There are more things in prehistory than are dreamt of in our urheimat

By Razib Khan | August 23, 2012 11:59 pm

A new paper in Science claims to have ascertained the locus of origin of the Indo-Europeans, Mapping the Origins and Expansion of the Indo-European Language Family. These are bold claims, and naturally have triggered a firestorm. No surprise, the same happened with these researchers when they published the result in 2003 that Proto-Indo-European flourished ~9,000 years ago, in alignment with an “Anatolian hypothesis,” as opposed to a “Steppe/Kurgan hypothesis.” The original paper in 2003 utilized phylogenetic methods which are common within biology, and applied them to linguistics. This second paper now incorporates spatial information into their model, to generate an explicit locus of origination, in addition to the dates for the bifurcations of the node.

In relation to results I think that the figure to the left is the most important, because it gives us their inferred dates of separation between various Indo-European language families. Observe that Italic and Celtic did not diverge in prehistory, but in history (i.e., the Sumerians and Egyptians were flourishing at the time). Additionally, the diversification pattern is not a simple “rake,” there is internal structure. They may date the origin of Indo-European languages to the early Holocene, but the diversification seems to have happened in steps and pulses. Though the authors support the Anatolian hypothesis, they also  seem quite comfortable acknowledging that the real story is more complex, though you wouldn’t get that from the media.


But speaking of complexity, who really knows what’s going on in this paper? I have a handle on the general framework, but haven’t used all the algorithms. As I indicate below in population genetics a good intuition on the kinks and tendencies of clustering algorithms can be obtained only through usage. And of course few people will read the supplements. For example, in Nick Wades’ piece in The New York Times David Anthony, author of the magisterial The Horse, the Wheel, and Language: How Bronze-Age Riders from the Eurasian Steppes Shaped the Modern World , makes a criticism which is addressed within the paper (in the supplements):

Dr. Anthony, noting that neither he nor Dr. Atkinson is a linguist, said that cognates were only one ingredient for reconstructing language trees, and that grammar and sound changes should also be used. Dr. Atkinson’s reconstruction is “a one-legged stool, so it’s not surprising that the tree it produces contains language groupings that would not survive if you included morphology and sound changes,” Dr. Anthony said.

Dr. Atkinson responded that he did indeed run his computer simulation on a grammar-based tree constructed by Don Ringe, an expert on Indo-European at the University of Pennsylvania, but that the resulting origin was, again, Anatolia, not the Pontic steppe.

There’s an asymmetry here. The historical linguists have compelling and transparent rationales to make for why the Steppe thesis should be preferred over the Anatolian one. Lay persons can make assessments about historical linguistic models which are based on common sense such as words which span all Indo-European languages, and might give clues to the geographical and temporal point of origin. In response, you have Bayesian phylogenetics. At some point in the future I suspect all of this research will make recourse to Bayesian phylogenetics, but at this stage of the game even most people who use Bayestian phylogenetic packages don’t really understand how they work.

I may not grok the methods in detail, but I do appreciate that the authors simulated data to test their methods, and, that their methods worked for cases where we know the answer. For example, the method correctly inferred the geographical origin of the Romance languages, and their time of diversification. But in this situation we know the answer. How about in cases where we don’t?

I noticed this strange plot in the supplements. I’ve highlighted Romani, the language of the Roma. The fact that Romani is an outgroup to Indo-Aryan langauges, illustrates some deep problem with their method. Romani did not start diverging from other Indo-Aryan languages 3-3,500 years ago. It started diverging 1-1,500 years one. We know this because that’s when the Roma start showing up in the Islamic world and parts of southeast Europe. It may be that it just happens to be that the most diverged Indo-Aryan language also happened to be the one which migrated out of India, but I don’t think that’s the case. Rather, the non-Indo-Aryan influences on Romani must be impacting its affinity to other Indo-Aryan languages, even if they are core words.

With that skepticism entered into the record, I can broadly credit the possibility proposed here in the most general sense. We know from genetic clustering algorithms that Indo-European populations within Europe seem enriched for a “West Asian” element vis-a-vis their non-Indo-European neighbors. I’m talking here mostly about the Basque and Finns, though arguably the Sardinians were Indo-Europeanized only during the Roman era, and they should count as well. But, I’m pretty sure that the Indo-Aryans are the ones who brought the “European” component found in low levels across northwest South Asia to the subcontinent. The Indo-Iranians diverged from the European Indo-Europeans ~4,000 BC, and I’m suspecting this may have happened along the broad trans-Caucasian and Russian fringe. This is where contact was made was Uralic peoples. The authors of the paper themselves point to the viability of the Kurgan hypothesis in this modified form in the text. I don’t see why the archaeologist are all worked though (unlike the historical linguists).

CATEGORIZED UNDER: Anthroplogy
MORE ABOUT: Indo-European
  • Paul Ó Duḃṫaiġ

    Well people have been talking about a common Proto Italo-Celtic grouping for a long time. It is interesting when you look at maps of R1b distribution in Europe that the Italic and formerly Celtic speaking parts of Europe are dominated by P312 and it’s subclades (with distinct geographic spread of each clades). Whereas P312 brother (under L11) U106 appears dominant in areas of Germanic language dominance.

    There were some maps published 2 years ago or so as part of the report on M269 in Europe (Busby?)
    U106/S21
    http://compsoc.nuigalway.ie/~dubhthach/U106-S21-poE-CT.png

    U152/S28 (under P312)
    http://compsoc.nuigalway.ie/~dubhthach/U152-S28-poE-CT.png

    L21/S145 (under P312)
    http://compsoc.nuigalway.ie/~dubhthach/L21-S145-poE-CT.png

    It doesn’t really help that they use different scales in the third map compare to the other two, I know there’s a better L21 map available from Eupedia:
    http://compsoc.nuigalway.ie/~dubhthach/Haplogroup-R1b-L21.gif

    The following tree from August last year gives a good idea of the then known structure of L11 in Europe. Though there’s been some big discoveries even since then:
    http://compsoc.nuigalway.ie/~dubhthach/R1b-deepClade.png

    The ISOGG tree is considerably more up to date.
    http://www.isogg.org/tree/ISOGG_HapgrpR.html

    Some are thus saying that R1b-P312 is a signature of those speakers of “Western Proto-IE” who would subsequently would give rise to both Proto-Itaic and Proto-Celtic.

