I’m slowly working my way through Nicholas Ostler’s Empires of the Word: A Language History of the World, a book I’d been dying to read and finally got last summer, and I just hit the brief excursus on Elamite (which is probably related to Dravidian). I had not realized that Elamite was still spoken when Alexander conquered the area (and possibly as late as the Arab conquest), nor had I realized that Elam became the heart of the Old Persian empire:

Two generations later, in 522 BC, Darius (Dārayavauš), the Persian heir to Anshan, took control of the whole Persian empire, which by now extended from Egypt and Anatolia to the borders of India. Despite two abortive Elamite rebellions shortly after his accession, he chose Elam as the hub of this empire, with Susa itself (known to him as Šušan) as the administrative capital, and Parša, i.e. Anshan, as the site for a new ceremonial capital, to be better known in the West by its Greek name of Persepolis.

He goes on to make the following interesting observation:

The Persians had never prized literacy very highly. Famously, their leaders were educated in three things only: to ride a horse, to shoot a straight arrow, and to tell the truth. So their Elamite neighbours, with two thousand years of cuneiform education behind them, were well placed to be extremely useful in the more humdrum side of empire-building.

Which means the Elamites played the same role with respect to the Old Persians as the Persians played with respect to the Turks a millennium and a half later.
An amusing sidelight: “Nevertheless, Elamite must have continued to be spoken in Elam [after a long period of Akkadian domination], since in 1300 BC it springs back to life as the official language, replacing Akkadian for all written purposes, except curses.” (Emphasis added.)


  1. My theory is vindicated. Just fill in the blanks.

  2. See, I almost put in “be still, John Emerson’s beating heart” after the Dravidian reference, but I thought better of it.

  3. Nah. Proto-Hungarians, the lot of them.
    Ostler’s book is wonderful, but so dense. I take it up from time to time, chip off a few chunks of cortex candy, then set it aside while I relish those. Particularly useful for all that Mesopotamiana.
    Does anyway know his account of the development and spread of Latin, Ad Infinitum? I’ve only caught the odd short description, but it appears to be equally fine. Reviews at Amazon are *****x7.

  4. After reading about Ad Infinitum on this here never-to-be-praised-enough (onvolprezen) blog, I ordered it for the shop I work in, and have not looked back since. My customers are enthousiastic and they are usually classicists ( is that a word?… students/scholars of Latin & Greek) so they should know. Their verdict is that it is much but much much much better than “Latin” by Tore Jansson. I’ve bought them both for myself so soon enough I’ll see.

  5. and Parša, i.e. Anshan, as the site for a new ceremonial capital, to be better known in the West by its Greek name of Persepolis.
    But that’s not true.
    Anshan lies 43 km west of Persepolis. I can’t find any reference to Ostler’s theory. I wonder if he just goofed.

  6. Ostler’s book is wonderful, but so dense. I take it up from time to time, chip off a few chunks of cortex candy, then set it aside while I relish those.
    I wonder if he just goofed.
    Quite possibly. He is, after all, writing a popularization of a huge field of knowledge, and you can’t expect every detail to be perfect (I nitpicked a couple of things in my review of Ad Infinitum). If one is particularly interested in any detail, one should check it against other sources. But then, that’s true of just about any book.

  7. Ahem… I see. Reviewed here while I was preoccupied with my Wikipedic adventures, it was. That’ll teach me.
    OK, I’ll order Ad Infinitum and ever so slowly devour it also.

  8. For all written purposes except curses? Harumph.

  9. I think that the Wikipedia article gives an overly optimistic impression of the status of the Elamo-Dravidian proposal. Although it is not considered a crank hypothesis (the evidence put forward by McCalpin was valid in type) I think that it is generally regarded among historical linguists as unproven.

  10. Were these serious formal curses, the kind where you try to get lightning or leprosy to strike the guy, or the kind of casual curses you ake when you drop something on your foot? My guess is that it was the former — cursing in those days was taken more seriously than it is with us.

