Once more on Sinitic *mraɣ and Celtic and Germanic *marko for "horse"

« previous post | next post »

Jessica Hemming, in consultation with Joseph Eska (personal communication), writes:

In the debate about whether Sinitic ‘mra’ could be a borrowing from an Indo-European language, given that only Celtic and Germanic have horse words in *marko, it may be of use to know that proto-Celtic is now conventionally dated to no earlier than c.1000 BC and proto-Germanic to c.500 BC. These would seem to be too late to be options. Also, the native word for horse in Celtic is *ekwo; nobody is quite sure where *marko came from, although later it became the standard medieval word for horse (especially in the sense of ‘war horse’ or ‘steed’ in Middle Welsh).

I checked with one of my CSANA* colleagues, linguist Joe Eska, this morning and he says: ‘As you say, *marko- is only known in Celtic and Germanic — though that doesn't mean it wasn't in Tocharian or Indic (though you'd have to say that they were lost in those languages after having loaned to Chinese, not a compelling position to hold). Also, *marko- seems to have been borrowed early in Celtic from some unknown source.  It's not attested in Celtiberian — though *ekwo- is — but that could be owing to accident of lack of attestation.  So we can't be sure that it was borrowed into proto-Celtic.  It may have been after Celtiberian separated from proto-Celtic.’   So, this may suggest that the *marko is even later.

*[VHM: Celtic Studies Association of North America]

In the past, I have cited the following Old Sinitic reconstructions from the Wiktionary article on mǎ 馬 ("horse"):

(BaxterSagart): /*mˤraʔ/

(Zhengzhang): /*mraːʔ/

and added:

Schuessler (2007) /*mrâʔ/

I do not recall previously having taken notice of this etymological note in the Wiktionary article:

“Horse” – from Proto-Sino-Tibetan *k-m-raŋ ~ s-raŋ. The sense of “big” is derived from “horse”; compare the English uses of horse.

Incidentally, Jessica Hemming originally wrote to me in connection with the dating of the invention of stirrups, so if anyone has any information — especially recent scholarship — on this subject, please do not hesitate to add it to the discussion thread.

 

Selected readings



18 Comments

  1. Laurent Sagart said,

    April 28, 2020 @ 8:34 am

    In the supplementary material to our recent paper on Sino-Tibetan (https://www.pnas.org/content/pnas/suppl/2019/04/30/1817972116.DCSupplemental/pnas.1817972116.sapp.pdf)
    we argued that the Chinese word for 'horse' was borrowed from a more westerly Sino-Tibetan language. We think a connection to IE is fanciful.

  2. Pamela said,

    April 28, 2020 @ 9:00 am

    so the borrowing was the other way around? or both came from an unknown third source?'

    as for stirrups –mountains written on this. hemp or leather stirrups (for resting the legs) could be about as old as proper (treed) saddles, but they are not depicted nor found in tombs. iron stirrups are generally agreed to be of Central Asian origin (like much technology related to iron, including chain mail), 5th-6th-no later than 7th century. as I have argued in several publications, (Hammer & Anvil most recent), stirrups are not terribly important –changes in saddle construction are. I have to remind students –stirrups are suspended from the saddle.

  3. Pamela said,

    April 28, 2020 @ 9:06 am

    So am I understanding that the common attribution of mar/marko etc to "Indo-European" is not precise, and that this should be something like" proto-Celtic/proto-Germanic"? or, is the hypothesis that there was an Indo-European root that has survived/been evinced only in proto-Celtic and proto-Germanic?

  4. Andreas Johansson said,

    April 28, 2020 @ 11:42 am

    @Pamela

    As Jessing Hemming writes, it's possible the word goes all the way back to PIE, but only survived to be attested in Celtic and Germanic.

    If it seems likely to be an early loan in Celtic, though, the obvious scenario would be it's a Germanic innovation (or loan from a non-PIE language) that was then borrowed into Celtic.

  5. Robbie said,

    April 28, 2020 @ 2:09 pm

    Two random and completely un-expert thoughts:

    It struck me that ma/mra/mrh sound like they could be onomatopoeic for the sound of a horse. If so, something like that word could have arisen spontaneously twice, once in China and once somewhere in Europe.

    Horses are big, powerful animals, and financially important too. Is it possible that *marko supplanted the older word *ekwo as a taboo name?

  6. Chris Button said,

    April 28, 2020 @ 5:24 pm

    @ Laurent Sagart

    we argued that the Chinese word for 'horse' was borrowed from a more westerly Sino-Tibetan language. We think a connection to IE is fanciful.