    Anyways as someone who can speak both a Celtic (Irish) and a Germanic (English) language I often see cognates that are more obvious between Irish and say Romance languages, obviously it doesn’t help that 50% of english words are derive from latin/romance languages. Still some obvious connections such as:

    Rí vs. Rex (King) — in related Gaulish this was Rix (old Irish: Rig)
    Dia vs. Deus(God)

    Anyways hopefully we’ll start seeing some interesting results from Ancient-DNA over next couple years, I think that will make things more interesting. I know the recent ancient-DNA sample conencted to Bell Beaker was R1b-M269 (two steps above L11) but U106-, I don’t believe they tested it for L11 or P312.

    -Paul

  • http://www.genomesunzipped.org J Pickrell

    One thing I can’t figure out from a quick reading: does does model try to infer the location of the modern population most closely related to the root, or is there an attempt to account for population movements/extinctions and thus infer the location of the ancestral root population. Seems like it must be the former, since the latter is really hard. Not sure if that matters for their conclusions though.

  • Rimon

    I read the NY Times article and was irritated by how stupid most of the comments were. too bad you can’t ban them there, Razib :)

    I thought it was odd that in the article Dr Anthony thought it was so implausible that Tocharian could have an Anatolian urheimat. Obviously, they could have just gone north from Anatolia and then headed east.

  • Nirjhar

    Naah! It was Mehrgarh guys.

  • Charles Nydorf

    As an anecdote to models like this that ignore or grossly underestimate the effect language contact on divergence, I would recommend a very old essay by N. S. Trubetskoy “Thoughts on the Indo-European Problem.”

  • Karl Zimmerman

    While I, as a laymen with an interest in linguistics, have always been more partial to the Anatolian hypothesis, I’m not sure how to square it with the knowledge that there appears to have been an agricultural “first wave” which preceded the Indo-European expansion. The decedents of two different groups of Near Eastern farmers would be roughly technologically matched, and certainly have similar population densities. There would have been no reason for the first wavers to be culturally swamped to the degree they have been without some military advantage on the part of Indo-Europeans.

    At the same time, I’ve long been suspicious of the idea of a loosely-bound steppe nation conquering Europe in the Bronze Age, because the local geography is quite different from Central Asia and India. Most of Europe would have been dense primeval forest except for wherever the “first farmers” cleared a significant amount of land. Individual horses might be able to navigate through dense forest, but I don’t think chariots would have had much luck.

  • http://www.textonthebeach.com Seth

    @4 You took the words right out of my keyboard. I haven’t looked at this paper in detail yet, but it appears that it severely underestimates how drastically language(s) can change in linguistic contact zones.

    One needn’t look further than English to see that language family trees are too linear to capture reality. Beowulf is truly Germanic, written around 1000 AD, and it is unintelligible to modern speakers. The Canterbury Tales and Le Mort d’Arthur, written around 1400 AD, are completely intelligible. What happened in between? The Norman Invasion, of course! English was born in the mixing of a Germanic and a Romance language over the course of a few hundred years. English is essentially a German phonetic and inflection system spoken with Latin syntax and a Latin idiom. (This is why it’s much, much easier for Americans to learn Spanish grammar than German grammar but easier to learn German pronunciation.)

    There is not a direct line from English back to Old Germanic, and any family tree that shows such a line is not just an incomplete model; it’s absolutely false. So, I’m generally skeptical of any evolutionary models that don’t show criss-crossing branches.

  • http://dienekes.blogspot.com Dienekes

    @Joe Pickrell,

    the authors considered both a set of extant languages (which excludes the now extinct Anatolian languages) and a set of 20 ancient languages, as well as the set of all languages. They all concluded the same thing.

    @Seth

    the network model has also been considered before, and apparently there is work on it specifically for IE by Paul Heggarty:

    http://dienekes.blogspot.com/2011/11/splits-or-waves-trees-or-webs.html

  • Nirjhar

    The ages of the languages aren’t scientific and words in truth don’t follow the rules of physics neither the rules set by ‘language-scientists’ specially in the case of unrecorded history.
    Only direct tokens from a time gives a true clue eg. aDNA, Scripts, tools etc.

  • Dwight E. Howell

    I’m unaware of any historic documents that suggest any sort of cultural or linguistic unity between the Latin speaking tribes and the Celtics. By the time these people appear in historic documents they were pretty distinct.

    In reference to another poster. A light two wheel cart/chariot pulled by two small horses doesn’t even need a path. Most trees in most natural old growth woodland are going to be far enough apart they get get through. If they aren’t go around. Villages would have had networks of paths connecting them. There would have been open land around the villages.

    While not suitable for use in combat in a forest the chariots known from the region could have been used in the forest tracks and around the villages. Rough broken hilly and mountainous conditions could be used judiciously to negate the advantages offered by chariots.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    Nirjhar, stop asserting concisely or i’ll ban you. make an argument or shut up.

  • Rahul Jain

    I’ve looked at the words in Gaulish (a Celtic language) that we know and they are shockingly similar to Latin. These are taken from historic artifacts, mostly, so ok, maybe no “documents”.

  • pconroy

    @10, Dwight,

    Have you ever heard of Celto-Ligurian? The Ligures were a people who lived in today’s NW Italy and SE France, and spoke a language intermediate between Italic and Celtic. They were finally subdued by the Romans in the 2nd century BC. Their territory formed parts of the provinces of Gallia Cisalpina and Gallia Transalpina later on.

    http://en.wikipedia.org/wiki/Ligures

  • Karl Zimmerman

    10 –

    I understand that old-growth forest isn’t impassible on horseback, but the speed advantages of cavalry over infantry are considerably lower if there isn’t a set path. In addition, given archery was widespread even before neolithic times, ambushing chariots in a forested environment would have been comparably easy (although not as easy as in rough terrain).

