More on IE wheels and horses

« previous post | next post »

Don Ringe's answer (" Horse and wheel in the early history of Indo-European") to the question that David Marjanović asked about Don's earlier post ("The Linguistic Diversity of Aboriginal Europe") stimulated some further questions, included these from Robert:

What was the Anatolian word for wheel? Given its lack of mention above, I'd assume it isn't cognate with the Indo-European term. Is it thought to be a borrowing from some other language, or is its origin unknown? if it's a borrowing, that would presumably give a handle of when those languages moved into that area.

Are there any other unexplained e to i transitions in Greek? If a dozen other words were affected, with no apparent pattern, I'd guess that would change the relative likelihood of the possible reasons.

Were horses domesticated just once, or many times. While a word for horse can predate domestication, it would seem plausible that it was repeatedly borrowed along with other horse related terminolgy as domestication spread, even into different language families. Conversely, if horses were domesticated independently by two cultures, they're unlikely to have borrowed the word from each other, even if there's a strong resemblance.

I've posted Don's response below — as before, a backup .pdf form is here in case some characters or formating got screwed up.


[Guest post by Don Ringe]

Many thanks to David, Robert, and the other bloggers for the kind words! I'll certainly keep sending Mark chunks to post. Here are some quick answers to Robert's questions.

The Hittite word for 'wheel' is ḫūrkis, and it resembles Tocharian A wärkänt and Tocharian B yerkwantai (oblique case; the nominative is not attested) enough to make us want to derive them from the same inherited root. The obvious choice is the ancestor of Vedic Sanskrit vr̥j- 'twist' (aorist 3sg. injunctive várk, subjunctive várjati, etc.), and it's straightforward to reconstruct a PIE root *h2werg- 'turn'. (There are probably other cognates too—Latin vergere 'to bend, to incline' looks like it ought to fit—but I haven't got all the relevant references here at home.) But when you investigate these 'wheel'-words in detail, the results are dismaying: even the Tocharian words can't reflect exactly the same preform, and the Hittite word is obviously an independent derivative. Here's what I mean about the Tocharian forms. From the fact that an initial *w- has been palatalized to y- in TB we can infer that the following e must reflect pre-Tocharian long *ē. But TA ä can't reflect *ē; the closest sound that it could reflect is short *e. Moreover, if the preform had a second *w after the *k, as the TB form clearly does, then the second TA ä should have been rounded to u; but clearly that has not happened. For those reasons we can't even reconstruct a Proto-Tocharian form for 'wheel'; we can't say exactly what Tocharian word replaced the inherited word when the latter's meaning was shifted to 'wagon'.

So though the Hittite and Tocharian words are probably derived from the same PIE root, it seems clear that they were derived independently—and that obviously offers no support at all for a PIE word for 'wheel'. In the current state of our knowledge, the most we can say is that the common ancestor of the non-Anatolian branches had a word for wheel (which of course could have been invented not long before the pre-Tocharians took off for the Altai).

As for cases of *e becoming i in Greek: there are few, and most of them are confined to single dialects and are obviously very late developments. (For discussion of individual cases see e.g. Carl Buck's book The Greek dialects, U. of Chicago Press 1955.) But now that you mention it, there's an odd word that shows an unexpected vacillation between e and i. In Homeric Greek it's δέπας /dépas/, and it refers to some kind of drinking vessel. The same word apparently shows up on the Linear B tablets, roughly half a millennium earlier, but there it's spelled di-pa (the syllabary doesn't allow for the spelling of most syllable-final consonants). Since there is no good IE etymology for such a word, the best guess is that it was borrowed from some non-IE language which had a different vowel system. It wouldn't be that surprising if 'horse' were somehow implicated in all this. But of course we need more evidence (a lot more!).

On the domestication of horses the leading authority is David Anthony, who has been working on that and related problems for almost thirty years. (He finished his dissertation in our Dept. of Anthropology in 1985.) He says a good deal about it in The horse, the wheel, and language (Princeton U. Press, 2007), which is a good read; I found the hard-core archaeological chapters in part two especially interesting and informative.