    I would actually argue the opposite. I think Tibeto-Burman *raŋ (be it the *s-raŋ or *k-m-raŋ as cited in the OP and depending on the prefix) is far harder to connect with Old Chinese *mráɣʔ than a putative Proto-Indo-European form.

    @ Andreas Johansson

    As Jessing Hemming writes, it's possible the word goes all the way back to PIE, but only survived to be attested in Celtic and Germanic

    Yes, and I don't see that as a major stumbling block when combined with good archeological evidence and a linguistically reasonable association with Old Chinese *mráɣʔ

    @ Victor Mair

    Once more on Sinitic *mra and Celtic and Germanic *marko for "horse"

    I don't think the title does justice to the Sinitic form by omitting the final velar component /ɣ/ perhaps surfacing as [ɰ]. We have the same situation with 巫 *màɣ with which I believe you have convincingly compared with Old Persian "maguš".

    I do not recall previously having taken notice of this etymological note in the Wiktionary article:

    "Horse" – from Proto-Sino-Tibetan *k-m-raŋ ~ s-raŋ. The sense of "big" is derived from "horse"; compare the English uses of horse.

    Well there is Burmese မြင့် "high, tall" which differs from မြင်း "horse" in tone, but I don't know if anyone's ever suggested an etymological link there.

  7. Victor Mair said,

    April 28, 2020 @ 8:15 pm

    @Chris Button

    I don't think the title does justice to the Sinitic form by omitting the final velar component /ɣ/ perhaps surfacing as [ɰ].

    I completely agree with you. I was conflicted over how to handle that problem because, in the first line of the quotation, Jessica Hemming just gave *mra, and I didn't want the title to contradict what she wrote. In retrospect, I should have asked her if she wouldn't mind my adding the velar. I have done so now.

  8. David Marjanović said,

    April 29, 2020 @ 7:42 am

    so the borrowing was the other way around? or both came from an unknown third source?

    The idea here is that either it's a chance similarity or parallel borrowings from an unknown third source. What such a source could be is a good question – it's too early for Avars or Huns. (Specifically, the Germanic word shows the effect of Grimm's law, which had run its course by the time the Romans showed up, and probably quite a bit earlier than that.)

    So am I understanding that the common attribution of mar/marko etc to "Indo-European" is not precise, and that this should be something like" proto-Celtic/proto-Germanic"? or, is the hypothesis that there was an Indo-European root that has survived/been evinced only in proto-Celtic and proto-Germanic?

    Four options:

    1) The word was present in Proto-Indo-European, but lost without a trace in every single branch except Celtic and Germanic. This is definitely not the most parsimonious hypothesis, but it's also definitely not impossible either. And keep in mind that, while the word has not been found in Tocharian, that could have any number of reasons. It definitely wasn't the default, neutral word for "horse" in either Tocharian language, but perhaps it had acquired a very specialized sense or had become pejorative or was for some other reason unsuitable for use in the Tocharian texts we have (most of which are translations of Buddhist classics). Compare Gothic: if a word is not attested in Gothic, that could mean it wasn't there, or it could mean it wasn't suitable for use in a Bible translation.
    2) It is possible that Germanic and Italo-Celtic (not Celtic alone!) are each other's closest known relatives. If so, the word could have been present in "Proto-West-IE" without being inherited all the way from PIE. It could then be a borrowing dating from, I would say, fairly early in the Bronze Age.
    3) The word was not present in PIE or PWIE, but somehow appeared in Celtic and was passed on to Germanic (Proto-Germanic has a good number of Celtic loanwords). "Somehow" could of course mean a borrowing from somewhere east, but we lack concrete candidates.
    4) The word was not present in PIE or PWIE, but somehow appeared in Germanic and was passed on to Celtic (Proto-Celtic has a few possible Germanic loanwords). "Somehow" could of course mean a borrowing from somewhere east (Celts got very far east around 300 BCE, definitely into Ukraine, not unlikely into Russia), but again we lack concrete candidates.

    As Jessing Hemming writes, it's possible the word goes all the way back to PIE, but only survived to be attested in Celtic and Germanic

    Yes, and I don't see that as a major stumbling block when combined with good archeological evidence and a linguistically reasonable association with Old Chinese *mráɣʔ

    I don't understand what you mean by "good archeological evidence" when you refer to a word, rather than to the horses themselves.

    (…By the way, given that two of the top experts are present in this thread, I personally would love to see a discussion of "the final velar element" so I can form an opinion about it.)