    Again, I’m not saying that this would be enough to negate any advantage for chariot-bound warriors, but it should have evened the score for long enough. Native Americans in the Great Plains figured out mounted combat within only a few centuries, so a slow enough incursion would have resulted in a likely stalemate once both sides reached technological parity.

  • pconroy

    @1, Paul,

    I’m no linguist, but there are a lot of English words that supposedly originate from the Norman French, but could also have arisen from the Celtic speaking peasant Britons, or the Latin speaking elite Britons, or indeed from the Hiberno-Latin – Latin larded with Gaelic, Greek and Hebrew words – written by the Irish monks, who brought enlightenment to the newly acquired “English” lands of the Germanic peoples (Angles, Saxons and Jutes).

    http://en.wikipedia.org/wiki/Hiberno-Latin

    Note: James Joyce’s “Finnegans Wake” was inspired by the Hiberno-Latin written by Irish monks.

  • Grey

    If the Hattities etc were non-IE then i don’t think Anatolia itself makes much sense as the IE urheimat but a region adjacent to Anatolia* which was also relatively unsuitable for the agriculture developed in Anatolia – thereby preventing expansion from the Anatolian farmers into that region – seems very likely especially if those mountain(?) IE then picked up pastoralism from the Anatolian farmers which led to them expanding in various directions – initially with their flocks – and maybe at some later time in a different place after their draft oxen had been bred into milk and meat producing cattle, with their herds?

    (*or an internal sub-region of Anatolia but a position north/east of Anatolia would give convenient access for Tocharian and steppe expansions.)

  • RKM

    There are two historic examples of massive linguistic shift perpetuated by Steppe folk conquering and culturally dominating densely populated agricultural regions: Magyar in Hungary and Turkish in Anatolia. While each of these linguistic/cultural shifts had its contingent proximal causes, together they do at least suggest the possibility of an ‘elite dominance’ mode for the spread of IE languages in Europe from the Pontic Steppe. BTW, there is no significant Uralic or Central Asian genetic signature among the present-day populations of Hungary or Turkey, respectively, though we know that Magyar and Turkish came from the East in the relatively recent historic period.

  • http://lyingeyes.blogspot.com ziel

    #15 – the only Irish words I know – cead, mile, failte, feis, and slainte – all have obvious Latin counterparts. Do you think these have shared origins or were they borrowed from Latin? Since they’re common concepts, my guess would be the former.

  • Dm

    We are getting used to the idea that genetic phylogeny trees aren’t in reality quite shaped as trees, because of the admixtures early and late. Admixtures from dissimilar populations make one overestimate TMRCAs in simple tree models.

    So I wonder how much admixing between languages could have led to overestimation of cognate replacement rate, and hence TMRCA in the study?

    Anecdotally, modern English is heavily admixed (but within IE); proto-Uralic is described as a recipient of a more ancient substantial IE admixture.

  • pconroy

    @18, ziel,

    Shared origin most likely.

    An example of a word borrowed from Latin into Gaelic (most likely) would be:
    Long (pronounced “lung”) = ship

    In Latin:
    Navis Longa = Ship (literally Long Boat)

  • RKM

    @19, Dm

    Proto-Uralic has layers of borrowings, earlier ones from PIE and later ones from IIr; this has always supported a Steppe PIE/IIr homeland over an Anatolian one.

    Interestingly, Atkinson et al explicitly state in their paper that the inclusion of the three extinct Anatolian languages may have skewed their results in favor of an Anatolian homeland. In fact, when these languages were excluded and only contemporary languages were included in the analysis, the strength of support (i.e. value of the Bayes factor) for an Anatolian homeland vis-a-vis a Steppe homeland diminished significantly (Table 1).

    Also, the effect of admixture on the estimation of the rate of cognate replacement is clearly seen in the case of Romani, as pointed out by Razib above. Romani is an Indo-Aryan language that spread from India with the Roma in the early medieval priod (~1000 AD). However, it has been influenced heavily by the Balkan languages and Greek. Because of this, in the analysis of Atkinson et al, Romani is erroneously shown to separate from the other Indo-Aryan languages 3500 YBP (Supplemental Figure 1). Clearly, admixture and language contact can influence the estimated rate of cognate replacement. This ‘sprachbund’ effect on vocabulary can seriously confound the methodology used in this paper.

  • pconroy

    It’s interesting how many other reconstructions show some of the earliest branches to be:
    1. Anatolian
    2. Tocharian
    3. Celtic

    In this paper Celtic is seen as one of the younger clusters?!

    I note that Albanian is maybe thee youngest cluster, and Thracian doesn’t make a showing at all?!

  • Patrick Wyman

    Yeah, I’m not buying this study. If you have a model whose results don’t fit known facts (i.e. the Romani language, as Razib points out) then there’s something wrong with the model.

    As far as Celtic and the Italic languages are concerned, most Celtic linguists (as of three years ago, when I was working on the topic) don’t accept the idea of a common Italo-Celtic branch. This used to be a common view, but my understanding is that the evidence is pretty strongly against common origin and points more toward borrowings in both directions.

  • Karl Zimmerman

    17 –

    The problem is, both of these scenarios involve language replacement in areas which are conducive to horse-based pastoralism. The Hungarian basin is a small, isolated grassland cut off from the nearest other grasslands (in Ukraine) by mountains and forest. And central Anatolia is semi-arid steppe. Indeed, the more heavily forested areas of Asia Minor, like the Pontic region, didn’t really go over to Turkish culture until the Greek-Turkish population exchanges in the early 20th century.

    Obviously if nomadic horsemen settle in a land where their way of life can remain unchanged for many generations, it becomes more likely that they will sway the majority. But if they’re in a land which is better suited to the farmer’s way of life than their own, it makes it more likely they will acculturate instead, as happened with the Bulgars and Alans in the Balkans, and the Mongols in Russia and China.

  • http://dispatchesfromturtleisland.blogspot.com ohwilleke

    @2 “does model try to infer the location of the modern population most closely related to the root, or is there an attempt to account for population movements/extinctions and thus infer the location of the ancestral root population.”