  11. And Poser is a partypooper.

  12. He’s just doing his job. And I agree, the former is more likely.

  13. Hi! Can I also be a partypooper? I actually spent time hoping to convince myself that Elamite and Dravidian were related, and ended up with nothing specific. Results are here, if anybody’s interested: http://starling.rinet.ru/Texts/elam.pdf. Sorry John Emerson. 🙂

  14. David Marjanović says

    There I want to bring up the paper that says there’s no evidence so far that Elamite is closer to Dravidian than English is, only to find that the author himself has already been here! 🙂
    Elamite is a fascinating language as long as you don’t have to learn it. It’s one of two languages known to conjugate nouns: sunkik “I, the king”, sunkit “you, the king”, sunkir “(he,) the king”. The other noun-conjugating language is Nàmá.
    However, let me mention another pet peeve: Of all scientists, only historical linguists ever use words like “proof”, and they really, really, really shouldn’t. “It’s unproven” is what creationists say about evolution because they don’t know what they’re talking about. (Of course it’s unproven, duh — science cannot prove, only disprove.)

  15. Truth must always fight its way through a thicket of lies and error, but in the end it will always prevail.
    David, George, Bill, and future John Doe partypoopers 1……n are all forgiven, of course, because the truth is unintelligible for those whose brains are infested with thetan engrams.

  16. David Marjanović says

    Clearly we are educated stupid.
    (Warning: Follow this link at your own risk. It may damage your sanity. It’s legendary enough to have a Wikipedia entry, and the blog I read most frequently uses it as the measure of insanity on a logarithmic scale.)

  17. Leo Caesius says

    Ostler’s book is entertaining, but I’d be sure to double-check anything he says. I wouldn’t want to claim any special expertise on the non-Middle Eastern side of things, but he makes plenty of goofs about that region. For example, he mentions that only Christians have managed to preserve Aramaic to the present date, although the Mandaic dialect was spoken up until the 8th century — which would come as a surprise to my Mandaean informants, who continue to use the modern form today in their daily life. Some of the stuff he says about Phoenician doesn’t sit well with me either; he says that it was unknown outside of its own settlements, which doesn’t explain why dialogue in Phoenician appears in one of Plautus’ plays, or why the last attestation of Phoenician is found in the hinterlands way to the south of Tunis and Tripoli.
    He describes Ethiopia as a “Christian country,” which is somewhat inaccurate (perhaps a third or more Ethiopians are Muslim) and claims that Zoroastrianism survives only India (there are still perhaps 45,000 Zoroastrians in Iran itself).
    There are other similar blunders, which do not exactly inspire me with confidence about the sections of the book that do not deal with the Middle East. I will say that I think he’s being more than a little silly when he says that the writing is on the wall for Chinese (p. 173).

  18. David Marjanović says

    There are Jewish and Muslim speakers of Aramaic, too.
    (And BTW, there are dialects that have preserved the emphatics as ejectives to the present day…)

  19. Actually, we historical linguists do not speak of “proof”. There’s a difference between saying that something is “unproven” and using the term “proof”, much less confusing scientific evidence with mathematical proof. The problem is what to say to a non-specialist audience. If we say “there is no good evidence that X is related to Y”, many non-specialists will take that to mean that nothing much has been adduced, when often it means that what evidence has been adduced, which may be quite a lot, is not probative. They will then look at a long list of putative cognates of the Greenbergian sort and say “what do you mean there’s no good evidence? There is piles of it?”
    In any case, other scientists do also casually speak of “proof”. Here’s a biological example
    Here’s a statement that: “a direct connection between aluminum, AD and disruption of Krebs cycle enzymes is yet to be fully proven.” from the European Journal of Biochemistry. Google will provide additional examples.

  20. Michael Farris says

    When talking about language relationships, I find the word ‘accepted’ to be useful.
    The most widely accepted theory is that Vietnamese and Cambodian are related while Cambodian and Thai are not.
    The grouping Ural-Altaic is no longer generally accepted as valid.
    Not ideal, but not the worse either.

  21. Actually, substituting “good evidence” for “proof” or vice versa does not solve the problem. If we say “this bunch of Greenbergian evidence does not prove that X is related to Y”, there’s nothing to stop non-specialists from questioning the validity of that statement either. The easiest way low probability relationship theories may be countered is through falsifying them with competing theories. For Elamite, I have tried to show that it is not difficult to put forward equally “strong” competing theories that relate it to almost any other family in Eurasia. I cannot say that this demonstration invalidates McAlpin’s hypothesis, or “disproves” it – it just shows that a close “Elamo-Dravidian” relationship isn’t the most probable reason for observed similarities between Elamite and Dravidian, not for now, at least.

