Huns) + Celestial Chions(Celestial Huns) + Ouarkhonitai(Avars)+ Alchon
    + Xiongnu
    Title: A Western Eurasian Male(R1a1) Is Found in 2000-Year-Old Elite Xiongnu Cemetery in Northeast Mongolia. Author(s): Kim K, Brenner CH, Mair VH, et al.
    Source: AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY Volume: 142 Issue: 3 Pages: 429-440 Published: JUL 2010

    ReplyDelete
  43. Scripts, words, grammar and Genetics. where to draw the line for Indo European question.

    Drawing Inferences based on speech and words not applicable to other.

    Hope somebody does similar research for all Language isolates and non Indo European languages in the same regions and it provides some clue.

    How much Indo European is Greek?. really?

    IN case of Persian they picked Semitic script to Avestan script. Is it regression or progress? Neither. People /societies just go with the flow. Does it complies with Authors model?. I doubt it. Just applying the authors model to all Iranian dialects will make it fall flat. These models cherry pick the popular concepts, root words etc.

    That is like saying Gangnam popularity will make society Korean,

    Indo European root word spread also depends on centuries of commercial and cultural links and popularity. They may not answer all genetic questions..

    ReplyDelete
  44. "That is a gross understatement of the achievements and accuracy of Bayesian statistical computations in genetics."

    Demonstrating again you have no idea how Bayesian or other statistical methods actually work. Obviously, the accuracy and precision attainable with such methods will depend on the nature and amount of data one has to work with, the question one wants to answer, and the details of the analysis actually performed.

    It's not a slight against "the achievements and accuracy of Bayesian statistical computations in genetics" but merely statement of a basic fact that goes without saying to anyone passingly conversant with these tools to point out that even when applied to genetic data, you will rarely get or expect "the" correct answer. You will get a more or less useful approximation of what actually happened.

    My contention is that Gray and Atkinson's results are far to the less useful side. The estimates of error they report (internal to their model and choice of data) don't begin to capture the numerous other likely and potential sources of error, and the true level of uncertainty renders their model useless for discriminating between the Steppe hypothesis and the 1987 Colin Renfrew fantasy model.

    One of the few ways they could begin to prove otherwise is to show that their model performs well where there is actual ground truth in the form of direct historical evidence. It doesn't, so we're going to need more than your say-so that the results magically get more accurate the less subject they are to confirmation. Repeating the word "trivial" is not an argument that will make this issue go away.

    If they were serious, I'd also expect them to put together independent data sets for the same languages (other types of word lists, etc.) and compare the resulting trees. I'm confident the results of such an exercise would remove any doubt the trees they're building are dramatically less accurate and precise than their internal estimates of error/certainty would indicate. Again, to the point of uselessness in distinguishing between IE origins 6000 years ago vs. 9000 years ago. They could also, for example, attempt to show their geographic results are robust to locating the Indo-Iranian homeland correctly (on the steppe), but I don't see them doing this.


    "They make no claim in the paper that they are differentiating between different IE origin models"

    Could have fooled me: "test these hypotheses. We found decisive support for an Anatolian origin over a steppe origin."


    "They even state that Kurgan expansion may have played a role."

    A passing acknowledgment that the Anatolian Neolithic model doesn't actually make sense doesn't make up for the other 95% of the paper, or the abstract, or the press coverage. Renfrew similarly later attempted to incorporate steppe dispersals as an important part of his model. This was not out of charity to historical linguists. It's because the Anatolian Neolithic model has never worked.


    "The accepted plural for the newly-evolved English noun is ad hominems"

    I was mocking you, dude -- not commenting on your grammar. You whine about "ad hominem" when I briefly let it be known I have no respect for you then go on in the same thread to fill entire posts with odd/inept speculation about my mental state. It was a throwaway comment making fun of your hypocrisy with respect to your pseudo-rational wannabe science nerd-ism. To be clear: none of this actually matters and your posts display about the same level of reasoning with or without name calling.


    nuadha,

    I don't have a problem with you. You seem to be interested in learning, not playing pretend senior internet scientist dress-up. I'll reply when I get around to it. Trolling won't speed that up.

