The story of Bedřich Hrozný realizing that Hittite was Indo-European when he saw that wa-a-tar must mean ‘water’ is well loved among historical linguists and is familiar to anyone who’s read anything about the decipherment of Hittite. But it’s easy to misunderstand, and Piotr Gąsiorowski devotes his latest Language Evolution post to explaining exactly how the correlation works and why it is so important. Much has been learned since I left the field several decades ago, and I confess I find it both exciting and moving to see the correlations laid out so clearly and convincingly; it’s one of those things that makes me intensely nostalgic for the days when I had my nose in dusty volumes of Kuhns Zeitschrift. Here’s his conclusion:

To sum up, the fact that Hittite wātar is similar to English water is interesting but not particularly impressive as an isolated observation. Similarities can be found between any languages chosen at random. It’s far more significant that the inflectional pattern visible in Hittite helps us to understand the origin of the diversity displayed by cognate ‘water’ words elsewhere in the IE family and is part of the evidence used in the reconstruction of the PIE morphological system. It’s those pervasive shared patterns that demonstrate the membership of Hittite in the IE family.

But I urge you to read the whole thing; it’s a mini-course in Indo-European, and if you can get it under your belt you’ll never again be fooled by flashy claims based on surface similarity.


  1. dearieme says

    It makes me wonder how often water’s cognates get pressed into service to mean river (e.g. Water of Leith) or lake (Derwentwater) in languages other than English.

  2. Trond Engen says

    Vatn (with the n-stem) is the native word for “lake” in non-Danish Scandinavian. The other “water” word, *akW-, is the origin of å, the native word for “river”.
    (And by the way, å and eau is a nice example of convergent evolution.)

  3. Thanks for this link. It brings up nostalgic memories for me, too. I was taking Summerian and Indoeuropean studies classes in the same building where Hrozný had worked from someone who had been taught by several of Hrozný’s pupils. I was even there for the last series of Schindler’s lectures (who sadly died before he could complete them). I was lured away by far more theoretical pursuits but I do miss those days.

  4. I posted this over theresince I thought piotr would enjoy it:
    “I have an example of the exact opposite to the careful process and pattern of actual evidence you are talking about here. The is the kind of thing you are criticizing.
    New proto-world etymon! English ‘salad” and Lushootseed ‘sʔəɬəd” “food”. Pretty undeniable, right?
    Well no. ‘sʔəɬəd’ consists of a root ‘ʔəɬ’ “eaten” (all Lushootseed roots are monovalent) and the transitivizing suffix ‘əd’ and the ‘s’ is a nominalizer. So it means “what [we] eat.”
    Whereas ‘salad’ consists of a root for salt with a stativizing suffix.
    Dearieme, as I recall that happened in Chinese, where the word “shui3” originally referred to rivers and then shifted to water as ‘he2’ came to refer to rivers, or in the south ‘jiang1’. There is a fair amount of this kind of thing in Chinese, completely unrelated or remotely related words for the same sememe. The closest I ever got to an explanation was Zhang Kun saying years ago that there was a theory that the people who set up the Zhou Dynasty spoke a different language than the Shang they replaced, and the same could be tru for other eras too, with regional and non-Chinese words coming in as adstrate.

  5. There’s also this famous story told by Robert Dixon in his Languages of Australia (p. 196 in the 2011 edition):
    In 1964 I found, after considerable search, what turned out to be the last speaker of Mbabaram …. One of the first words obtained was /dɔg/ for ‘dog’! Hale later pointed out that it too is a reflex of */gudaga/, involving raising of the second /a/ to /ɔ/ after an initial velar, then loss of initial C₁V₁ and loss of the final vowel. Other examples were found of each of these changes, showing them to be as regular as the changes described for language further north
    [Proto-Paman *gudaga ‘dog’ also yields Yidiny gudaga, Dyirbal guda, Uraði /utaɣa/, Mbiywom /two/, etc.]

  6. Corretion: languages further north

  7. Yes, I posted about that wonderful coincidence here (I can’t believe it’s been over seven years!).

  8. “Hale later pointed out that it too is a reflex of */gudaga/,”
    Yes!! But don’t you see, that’s cognate with “good dog”, so there’s a connection????? They are all Japhetic languages!!!!!