[Update 1/11/2009:

This is a couple of quick replies to James D and Etienne's comments. (Backup .pdf is here.)

I think it's true that Epo- only occurrs in three-member Gaulish names, though I haven't had a chance to check. But they're unproblematically interpretable: Epo-so-gnātus is 'horse-well-knowledgeable', i.e. 'Well-versed-in-horsemanship', while Epo-rēdo-rīx is 'Horse-rider-king'. Unless I'm mistaken, Welsh ap is a reduced form of map, cognate with OIr. mac 'son' and reflecting a preform *makwkwos. It's also true that the caballus-words seem to have been borrowed from Latin into the Insular Celtic languages; the Welsh and Irish words don't match the way they should if they were cognates. There is a further Celtic word for 'horse', which also shows up in Germanic but has no other unarguable external cognates: OIr. marc, Welsh march < *markos; ON marr, OE mearh, OHG marah < PGmc. *marhaz < *márkos. (The pre-Germanic accent can be recon­structed in this case because the *h did not become voiced. A derived feminine survives in Modern English 'mare'.) As is often the case, we can't tell whether this is a shared inheritance or an early loanword, but in any case it seems restricted to northwestern Europe. (The interaction between Germanic and Celtic will come up again in a later post; I need to check a bunch of things in it before sending it to Mark.)

I write Latin equos for a specific reason. The standard form cer­tainly is equus, but I think that actually first appears in inscriptions in the 1st century CE. What hap­pened was approximately this. (Some of the dates may be a bit off; again, I don't have all the relevant references to hand right now.) Short o in most final syllables became u sometime late in the 3rd century BCE, I think—the spellings with o in the Senatus Con­sultum de Bacchanalibus, datable to 186, are deliberate archaisms—but the change was inhibited by a preceding u or v (spelled the same, of course; but I mean a preceding high back round vocalic, whether syllabic or not). Cicero still pronounced equos, servos, mortuos, etc., and wrote those forms with an o. During the Augustan period, I think, a further change of vo to u occurred, and we start seeing ecus in inscrip­tions. Finally the following generation (more or less) regularized the paradigms—this was an "analogical" change, not a regular sound change—and produced equus. Of course isolated words, and paradigms that didn't have a relevant alternation in them, weren't "restored"; thus secundus wasn't changed because it was no longer felt to be part of the paradigm of sequor. This is all spelled out in Ferdinand Sommer's Handbuch der lateinischen Laut- und Formenlehre (I don't have the page refs. to hand). Which form you prefer to use is basic­ally a matter of taste; I use the older one in an attempt to be chronologically con­sistent, since most of our Classical Latin grammar is Ciceronian.

Obviously a detail like this makes no difference in this context; so why bother to write equos when everyone who can read Latin expects equus? Basically as part of an attempt to get into the habit of paying attention to every detail every time—and because I've spent about a third of my career wrestling with the chronology of language changes, I notice this particular detail. It doesn't seem possible to avoid errors completely (at least I can't do it!), but cultivating this kind of habit minimizes them. Of course it also shows that you have to be obsessed with language change to pursue this kind of work profes­sionally; but maybe that was obvious already. ]



15 Comments

  1. David Marjanović said,

    January 11, 2009 @ 1:03 pm

    reflecting a preform *makʷkʷos

    I hear that /makʷi/ is attested in Ogham inscriptions, if we assume that the Ogham letter that corresponds to Q stood for /kʷ/.

  2. David Marjanović said,

    January 11, 2009 @ 1:57 pm

    There are probably other cognates too—Latin vergere 'to bend, to incline' looks like it ought to fit—

    What about German würgen "strangle"? No idea where the vowel comes from, though.

  3. Etienne said,

    January 11, 2009 @ 3:30 pm

    Professor Ringe: thank you very much for taking the time to reply. There is a core assumption in your posts: namely that the proto-Indo-European form in fact had the meaning "horse". But let us imagine that the original meaning was less definite, perhaps "large quadruped" or the like (the meaning "donkey" of the Armenian reflex is worthy of notice in this context). If we imagine a spread of Proto-Indo-European that took place before the domestication of the horse, it is more than plausible that the subsequent spread of domestic horses would lead to the inherited Indo-European word (*whatever its phonological form had in the meantime become in various Indo-European-speaking communities*) everywhere undergoing a process of semantic narrowing and becoming the word for "horse".