    The do seem to be inappropriately looking at modern distributions when we know what older distributiosn were like. I’ve seen a podcast at Dienekes of a “time evolution” video map, for example, that like the map with the family tree on it, doesn’t acknowledge that Celtic languages were ever spoken in Gaul or Iberia. Yet despite a lack of historical Celtic writings from that era, Celtic material culture is distinctive enough that we can map where and when it arrived and be confident of its linguistic affiliation, but these models ignores that kind of data. The model also does nothing to acknowledge just how recent (ca. 1500 years before present +/- a few centuries) Slavic language expansion is relative to the other parts of the Balkan IE language origins. It seems to show an expansion of Indo-Iranian languages from Armenia to Pakistan, instead of in the other direction, which doesn’t fit what we know about how that happened. The pre-Tocharaian story is a bit fuzzy, but we have a pretty good idea when it showed up in the Tarim Basin (2000 BCE to about 600 CE) and some of the likely archaeological culture antecedents of the Tocharians as pointed out by Mallory, et al. are some of the strongest pointers towards IE at a Pontic Caspian steppe at about the Kurgan hypothesis time period.

    It is quite infuriating to see people do Bayesian statistics, which are supposed to be all about incorporating prior distributions based on what you already know, ignoring the wealth of IE structure and time depth information we already have from other reliable sources.

    @5 “models like this that ignore or grossly underestimate the effect language contact on divergence”

    I totally agree. One very plausible reason for Anatolian languages to look old is that they absorbed a lot of Anatolian non-IE substrate influences, while other IE languages may have shared substrate influences that were mutually more similar to each other than they were to Anatolian non-IE substrates.

    @16 “If the Hattities etc were non-IE then i don’t think Anatolia itself makes much sense”

    There are significant swaths of Anatolia and Western Iran that we know were non-IE ca. 2000-1200 BCE prior to being conquered by the Hittites, and the Mittani at roughly the Iran-Turkey border ono a modern map spoke a language derived form Indo-Aryan at least at the elite level, which would have been derived from migrants sourced in the general vicinity of Northern Pakistan and Northwest India ca. 1500 BCE and necessarily had to be a recent arrival since we can date likely early Indo-Aryan archaeological traces and early Rig Vedic references.

    The story of Hittite expansion from just a couple of city-states ca. 2000 BCE (immediately following a major regional drought period) to almost all of Anatolia and the Northern Levant to utter collapse ca. 1200 BCE is pretty well documented. Early Hittite expansion is very close in time to early Indo-Aryan expansion. Some of the metalurgy technologies distinctive to both seem to be Caucusas sourced from what is usually believed to have been a non-IE society.

    Most IE Anatolian language documents derives from late-Hittite or post-Hittite collapse sources so they aren’t highly probative about the linguistic state of Anatolia pre-2000 BCE. Most pre-Hittite Anatolian IE languages that are attested were in the far West of Anatolia and were part of a larger Aegean oriented IE society (think Trojans). Outside of this “contact zone” with the Aegean societies it isn’t at all clear to what extent IE in Anatolia say, as of 2500 BCE, was pervasive or was a few dotted IE colonies from elsewhere in a sea of substrate non-IE speakers. Hittite expansion began much further east in Anatolia and seemed to be over mostly non-Hittite conquered peoples.

    It is not at all clear how old Armenian really is and its presence at a crossroads and boundary area of many languages make pure linguistic estimates of its origin place and time just as questionable as it is for the Roma. Armenian is not atttested historically until quite late by historical linguistics standards, but absence of evidence is not necessarily evidence of absence especially in ancient history.

    We do know as a matter of historical fact that there were no IE languages in Mesopotamia, the Levant, or the mountains of Iran immediately to the east of Mesopotamia in the time frame of ca. 3500 BCE until the Hittite expansion and I am somewhat skeptical of the notion that a predominantly IE speaking Asia Minor could get by with no attested IE languages for 1500 years of written Fertile Crescent history prior to the Hittites. We also have some indications from the shared PIE rooted lexicon about what the IE urheimat must have looked like that aren’t a great fit for an Anatolian hypothesis.

    @17 “BTW, there is no significant Uralic or Central Asian genetic signature among the present-day populations of Hungary or Turkey, respectively, though we know that Magyar and Turkish came from the East in the relatively recent historic period.”

    In Hungary you are right. In Turkey, you are wrong. The level of Central Asian genetic admixture is on the order of 8% overall, and I suspect that this is probably higher in some populations (e.g. the Alevi) who have greater residual Central Asian cultural ties, and lower in others. At any rate, there is a clear genetic signature there and that signature follows a clear West to East cline across Central Asia all the way back to a NE Asian Turkic homeland. In contrast, the subsequent Mongolian Empire left very little in the way of either a linguistic signature outside its core, despite its once immense span, while it has left somewhat more of an genetic signature, particularly in the patrilines.

  • ryan

    #12 – I don’t know much Gaulish, but a quick look at the wiki page for the Coligny Calendar shows me little similarity. Sure, the months end in -os (similar to Latin, but also to common Greek word endings); and anagantios may have a present participle syllable in it. But not only are the names of months completely different, I’m just not seeing close relationships to Latin roots in their stems. For dumannios, they cite both Latin fumare and Sanskrit dhuman.

    The declension of some cases of noun seems similar. But word order is apparently subject verb object rather than the typical subject object verb of Latin. The ordinal numbers don’t seem particularly Latinate to me.

    I recognize virtually nothing in the curse tablet of L’Hospitalet du Larzac.

    My Latin is rusty, and I’m not a linguist, so it’s very possible I’m missing something. I’m wondering if you can offer some examples for why Gaulish seemed “shockingly similar” to Latin?

  • http://dispatchesfromturtleisland.blogspot.com ohwilleke

    I sum up the evidence related to pre-IE linguistics in the Aegean and Anatolia, as a number of off topic points, in a lengthy and fairly heavily sourced blog post of June 22, 2011.

    News since then, like the discovery of a newly discovered ancient language isolate that existed in the Zargos Mountains until the 6th century BCE probably near the northern boundary of the non-IE Elamite linguistic area (ca. 2800 BCE to 550 BCE), generally supports the analysis and suggests that prior estimates of the time of arrival of the Persian language in the area may have been something on the order of 900 years too early.