  9. marie-lucie says

    I started to write a comment in praise of Piotr’s article, but an error message told me that I was not allowed to post on this site.

  10. marie-lucie says

    I saved it, so I’ll try again:
    (I pasted my comment right here, but it was rejected again. I fail to see where I might have used a forbidden sequence of letters, or done anything else unacceptable).

  11. The *akwa- words in English are few, other than the obvious Latin borrowings: there is eyot, more etymologically spelled ait; there is island, where the s is unetymological and comes from isle < Latin insula; and there is (very surprising to me) ewer and sewer < Normand < Latin aquaria (> French aiguière) and exaquaria ‘drainage channel’.

  12. On the other hand, toponymic/hydronymic reflexes of OE ēa ‘river’ and īġ ‘island’ are extremely common throughout England.

  13. m-l: If you’ll e-mail me your comment, I’ll try and figure out what the problem is. Sorry about that.

  14. And in Scandinavia.

  15. Trond Engen says

    And in Germany. But I think the point was that toponyms show that the *akwa words used to be common in English. For most Scandinavian you don’t need that (although elv has replaced å as primary form in many regions).

  16. Trond, I meant the word ‘Scandinavia’ 🙂

  17. Heh.

  18. Trond Engen says

    Heh too!

  19. Trond Engen says

    Something like *skandin aGwijo, i.e. Skånøya, possibly used for the southern tip of the Scandinavian peninsula, possibly a misnomer from the ancient marketplace at Skanör.

  20. John Cowan: “aiguière” is a Provençal loan into French. Diachronically intervocalic Latin /kw/ yields /v/ in French and /g(w)/ in Provençal. A better comparandum to English “ewer/sewer” would be Modern French “évier”, which must derive from AQUARIUM and not AQUARIA, however.
    The fact that French and Norwegian both share /o/ with the same meaning and etymology (good point, Trond), when you’ve forms such as AIGO, AGUA or ACQUA (all from Latin AQUA) in Provençal, Spanish and Italian respectively, does show that there are no easy shortcuts when it comes to language classification.

  21. Trond, there’s probably no need to include the ‘G’, if you’re talking anything Germanic. The PGmc form seems to have been pretty clearly *aujō, which is (possibly with a little bit of analogy) from pre-Gmc *agʷjō

  22. Trond Engen says

    Nelson, you’re right. I misremembered the details.
    (It’s probably time to confess that I’m writing off the top of my head. On a nice day like this I can’t be bothered to go indoors and check my books.)

  23. Trond Engen says

    Etienne: Thanks. But it’s not an exact semantic match, since it can’t mean “water” in Scandinavian, only “(small) river”.

  24. marie-lucie says

    Etienne : about aiguière, évier, both ultimately from Latin AQUA:
    Definitions: une aiguière : (from online dictionary) ‘a decorative pitcher-shaped usu. tall and slender vessel with a handle and spout’
    un évier: ‘sink, esp. kitchen sink’ (a bathroom sink is called “un lavabo”, but “un évier” could be in a kitchen, a basement, a lab, etc).

  25. Yes, I posted about that wonderful coincidence here (I can’t believe it’s been over seven years!).
    The original post and comments were in 2006, but more was added six years later:
    Yeah, it looked to me as if Noetica took a position without really thinking it through, then dug in and didn’t want to rethink it. His position really doesn’t make sense.
    Posted by: languagehat at July 26, 2012 02:35 PM
    I was doubly surprised to find such commentary after so long a break, because the “position” mentioned is nothing of the sort. Noetica’s “position” was stated in these terms:

  26. [After actually hitting “post”, on a handheld device:]
    “It would take some argument to convince me that the Mbabaram word is not derived (simply or complexly) from the English word dog.”

    “I wanted an argument, recall accurately, that the Mbabaram word is not derived (simply or complexly) from the English word dog. And I would want that argument to show that the current Mbabaram word is not etymologically overdetermined.”