    Here's a partial analogy: all Germanic languages today have a cognate of English "God" to refer to the Christian god. The original meaning of the proto-Germanic word was a non-Christian god, obviously: but if we had no knowledge of the chronology of the spread of Christianity compared to the chronology of the break-up of Proto-Germanic, we would have no way of knowing whether the proto-Germanic word (however accurately we are able to reconstruct it as far as phonology goes) referred to the Christian god or not. In like fashion, I accept the reconstruction of the phonological form of the Proto-Indo-European form of the word which in attested Indo-european languages meant "horse", but am less certain as to its original meaning in the proto-language.

  4. David Marjanović said,

    January 11, 2009 @ 5:18 pm

    The donkey-drawn Sumerian war chariots come to mind… and Armenian is a fairly early-branching subgroup… so I wouldn't be too surprised if the original meaning was in fact "donkey". But still, when no meanings other than "donkey" or "horse" are attested, and "donkey" is that much limited to Armenian that even the attested Anatolian reflexes (Anatolian being the sister-group to all the rest together, and the Anatolian languages having been spoken in regious where various donkeys occur in the wild) all mean "horse"*, this does seem less likely than the alternative.

    * How securely is this actually known, though? Any philologists of Anatolian languages out there…?

  5. Etienne said,

    January 12, 2009 @ 1:43 pm

    David: I would agree with your point if the only possible assumption was that all branches of Indo-European other than Armenian had shifted the meaning from "donkey/large quadruped" to "horse" independently of one another: but a contact explanation is possible, whereby this semantic innovation began in one branch of Indo-European and thence spread to other branches (perhaps with the spread of domesticated horses), without different reflexes of *EQUOS being actually borrowed from one branch to the next of course. It was Leonard Bloomfield who pointed out that a careless linguist might reconstruct a proto-Algonquian word for "whiskey", which in all Algonquian languages is a compound of the words "fire" and "water", each of which is inherited from proto-Algonquian: the innovation, whereby the two words are compounded with the meaning "whiskey", is of course a post-proto-Algonquian innovation, which involved a spread of new meanings rather than new forms: and unfortunately, in comparative and historical linguistics, whereas the regularity of sound change allows us to weed out inherited from borrowed *forms*, there is no way to weed out inherited from borrowed *meaning*.

  6. David Marjanović said,

    January 12, 2009 @ 7:44 pm

    Concerning the Greek e/i problem, I had completely forgotten that Steven R. Fischer, in his deciphering of the Phaistos Disk, finds what he calls Minoan Greek to have differed from Mycenaean Greek by not only an a/o but also an e/i flip-flop (for a very well understood vowel flip-flop, see here). However, why anyone in their right mind would import their word for "horse" from Crete is beyond me.

    a contact explanation is possible, whereby this semantic innovation began in one branch of Indo-European and thence spread to other branches (perhaps with the spread of domesticated horses), without different reflexes of *EQUOS being actually borrowed from one branch to the next of course.

    Very good point.

    Does anyone know what the Armenian word for "horse" is? The Armenian Wikipedia doesn't have an article on "horse", so I can't find that out easily.

  7. language hat said,

    January 13, 2009 @ 10:09 am

    The Armenian word for 'horse' is ձի /dzi/, gen. ձիոյ /dzioy/; Meillet says it's related to Sanskrit hayaḥ.

  8. Brian M. Scott said,

    January 13, 2009 @ 5:00 pm

    What about German würgen "strangle"? No idea where the vowel comes from, though.

    I-umlaut: the word is cognate with OE wyrgan 'to strangle' and must reflect PGmc. *wurgjan.

  9. Dušan Vukotić said,

    January 13, 2009 @ 7:20 pm

    The Hittite word for 'wheel' is ḫūrkis, and it resembles Tocharian A wärkänt and Tocharian B yerkwantai (oblique case; the nominative is not attested) enough to make us want to derive them from the same inherited root. The obvious choice is the ancestor of Vedic Sanskrit vr̥j- 'twist' (aorist 3sg. injunctive várk, subjunctive várjati, etc.), and it's straightforward to reconstruct a PIE root *h2werg- 'turn'.