    ReplyDelete
  45. Western Nostratic ("Indo-Kartvelian", IE + Afrasiatic + Kartvelian)

    Yakhontov comparisons of proto-languages
    http://ifotki.info/11/f1e140363b05267821edf133dc0af7dd4f6726131418339.jpg.html


    I compiled reverse table of Yakhontov comparisons (PHYLIP data)

    7
    modal 0 50 50 50 50 50 50
    Altayan 50 0 83 69 72 66 69
    Dravidian 50 83 0 89 89 89 83
    IndoEurope 50 69 89 0 69 58 75
    Kartvelian 50 72 89 69 0 52 80
    SemitHamit 50 66 89 58 52 0 75
    Uralian 50 69 83 75 80 75 0

    The result
    http://ifotki.info/11/5a641171e9f1c5b8d4bb9294a2675e094f6726131420034.jpg.html

    Polaco graphics
    http://s54.radikal.ru/i146/1212/3b/e3fa8e9991d5.png
    Some IE people are close to Kartvels and to Afrasiats, I puted them all in a red circle.

    ReplyDelete
  46. Mr Dienekes,
    Please change in my previous post, the
    http://ifotki.info/11/f1e140363b05267821edf133dc0af7dd4f6726131418339.jpg.html

    to
    http://f11.ifotki.info/org/f1e140363b05267821edf133dc0af7dd4f6726131418339.jpg

    and change the http://ifotki.info/11/5a641171e9f1c5b8d4bb9294a2675e094f6726131420034.jpg.html

    to http://f11.ifotki.info/org/5a641171e9f1c5b8d4bb9294a2675e094f6726131420034.jpg

    ReplyDelete
  47. n/a,

    I still have not seen any arguments at all with intellectual merit from you. Statements like:

    you have no idea...

    I was mocking you, dude...

    I have no respect for you...

    It was a throwaway comment making fun of your hypocrisy...


    ...in addition to those I listed above leave me and most anyone else with only one plausible impression: you are not actually here to contribute to the discussion -- but you are here due to a mind set or condition that is better dealt with elsewhere.

    ReplyDelete
  48. I find it disturbing that none of the discussion here refers to the role of advection,or the referenced Davison et al. paper (http://www.mas.ncl.ac.uk/~n0072132/NeolithicWaterways.pdf). Although Davison et al. study the role of the Danube and Rhine in the spread of agriculture, a similar role might well apply to PIE origins for the Pontic rivers and the coastlines of the Black and Caspian seas. Depending on detailed assumptions, NW Anatolia and much of the Pontic might have been "diffusionally close", as close as or closer than the Aegean and NW Anatolia.

    I'll also point out that the assumption that PIE started in one location and spread out is totally unwarranted. It might well have actually originated as a "lingua franca" spoken along a distributed network of trade routes, both land and water-borne.

    IIRC some scholars have found traces of an early "isolating" substratum, which could have derived from a creole used by speakers of a number of different, variously related, languages associated with such a distributed trade network.

    I'm not proposing any specific alternative here, just trying to show that the current scenarios are far too limited, based on unwarranted "simplifying" assumptions.

    ReplyDelete
  49. Depending on detailed assumptions, NW Anatolia and much of the Pontic might have been "diffusionally close", as close as or closer than the Aegean and NW Anatolia.

    AK,

    I totally agree - which is why I have hypothesized both above and before about a trans-Pontic Sprachbund - which could resolve a number of problems; among them, the long hiatus before some of the branching, and the anachronistic closeness of Proto-Iranian to PIE.

    ReplyDelete
  50. @ Eurologist:

    Except for you, I've never read anything hinting to a particular closeness of proto-iranian with PIE (it seems to go against what I have read actually).
    Can you provide a source for this (if possible something readable on the internet).

    ReplyDelete
  51. What I consider a shame is that they based most if not all of their arguments on possibilities and their own believes and assumpatations, listened and waited in vain for any evidences they would provide in support for their hypothesis and all I got was arrogant appearing talks of two linguists and their opinions.

    ReplyDelete
  52. In most areas of science, this would be regarded as a rogue and extremely desperate talk. It is acceptable that you spend a few minutes criticizing opponents, but largely, you need to provide your own evidence and research.


    You simply cannot criticize a new, rapidly-evolving and improving model just based on its trivial, known shortcomings. Such a thing is ludicrous and paints a truly bad picture of the talk presenters.



    Exactly this was also my observation. They might be true in some points but I desperately waited for any evidences they would provide us but all I saw was critics based on their assumptions

    ReplyDelete

Stay on topic. Be polite. Use facts and arguments. Be Brief. Do not post back to back comments in the same thread, unless you absolutely have to. Don't quote excessively. Google before you ask.