    One big difference between the approach that I took in that post, and the one taken in the most recent computational linguistics paper, is that my analysis isn’t conducted solely from the direction of Indo-European linguistic evidence. It also looks for the affirmative evidence of non-Indo-European languages in the same region as well as evidence internal to Indo-European languages. The case for an Anatolian language family ca. 9,000 years ago looks much weaker when weighed against the positive evidence that there was a region from the Aegean to the Caucasus mountains and Zargos Mountains in which a number of non-Indo-European languages were spoken. When the question is “which of two language families were spoken in Anatolia ca. 3000 BCE?”, the weight of the evidence is different.

    Even if Indo-European languages had an Anatolian origin, this must have been a distinctively minority dialect within Anatolia until some time ca. 2500 BCE or later. For example, there is some evidence that at some point contemporaneous with the use of Minoan language Linear A script in Crete that the same script may have also been used by the Trojans on the Anatolian coast.

    Some of the comments on this publication at Language Log are also quite apt and point to a garbage in, garbage out problem with the calibration points used in the model. For example, it has Welsh and Irish separating at about 1000 BC when the archaeology tells us that the Celts were still in Austria and Bavaria.

    Similarly, it is hard to give much credit to a hypothesis of Indo-European linguistic continuity all of the way back to the early Neolithic when, for example, there is a huge difference between the Y-DNA found in ancient DNA from the first wave Neolithic LBK population in Europe and ancient Y-DNA of the later Corded Ware culture in essentially the same geographic area about 1400 years later.

  • Nirjhar

    @11. Sorry! I don’t talk much but i say truthfully with no edition.

  • Nirjhar

    @27 Look! We have no i mean no true evidence of The I.E. Language before 2000b.c. Which is Hittite and Hittite is also filled with quite a few non I.E. Words!! And as we know Hittite etc were not preserved orally like of Rigved So the real evidence ends with it, the I.E. Languages analysis of before 4000YBP is nothing more than stories, now people will say differentiation but i will say different type adaptions of a new language by different groups! the language which came from somewhere else.
    Good day.

  • Nathan

    If PIE arose in Anatolia then why didn’t IE come to dominate the region , even as it came to dominate areas as far apart as the Western fringe of Europe and most of South Asia?

    If PIE arose in Anatolia then how do the Anatolian hypothesis proponents explain the presense of Hattic, a non IE language ,in the Hittite substratum?

    The Mitanni empire covered parts of Anatolia yet the language of the empire was Hurrian, a non IE language even while the elites were significantly composed of Indo-Aryans.

    This points to IE being intrusive into Anatolia and assimilating and absorbing non IE speaking people.

  • Paul Ó Duḃṫaiġ

    @ziel, @pconroy

    Regarding “Cead míle Fáilte” well míle (thousand) is a loan word from latin, the other two come from proto-celtic

    Céad: From Old Irish cét, from Proto-Celtic *kanton, from Proto-Indo-European *ḱm̥tóm.

    míle
    iā, f. (< Lat. pl. mīlia) a thousand.

    Fáilte: From Proto-Celtic *vāletiā, from Proto-Indo-European *wel. Cognate with Old English wele, wela, willan (“to wish”) (English weal, wealth, and will); Latin velle (“to want”); etc.

    You can find list of Latin, Old Norse (Bád for example = boat) French and English loanwords in Irish here:
    https://docs.google.com/viewer?a=v&q=cache:AT9i15caUHAJ:www.gaeilge.ie/dynamic/file/Loanwords%2520in%2520the%2520Irish%2520language_full%2520text%2520TL.doc+&hl=en&gl=ie&pid=bl&srcid=ADGEESgLG_51VVz0ZviT53gAGse7sdhcIciw7gHJ4Z7qvCEiCN7U7-e8DSqGZgcQohgTCASXub5tF7GWjlYh-JeKi7KY4O0GUtzZQcyv67Y-qxe3C8O8vY_mfonHXiT8Tgj4C3xrCqgH&sig=AHIEtbTAlqko2FisrksASYPzZOIBJ2edZw&pli=1

    Interesting note on the word Cross in that document:


    Cross is an example of a Latin loanword borrowed into Old English via Old Norse from Old Irish. Interestingly, the original Latin word was later borrowed unchanged into English as crux, apparently from Medieval Latin crux interpretum (commentator’s torment, i.e., a difficult passage in a text).

    -Paul

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    If PIE arose in Anatolia then why didn’t IE come to dominate the region ,

    this is a very stupid point. can we get off it? we know “why.” the turks. the objection re: hatti and hurrians seems valid, but let’s think before we type. imagine you’d ask yourself: “if celtic languages came from the continent, why is the only celtic language on the european continent derived from a british variant?” again, we know why.

  • Nathan

    I thought it was obvious I was referring to antiquity , since we are talking about PIE homeland and I mentioned Hattic and Hurrian.

    Re. account of Turkic displacing IE in some parts of Eurasia and coming to dominate Anatolia ; that is my point, that we know of IE preceding Turkic in the region. Yet in Anatolia we don’t have record of Hittie or another IE preceding Hattic.

  • Nathan

    A poster by the name of ” goofy ” on the Language thread on this subject has pointed out how both Sri Lanka and Iceland show up as dark blue in the authors’ IE timeline spread; dark blue corresponding to 500 years bp. Indo-Aryan speakers reached Sri Lanka around 2500 bp.

    http://language.cs.auckland.ac.nz/

    http://languagelog.ldc.upenn.edu/nll/?p=4142

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    #33, is it known that hattian actually precedes nesili? (the hittites own name for their language). from what i know it’s actually just presumed. the hittite rulers conquered the hatti from kanesh. what you may have in the anatolian ‘homeland’ is a complex of very deep rooted language families. the analogy to genetics here is pretty straightforward…you stochastically sample one of those languages (the ancestor of indo-european), and it can population much of the rest of the world due to cultural advantages. but it may not have those advantages in the anatolian homeland.

    i don’t personally believe in this model, but it’s not crazy.