    It seems that there is something wrong with the PIE root *h2werg-. Following this root in Slavic languages we are getting the prefixed forms of the verb vrteti (Russ. за-вертеть; Cz. z-vrat 'turnaround'; Serb. za-vrteti 'rotate', 'spin'); OSl врьтѣти from the PIE root *werg-/*wert-. Paul S. Соhеп (Nostratic Centennial Conference: the Pecs Papers. Pecs, 2004. Pp. 51-62.) compared English quirk and Hittite ḫūrki- 'wheel'. I believe (maybe wrong) that Hitt. ḫūrkis is related to Greek κρίκος (ring), Latin circus, Germanic h/ring- and Slavic krug (circle; Russ. круг, Cz, kruh; hence 'car'; all from PIE *kirk-. On the other hand, English quirk might be in relation with Latin converto, Russ. оборот (turn; from *habar-ot), Serb. kovrt (turn), kovrča (curl) and zvrk (whirligig). Can we talk here about the root *h2werg- or about the H- prefixed PIE root *werg-; i.e. in this specific case*h2-werg-?

  10. David Marjanović said,

    January 13, 2009 @ 9:17 pm

    must reflect PGmc. *wurgjan.

    Oh, so it's yet another fossilized causative, and the non-causative version is extinct because it meant just about the same thing. (How do you "cause [something] to be twisted"? You just take it and "twist" it.) I see.

    compared English quirk and Hittite ḫūrki- 'wheel'.

    How does that work? Does the Hittite ever correspond to a PIE plosive or in fact anything other than PIE *h2?

  11. Dušan Vukotic said,

    January 14, 2009 @ 2:11 am

    Oh, so it's yet another fossilized causative, and the non-causative version is extinct because it meant just about the same thing. (How do you "cause [something] to be twisted"? You just take it and "twist" it.) I see.

    I think, at this place it may be interesting to mention the Serbian syntagm zavrnuti vrat (to wring the neck, strangle). Again, there is in question the Slavic verb *vert-et (or *ver(g)n-ut) in its za- prefixed form; even the Serbian noun vrat (neck) comes from the same root. In addition, the Serbian verb brinuti (worry) is obviously connected to the verbs obrtati, vrteti (turn, twist) in the same way as OE wyrgan is related to English worry; hence also the relation between za-brinuti se (to get worried) and za-vrnuti (to twist, to suppress, turn off).

  12. Dušan Vukotić said,

    January 16, 2009 @ 4:21 am

    But let us imagine that the original meaning was less definite, perhaps "large quadruped" or the like (the meaning "donkey" of the Armenian reflex is worthy of notice in this context).

    Or might it be more definite?

    [(myl) The rest of this interesting ~2,000 word essay on metonymy, and other reasons why approximate semantic and phonetic similarity are not reliable sources of evidence, has been moved here, in keeping with our comments policy. In general, if you have something to add to the discussion in our comments section that doesn't fit within a single screen, please post it elsewhere and link to it with a summary.]

  13. Shackled Cattle « Xur-Bel-Gon said,

    January 16, 2009 @ 6:17 am

    […] But let us imagine that the original meaning was less definite, perhaps "large quadruped" or the…). […]

  14. Dušan Vukotić said,

    January 16, 2009 @ 8:06 am

    I see… You are absolutely right. I didn't want to disturb anyone. I was writing those lines right here on your blog and I was not aware of the text length which appeared to be inappropriate for this commenting page. I realized it when I pushed the submit button, but then it was to late.
    Please, accept my apology.

    Dušan Vukotić

    [(myl) It's not a problem, and your comments are appreciated.]

  15. The Prehistoric Mother Tongue (minicast) | A Way with Words said,

    September 15, 2009 @ 4:26 pm

    […] Language Log posts in the order they should be read: The Linguistic Diversity of Aboriginal Europe Horse and wheel in the early history of Indo-European More on IE wheels and horses […]

RSS feed for comments on this post