  22. David Marjanović says

    They will then look at a long list of putative cognates of the Greenbergian sort and say “what do you mean there’s no good evidence? There is piles of it?”

    Then just say “you said good evidence — this isn’t good evidence”.

    In any case, other scientists do also casually speak of “proof”.

    Yes, but it’s remarkably rare, if I may switch into descriptive mode. Compare the usage of “shown” or “demonstrated”. (For example, instead of “is yet to be fully proven” in the example you quote, the people I know would have written “is yet to be demonstrated in detail”.)

    The easiest way low probability relationship theories may be countered is through falsifying them with competing theories.

    Showing that one phylogenetic hypothesis is less parsimonious than another is how we do phylogenetics in biology. You have shown that for Elamite several previously overlooked hypotheses are just about as parsimonious as the Elamo-Dravidian hypothesis, demonstrating that (based on the evidence known and examined so far) this hypothesis should not get preferred treatment, and also demonstrating that more research needs to be done.

  23. Michael Farris says

    IIRC Greenbergian methodology isn’t designed to ‘prove’ reations, only to disprove them. That is using his methods you can say two languages aren’t related (but not that two languages are related). A lot of the larger groupings are presumably just waiting for people to show they aren’t related after all.

  24. I think even most “mainstream” linguists agree that you can’t ever ‘disprove’ relationships – you can really only say that the evidence is not good enough for suggesting a particularly close degree of relationship. So, to be more precise, if I say that there is no “Elamo-Dravidian”, what I really mean is that there is no closer link between Elamite and Dravidian than between Elamite and a whole bunch of other families, and that McAlpin’s evidence for establishing this closer link is insufficient. This doesn’t mean that Elamite and Dravidian can’t be related, along with other families, on a much higher level – but we certainly don’t have the sufficient evidence to demonstrate that now (although some hints look interesting).
    My opinion is that Greenbergian methodology is great for working out preliminary hypotheses. It gives you ideas to work with, based on some rough concepts of phonetic typology, which can be later either molded into more solid scientific theories or discarded. It’s only when preliminary hypotheses start being treated as categoric statements that trouble begins.

  25. Once again, I express my gratitude for the learned and stimulating comments that turn up on this blog, which is giving me at least as good an education as grad school did.

  26. The things got more difficult for the Elamo-Dravidian hypothesis on the ancient DNA front as well.


    The new paper of Shinde et al. characterizes the “Iranian-like” DNA component of the Indus Valley Civilization and finds out that it isn’t “sufficiently like” the known Iranian ancient samples, both from the proto-Elamite times and in the preceding millennia. The similarity between the Indus Valley component and the old Iranian specimens is consistent with a split between the two populations some 12,000 years ago, even before agriculture. (Of course it’s faintly possible that the dissimilarity isn’t really due to an ancient population split, but was caused by a more recent admixture from a different local population picked by the ancestors of the Harappans on their way East).

    But the simplest hypothesis seem to be that the ancestors of the Indus Civilization learned agriculture somewhere in the region of Eastern Iran / Afrghanistan / India by a culture transfer from the West…

    On a different note: is the linguistic prehistory of the Dravidian languages consistent with their arrival with the ancestors of the IVC?

  27. On the other had, the other of the two recent papers (an updated version of Narasimhan preprint already discussed here) ( https://science.sciencemag.org/content/365/6457/eaat7487 ) favors the hypothesis of the Dravidian nature of the Indus Valley Culture based, largely, on the linguistic arguments (the contemporary Dravidians are characterized by prominent Ancestral Sound Indian DNA connection, which is derived from the two major components, an IVC-type DNA as well as a more ancient substrate DNA of the peninsular South Asia from before the agricultural times). They follow the books of Fortson 2011 and Parpola 2015 to argue that known IVC seals represent Dravidian words and names.

    On the other hand, they cite Krishnamurti 2003 re: the proto-Dravidian reconstructions of the names of local animal and plants which (unless they were early borrowings picked by the invasive proto-Dravidians) would lend support for the local origins of the Dravidian languages.

Speak Your Mind