  • Grey

    @24 “The problem is, both of these scenarios involve language replacement in areas which are conducive to horse-based pastoralism”

    I think another possible example comes from the Arab conquests. If a relatively small elite-based expansion has an associated religion and their language is associated with that religion then that may increase the chances of the language taking over although that could only apply if the indo-european expansion had an associated religion.

    @30
    “If PIE arose in Anatolia then why didn’t IE come to dominate the region , even as it came to dominate areas as far apart as the Western fringe of Europe and most of South Asia?”

    Although i personally think PIE is more likely to be in the mountainous regions north and east of Anatolia than Anatolia proper, in the period between the Hittites and the Turks, Anatolia was a crossroads and fought over almost incessantly so it could just be a question of who got to be the ultimate (or most recent) winner.

  • random

    “If PIE arose in Anatolia then why didn’t IE come to dominate the region , even as it came to dominate areas as far apart as the Western fringe of Europe and most of South Asia?”

    -It’s happened with other languages, this is not a good generic objection.

    “If PIE arose in Anatolia then how do the Anatolian hypothesis proponents explain the presense of Hattic, a non IE language ,in the Hittite substratum?”

    -You’re presupposing that Hattic preceded Hittite in Anatolia, you aren’t mentioning why. If you want to argue that, you need to provide us with the relevant information if you have it.

    “The Mitanni empire covered parts of Anatolia yet the language of the empire was Hurrian, a non IE language even while the elites were significantly composed of Indo-Aryans.”

    -The eastern edges but what’s your point here?

    “account of Turkic displacing IE in some parts of Eurasia and coming to dominate Anatolia ; that is my point, that we know of IE preceding Turkic in the region. Yet in Anatolia we don’t have record of Hittie or another IE preceding Hattic.”

    -Are there records of Hattic that precede Hittite?

    “Sri Lanka and Iceland show up as dark blue in the authors’ IE timeline spread; dark blue corresponding to 500 years bp. Indo-Aryan speakers reached Sri Lanka around 2500 bp.”

    -You should read the note accompanying the map.

    Perhaps you don’t like this particular scenario for good reasons but the arguments you’ve presented aren’t particularly good – at least with no background information for some of them that desperately need it.

  • Onur

    in the period between the Hittites and the Turks, Anatolia was a crossroads and fought over almost incessantly so it could just be a question of who got to be the ultimate (or most recent) winner.

    It only seems to you so because of the usual salience of the periods of war compared to the periods of peace when looking at history. In reality, there were periods of peace in Anatolia during much of the period you mention.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    . If a relatively small elite-based expansion has an associated religion and their language is associated with that religion then that may increase the chances of the language taking over although that could only apply if the indo-european expansion had an associated religion.

    take a look at where the arabs spread their language: only in the afro-asiatic domains. in areas like the fertile crescent there were already a fair number of arabs, and the other languages were related.

  • Ken

    One of the assumptions of the phylogenetic clustering models is that once a species has split, the daughter species cannot mix genes any more. Another assumption is that any particular gene sequence arose exactly once (which is not always true, but is a high-probability assumption). Taken together, this means that if two species have some gene, they must have a common ancestor; and if they are the only two species with that sequence, their split happened after the common ancestor had split from the other species.

    I don’t see how these assumptions could work with languages, which can and do borrow freely. For example in the Romani case, the material it has borrowed would make it relatively dissimilar to the other Indo-Aryan languages, and the algorithms would thus position it as an outgroup to those languages. With enough such borrowing, it could even shift position in the tree as a whole, and group with(say) the Italic languages.

  • Grey

    @38 I exaggerated but i think the main point stands. If a terriotory is fought over then the ultimate cultural winner doesn’t have to have any connection to the original population.

    @39 Fair point.

    @37 “You’re presupposing that Hattic preceded Hittite in Anatolia, you aren’t mentioning why. If you want to argue that, you need to provide us with the relevant information if you have it.”

    I tend to look at things as a process of elimination which is the wrong way round, however…

    If farming started in Eastern Anatolia and then spread in the directions that were viable for farming at the time then either

    1) The original farmer population expanded into adjacent suitable regions

    or

    2) The idea of farming was transferred to adjacent populations in suitable regions

    *If* it was the first case then you’d expect peoples like the Sumerians to the south or other early farming groups to the west of Eastern Anatolia to show some cultural and/or genetic connection to the first farmers and from my understanding they don’t show any sign of being IE or PIE. So unless i’ve missed something i think that implies either it’s case (2) or the first farmers weren’t PIE.

    Case (1) seems more plausible to me on path of least resistance grounds so that would leave the first farmers not being PIE.

    However where the expansion of early farming out of eastern Anatolia was blocked in some way either by physical geography (mountains, marshes, deserts) or climate (too hot, too wet, too cold, too dry) and that blockage lasted for a long time then i think part of the farmer population would develop a purely pastoralist model along the edges of the blockage and that model might spread to the indigenous hunter-gatherers through their hiring as stockmen (if the blockage lasted long enough).

    http://resources3.news.com.au/images/2011/06/15/1226075/553655-180611-raparapa.jpg

    So the most plausible model to me is original farmers expanding where suitable from Eastern Anatolia – mostly to the west and south – sparking in their wake the creation of pastoralist societies all around the edges of the farming terriotory e.g. the Semitic tribes in the south and maybe the PIE tribes in the vicinity of the transcaucasus. If so, once pastoralized, that position would be very convenient for expanding around the black sea coast, to the steppe or to the Tarim basin – or over-running the non PIE farmers in Anatolia (while the Semitic pastoralist tribes did the same thing in the south).

    I think that works as a model – at least to my current level of knowledge.

    (I’m not saying it’s true just that it’s a plausible fit.)

  • pconroy

    @35, Razib

    ancestor of indo-european), and it can population much of the rest of the world due to cultural advantages. but it may not have those advantages in the anatolian homeland.