  9. David Marjanović said,

    April 29, 2020 @ 8:14 am

    Incidentally, this is as good a place as any to post the latest list of possible loanwords from Tocharian into Chinese. It's quite a bit longer than the second-to-latest list (referred to as "Židek" and "Židek's thesis"). Unfortunately, sources are only alluded to, not cited.

  10. Andreas Johansson said,

    April 29, 2020 @ 9:19 am

    With apologies to Jessica Hemming for mangling her name above …

  11. Chris Button said,

    April 29, 2020 @ 8:01 pm

    @ David Marjanović

    The list you mention appears to be from Lubotsky (1998) "Tocharian Loanwords in Chinese".

    We've discussed the evidence for the final velar fricative/approximant before on LLog. The problem isn't the evidence for it. Rather, the acceptance of it throws the symmetry of many reconstructions out of whack (i.e., if it occurs here, then surely it should occur here too).

    As for the archaeology, try Prof. Mair's "The Horse in Late Prehistoric China: Wresting Culture and Control from the 'Barbarians'." (2003)

  12. David Marjanović said,

    April 30, 2020 @ 5:48 am

    As for the archaeology, try Prof. Mair's "The Horse in Late Prehistoric China: Wresting Culture and Control from the 'Barbarians'." (2003)

    Well, I can't, it's only available on paper. Can't you summarize how it can tell us anything about the word *mark-?

  13. Victor Mair said,

    April 30, 2020 @ 9:31 pm

    As for the archaeology, try Prof. Mair's "The Horse in Late Prehistoric China: Wresting Culture and Control from the 'Barbarians'." (2003)

  14. Victor Mair said,

    April 30, 2020 @ 9:34 pm

    From an authority on ancient Celtic textiles:

    I was quite surprised that someone said Proto-Celtic is now dated at about 1000 BC. The textile development of the Hallstatt Culture, clearly the immediate and often indistinguishable ancestor of the La Tene Celts, goes back with its interesting idiosyncrasies at least to stuff dated 1300 BC in the Austrian Alps. Don't know if 300 years makes any difference to these arguments.

  15. David Marjanović said,

    May 1, 2020 @ 8:31 am

    As for the archaeology

    My original question was: "I don't understand what you mean by "good archeological evidence" when you refer to a word, rather than to the horses themselves." So, are you saying I was referred back to the horses themselves, which is not what I asked about?

    I was quite surprised that someone said Proto-Celtic is now dated at about 1000 BC.

    Generally, what people mean by such a statement is that that's the date when the attested Celtic branches began to diverge. Older – Pre-Proto-Celtic – stages could of course have been spoken much earlier.

    Sometimes, though, people instead mean the date of a particularly conspicuous innovation. For instance, the often-quoted date of 500 BC for Proto-Germanic is clearly not when East Germanic and Northwest Germanic became distinguishable; that's more like 100 or 200 AD! It's a guess at when Grimm's law might have happened; and some sources make this explicit.

  16. Nelson Goering said,

    May 2, 2020 @ 4:37 am

    Generally, what people mean by such a statement is that that's the date when the attested Celtic branches began to diverge. Older – Pre-Proto-Celtic – stages could of course have been spoken much earlier.

    This is a point of frequent confusion in historical linguistics, unfortunately. Some people use 'proto' to mean some kind of essentialized stage when the language took on its essence as that language (this is a thoroughly unscientific approach, and should be avoided, but it still shows up now and then due to the romantic legacies of much historical linguistic research), others use it to refer to a long phase (in which case it has to be dated with a range, not to a point), and finally, in what I think is the most sensible usage, it can mean the MRCA ('most recent common ancestor') before the proto-language diverged into later branches.

    For dating a proto-language, archaeology is usually only very roughly helpful. A clearly reconstructible term (or better, set of terms) can be a good time-after-which — but 'after' can be long after, potentially. Arguments of silence can sometimes be used to get an approximate time-before-which (people make this argument with regard to metals, for instance, in dating PIE to the Bronze Age), but this has to be done extremely cautiously, since there are other reasons why general vocabulary loss or replacement in a particular area might occur.

    I recently posted about some of the basics of 'linguistic palaeontology' in my History of English series: https://historyofenglish.joomla.com/index.php/14-12-wheel

  17. Nelson Goering said,

    May 2, 2020 @ 4:38 am

    (I seem to have mistyped the code for making the quotation appear as such — the first paragraph should be indented as a block quote.)

  18. David Marjanović said,

    May 7, 2020 @ 8:48 am

    Yes, thanks. It does not help that protos means "the first".

RSS feed for comments on this post