    “For all we know, on the evidence given here so far, the relations between dingo and dog are similarly problematic in indigenous Australian languages and cultures, and also in the complex pidgin and creole languages and the milieux associated with them.”
    I hope that none of this is “digging one’s heels in”. Of course there is evidence and rather compelling argument to consider, and some of it did emerge. A plea for caution and patient assessment of argument is not a “position” that anyone dug heels in to maintain, in that thread with its six-year hiatus. Sometimes one pleads in vain for interlocutors to “recall accurately”, I fear.

  27. Trond, the general view is that the word should have been Scadinavia, after the OE Scedenīġ, and that the standard form is a scribal error. That would make it either the ‘shadowed’ or the ‘dangerous’ (Eng. scathing) land-on-the-water; I always think in the latter case of Robert Hughes’s history of transportation to Australia, The Perilous Shore.
    N, that was one of the postings Hat reopened in 2012 for the tenth anniversary.

  28. Er, what terms? His basic statement was “It would take some argument to convince me that the Mbabaram word is not derived (simply or complexly) from the English word dog,” and that was fully refuted by various people who understood the linguistics involved (not to mention in the post itself). Noetica apparently had a hard time accepting the role coincidence plays in these matters, which is not surprising, since most people do.

  29. John Cowan says

    In any case there can be no fully convincing arguments that A is not derived from B in historical linguistics. It may be, at some level we can’t establish, that two similar words in random parts of the world are inherited from a common ancestor. We can explode any one argument, but not all possible arguments. In this case absence of evidence really is evidence of absence.

  30. The actually doable fallback point is to show that whatever surface similarities two words may have result from parallel developments, and that they go back to ancestors with more distinct shape and meaning.

  31. Exactly.

  32. David Marjanović says

    That’s the difference between proof and parsimony.

  33. Re the discussion of akwa above: a few days after this post, Piotr Gąsiorowski continued his series with a thundering condemnation of reconstructing that word beyond Germanic+Latin. Also, there was a re-run of the same discussion here less than a year later.

  34. January First-of-May says

    The actually doable fallback point is to show that whatever surface similarities two words may have result from parallel developments, and that they go back to ancestors with more distinct shape and meaning.

    It also helps if they can be proven to divide into very different morphemes.

    Previously on LH, featuring (among others) my list of incredibly similar words for “seven” in various languages across Eurasia (admittedly some of those might in fact be borrowings and/or substrate influence).

  35. thundering condemnation

    Ouch! I almost feel sorry for Bengtson & Ruhlen — I’m only an ‘umble etymologist, may it please yer ‘onour, a humble drudge.

    And ref etymonline from a few days ago, they’re still showing “from PIE root *akwa- “water.””.

    I’m not keeping up with the play: Eng. “island” is not cognate with “isle” < Latin insula; the ‘s’ in “island” is an interloper. etymonline again:

    1590s, earlier yland (c. 1300), from Old English igland, iegland “an island,” from ieg “island” (from Proto-Germanic *awjo “thing on the water,” from PIE root *akwa- “water”) + land (n.).

    PIE again. Shhh!

  36. https://en.wikipedia.org/wiki/Tambora_language

    Somebody compiled a wordlist for this language and one term leaped out: The word for “star” is kingkong.

    Did the folks who wrote the script about a giant ape see the wordlist?

    And I wonder if anybody tried to suggest giant apes from another star system came…

    And slightly more seriously, I wondered if both of the following is grammatically correct? Since Hittite is an extinct language…

    The Hittite word for water is waatar.
    The Hittite word for water was waatar.

  37. David Eddyshaw says

    thundering condemnation

    With ʔAQ’WA, Bengtson and Ruhlen, not for the first time, were handicapped by their ignorance of Western Oti-Volta. The form is clearly derived from the Kusaal ku’om “water” (stem ku’a-.)

    The Scandinavian cognates of course reflect common inheritance from proto-Scandi-Congo. The Latin form is borrowed from this via Gothic.

    The Hittite word for water is/was waatar.

    It’s even more complicated when you start talking about reconstructed protolanguages. Probably the best thing would be to write all works on historical linguistics in a language with proper evidentials.

Speak Your Mind