    As unlikely as this seems, IIRC, this is exactly what happened with Austronesian, as it only closely resembles one of the 5 language groups in Taiwan (aka Formosa) – the most southerly one – but seems to have spread from there to the Philippines and peninsular SE Asia and beyond… so I guess there is at least one instance of this rare event happening…

  • Onur

    @38 I exaggerated but i think the main point stands. If a terriotory is fought over then the ultimate cultural winner doesn’t have to have any connection to the original population.

    Actually the Turkic invaders did not leave much of a cultural impact in Anatolia other than language (even that is debatable, as the Turkic language that spread in Anatolia is not directly the continuation of the Turkic language of the Turkic invaders, but a form of Turkic that is greatly impacted by the Persian and Arabic of the Muslim ruling class of Anatolia and, especially at the folk level, by the local languages such as Greek and Armenian) and some products such as yoghurt, which spread not only in Anatolia but all through West Asia during the same time. Even the kind of Islam that spread in much of Anatolia is not the Turkic Islam but the Islam of the Hanafi Sunni Iranian religious elite that came to Anatolia with the Seljuq elite. Only in some small marginal areas of Anatolia (e.g., the relatively depopulated Byzantine-Seljuq border region) did the Turkic invaders leave a relatively strong cultural impact.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    Actually the Turkic invaders did not leave much of a cultural impact in Anatolia other than language (even that is debatable, as the Turkic language that spread in Anatolia is not directly the continuation of the Turkic language of the Turkic invaders, but a form of Turkic that is greatly impacted by the Persian and Arabic of the Muslim ruling class of Anatolia and, especially at the folk level, by the local languages such as Greek and Armenian)

    this sounds like bullshit to me. of course the substrate has an effect. but anyone who hears turkish sure as hell knows it sounds like a totally bizarre language. but that’s only because you are used to hearing semitic or indo-european languages (the same is true of finnish and hungarian; both shaped by the indo-european neighbors, but notably distinct even to the casual ear).

    Even the kind of Islam that spread in much of Anatolia is not the Turkic Islam but the Islam of the Hanafi Sunni Iranian religious elite that came to Anatolia with the Seljuq elite. Only in some small marginal areas of Anatolia (

    what is turkic islam that you speak of? be precise and specific in your response so i can evaluate it. don’t evade or expend excessive verbiage.

  • Nirjhar

    @42@44
    4500-800b.c. Guys can you pick what i’m trying to put?.
    Good day.

  • Onur

    #44,

    You accused me of literalism before, but you are now making the same literalism. If you just focus on my expression “even that is debatable” and take it out of the context, it may sound like bullshit to you. But the rest of the sentence quite clearly shows what I mean by “debatable”: “as the Turkic language that spread in Anatolia is not directly the continuation of the Turkic language of the Turkic invaders, but a form of Turkic that is greatly impacted by the Persian and Arabic of the Muslim ruling class of Anatolia and, especially at the folk level, by the local languages such as Greek and Armenian”. So it is clear from my word “directly” that the Turkic language that spread in Anatolia is some sort of a continuation of the Turkic language of the invaders but not so much due to the great impact from the non-Turkic languages that I mention (from top to down for Persian and Arabic and from bottom to top for Greek and Armenian). I hope I have now made myself clear.

    what is turkic islam that you speak of? be precise and specific in your response so i can evaluate it. don’t evade or expend excessive verbiage.

    The Turkic Islam I speak of is the kind of Islam that was prevalent among the Turkic invaders of Anatolia. It was quite different in many ways from the Islam of the Muslim establishment in Anatolia, which was established in Anatolia by the Hanafi Sunni Iranian religious elite that came with the Seljuqs. It is the Islam of the establishment that won the day in Anatolia.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    The Turkic Islam I speak of is the kind of Islam that was prevalent among the Turkic invaders of Anatolia.

    tell me about this islam in concrete terms. what made it distinctive from hanafi sunni islam.

  • Onur

    tell me about this islam in concrete terms. what made it distinctive from hanafi sunni islam.

    It was basically Tengriism with a Muslim clothing. It was neither Sunni nor Shiite Islam but a very different religion.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    #48, i know whereof you speak. many kirghiz and some kazakhs are somewhat in that stage now. i don’t think that’s a really credible claim for a distinctive turkic islam though. it’s just a transient stage, which is common in many newly converted populations. this is contrast with forms of chinese islam which arose during the ming dynasty which elucidated a form of the religion which was strongly colored by confucian philosophy. i.e., a syncretistism which was from on high, and not simply a function of incomplete assimilation toward muslim ideals. there are forms of javanese islam which also move in this direction, deviating from world normative islam through an interleaving of javanese cosmogony. to my knowledge the tengriism never reached that stage of elaboration.

    by analogy, one could speak of distinctive nordic christianity during the early years when the new religion had to accommodate with the old. but this was just a transient. in sharp contrast for various historical reasons irish christianity did develop its own distinctive cultural form which was NOT a transient, and the overturning of irish christianity for roman christianity was not an evolution toward orthodoxy, but a cultural rupture and rejection.

  • Onur

    #49,

    Your analogy is broken when we take into account the fact that the Seljuq Anatolia hosted for an appreciable period two very different and often clashing forms of Islam (the Islam of the establishment, and the Turkic Islam; they continued to clash as late as the mid-13th century, see the Babai revolt) and the fact that the Islam of the establishment was already the fully formed orthodox Hanafi Sunni Islam from the very beginning of the Seljuq rule in Anatolia, as it was imported from Persia via the Persian Hanafi Sunni religious elite that was brought to Anatolia by the Seljuq elite, who supported them from the beginning, so the Islam of the establishment had no influence from the Islam of the Turkic invaders. So, in Anatolia we do not see a transition from a Tengriist form of Islam to an orthodox form of Islam, rather, we see the concurrent and hostile (to the level of massacres) existence of the two from the very beginning and the ultimate victory of the orthodox Islam.

  • http://dispatchesfromturtleisland.blogspot.com ohwilleke

    “#33, is it known that hattian actually precedes nesili? (the hittites own name for their language). from what i know it’s actually just presumed. the hittite rulers conquered the hatti from kanesh. what you may have in the anatolian ‘homeland’ is a complex of very deep rooted language families.”

    The expansion of the Hittites from being a two city state petty kingdom to ruling almost all of Anatolia and the Northern Fertile Crescent to the fall of the empire is pretty well documented historically at every step, so we know on a pretty much province by province basis what parts of Anatolia were Hattic and then conquered by the Hittites. There are a few non-Hittite Anatolian communities that become client states, but these are a decided minority an appear only in pockets of the ultimate Hittite empire.

    Much of the surviving record consists of correspondence between different senior political leaders across Anatolia and in Egypt, and bragging accounts of their conquests and battles. The Hittites won their empire one little chunk at a time, not in one fell sweep. This is much richer material than the trade and welfare accounting records left behind by the Minoans, and is more similar in character to the contemporaneous Egyptian historical legacy.

    Moreover, the accuracy of the Hittite accounts are corroborated by parallel Akkadian and Egyptian accounts and have some archaeological corroboration as well. In the 13th millenium BCE, there was a border boundary that Egypt and the Hitties shared and the these kingdoms engaged in at least a couple of instances of royal bride exchange. The pre-Hittite non-Indo-European sequence of languages in Northern Mesopotamia (in a nutshell, Sumerian, Akkadian, Kassite-Hurrian, a Mittani superstate for a while) is documented quite far back. Similar, there is quite solid evidence of the non-Indo-European substrate in the Aegean.

    Moreover, while we do not know from whence the Hittites or other Anatolian communities came to Anatolia with any real certainty (eastern, western and maritime routes are all plausible), we do know that Indo-European Greek arrived in the Aegean first in mainland Greece from the North and then worked its way South, which would be a rather improbable trajectory for a language that had recently diverged from an Anatolian source.

    The survival of Hattic as a liturgical language long after the Hittite language had replaced Hattic in commerce and politcs and daily use is also instructive. This very strongly parallels the sequence of events in which Sumerian was superceded by Akkadian in Mesopotamia. Other examples, like the survival of otherwise extinct languages like Latin, Coptic, Sanskrit and Hebrew as liturgical languages also support this sequence.

    The pattern of Indo-European Anatolian language attestation in Anatolia looks much more like the pattern in which the Greeks and the linguistically diverse early Italic speakers established colonies in Southern Europe and the adjacent islands. Anatolian languages that predate Hittite (and most of the Anatolian languages arise after the demise of the Hittite empire in a process more like the diversification of the Romance languages from Latin or the diversification of the Indo-Aryan languages from Sanskrit), appear in little patches here and there that discontinous in Anatolia. Also, both the earliest Hittite metal working evidence, discovered archaeologically, and the earliest Indo-Aryan metal working evidence, discovered archaeologically, is contemparaneous, and in both cases shows antecedents in the Caucusas – suggestive again of intrusive colonies of recent migrants rather than a widespread continuguous indigenous linguistic community.

    Thus, even if the Anatolian languages are indeed a very basal branch of Indo-European suggesting a great time depth of a common ancestor of these langauges and other IE languages, putting their point of origin in Anatolia, rather than, for example, a source point for colonists in Anatolia somewhat in the Caucasus in a model, may be misleading in terms of identifying a probably location of origin for these languages.

    A similar point comes up with Armenian. One of the quite plausible possibilities suggested by some of the isoglosses between Armenian and Greek is that Armenian was a divergent and now lost Greek dialect of a people who migrated across Anatolia from the Aegean in the wake of the collapse of the Hittite empire and then experienced areal influences arising from language contact in the intervening centuries that made it mutually unintelligible with Greek. There is historical evidence that such a migration happened, but we don’t know how big it was, whether the migrants were successful when they arrived, and whether these migrants can be identified with the Armenians. Armenian itself is not attested in the historical record until much, much later – it was spoken in a place that was not on the “stage” that writers in literate languages were aware of for a very long time. Of course, if Armenian is indeed a basal branching from a most recent common ancestor of Indo-European languages, a source in or near the Caucasus is at least as plausible as one in Anatolia from a geographic perspective.

    Another key point is in regard to what in particular is distinctive about the Anatolian languages, or the Greek languages, for example, relative to other IE languages. In both cases, the evidence for the deviations being causally attributable to substrate influences is not insubstantial. We do not see in either of these languages a pattern of random differentiation by isolation; instead we see a pattern of significant lexical borrowing and notable phonetic and grammatical influence from the substrates.

    The story this evidence tells is that rather than the Indo-European contribution to Anatolian or Greek being particularly ancient or basal, it is telling us something about how very much different the substrated in the Balkans and Anatolia were from the substrates in other places to which the Indo-European languages expanded.

    If we want evidence of what the basal Indo-European languages were like, Tocharian, which involved a community in which there was no substrate and was settled basically as virgin territory (and in which there is also almost no evidence of admixture until very late) is a signal with a lot less noise in it. By analogy, Icelandic which is also an isolated language fringe with no substrate is also very conservative, experiencing just 3% lexical change over a time period when English saw a 32% lexical change. Language contact, substrate influence and intentional language differentiation by a community to separate their dialect from another community aren’t just side stories when talking about the rate of language change – they appear to be the dominant source of language change with random drift playing a comparatively insignificant role.

    We have no good evidence of what impact, if any, in Indo-European languages can be attributed to, for example, an LBK Neolithic or Harappan substrate linguistic influence on Indo-European languages. One plausible story of Indo-European ethnogenesis would be that it arose from the fusion of an LBK Neolithic and Uralic community involving a relatively equitable creolization in the vicinity of its purported urheimat. Another is that Indo-European ethnogenesis may have involved substantial borrowing from the populations at the fringes of the Harappan trade network at places like BMAC, perhaps magnified by the impact of the Harappan diaspora created when its empire collapsed and its people relocated in the wake of the drying up of the Saravasti River (notably, the Indo-Aryans, unlike many Indo-European populations, do not have an origin myth that involves a migration from somewhere else). An ethnogenesis born from the interactions of one or more languages of the Caucasus with Uralic would be another possibility. Tracing your way back to a time and place in which the most recent common ancestral language of all Indo-European languages doesn’t tell you anything about the origins of PIE itself.

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 »