Of precious swords and Old Sinitic reconstructions, part 2

« previous post | next post »

Part 1 in this series was posted here on 3/8/16 and dealt with a sword called Mòyé 鏌鋣 / 莫邪.  The post was followed by a vigorous discussion that revealed the existence of a large number of words for "sword" in other languages that sound like the reconstructed Old Sinitic form (roughly *mˤak-ja or /makzæ/), stretching westward across Eurasia.  Surprisingly, such words were found prominently in Slavic and Finno-Ugrian languages, but these were determined to be of Germanic origin.  There were also parallels in Caucasian languages.  All of this strongly suggested the possibility that further research along these lines would be rewarding.

Naturally, I hope that we will be able to make further breakthroughs in determining precisely how it could be that the name of this ancient (well over two milliennia ago) Chinese sword, Mòyé 鏌鋣 / 莫邪, could resemble so many west Eurasian words for "sword" and related terminology.  Greater attention to archeology, metallurgy, history, and so forth will undoubtedly be helpful in that regard.

Now I'd like to try the experiment again with another Chinese sword from antiquity whose name is clearly a transcription:  Jìnglù 徑路.  The normal meaning of jìnglù 徑路 is "path(way); shortcut", and it had this meaning already in ancient texts dating back to the I ching (1st millennium BC).  But jìnglù 徑路 also has another meaning, and that is the name of a type of sword favored by the Xiongnu 匈奴 (Old Sinitic /qʰoŋ.nˤa/, Wade–Giles Hsiung-nu, cf. Sogdian xwn]), a large confederation of Eurasian nomads who dominated the Asian Steppe from the late 3rd century BC to the late 1st century AD.  Their name is probably related to that of the later Huns (1st-7th c.) and perhaps also that of the Huna (5th-6th c.), but the composition of the people who made up the latter two groups — judging from historical, archeological, and physical anthropological sources — had changed by the time they ranged respectively as far west as Europe and the northwestern part of South Asia.

Among the Xiongnu, Jìnglù 徑路 was the name of a god whom they worshipped, so we may think of Jìnglù dāo 徑路刀 as "numinous / spirit-like sword".  Thus, although both are transcriptional names of precious swords in ancient China, the situation regarding Jìnglù 徑路, which is an epithet applied to a sword, is rather different than that regarding Mòyé 鏌鋣 / 莫邪, which essentially means "sword".

I recall reading somewhere long ago that jìnglù 徑路 is a transcription of an Iranian word for a type of dagger or short sword whose Greek form is akinakes (ἀκινάκης; cf. unattested Old Persian *akīnakah) and that its Turkic analog was supposedly kingrak, but I can't remember where I read that.

An alternative way of writing Jìnglù 徑路 as the name of a precious ancient sword is Qīnglǚ 輕吕.  On the surface, qīnglǚ 輕吕 doesn't really mean anything sensible:  "light pitch-pipe / musical note".  From the contexts in which it occurs (see here and here), however, we know that it must mean a sword of some sort.

Next step:  determining the Old Sinitic reconstructions of jìnglù 徑路 and qīnglǚ 輕吕.  As I pointed out in the first post of this series, there are a number of reconstructions of Old Sinitic.  Last time I cited the reconstructions of Baxter and Sagart and of Jonathan Smith.  This time I will cite the "minimal" reconstructions of Axel Schuessler.  For jìnglù 徑路 Schuessler gives / kêŋh râkh / and for qīnglǚ 輕吕 he gives / keŋ(h) raʔ /.

As I mentioned before in the first post in this series, if we have multiple written forms for a term that superficially do not make any sense when we interpret the characters one at a time, then we can suspect that they are serving to transcribe a borrowing.  As was the case with Moye, it is also in all likelihood true of jìnglù 徑路 / qīnglǚ 輕吕

As I was poking around and about to wrap up this post, by chance I came upon this blog post:

"Kingrak — 'the oldest Turkish word on record'" (4/7/13)

The post is an extended quotation from Ancient History of China to the End of the Chóu Dynasty (1908), pp. 65-70, by the great German-American Sinologist, Friedrich Hirth (1845-1927).  Since I did read Hirth once upon a time, this must be where I first encountered the Jìnglù dāo 徑路刀 ("numinous / spirit-like sword") of the Xiongnu.

Although I have enormous respect for Hirth, as I do for many other of my Sinological forefathers, I disagree with him in referring to the language in question as "Turkic", since I do not believe that there is any evidence by means of which we can document the existence of the Turkic peoples before about the 5th c. AD.  Of course, many words from the Xiongnu and other early nomadic groups during the first half of the first millennium AD must have fed into the vocabulary of the emerging Turkic ethnos.

Naturally, that leads us to wonder what type of people the Xiongnu were and what sort of language they spoke.  As for what kind of people the Xiongnu were, one can do no better than to consult the excellent book by the Austrian-American scholar, Otto J. Maenchen-Helfen (1894-1969), The World of the Huns:  Studies in Their History and Culture (Berkeley: University of California Press, 1973).  As for what sort of language the Xiongnu spoke, we only have a few words to go by, one being jìnglù 徑路 / qīnglǚ 輕吕, whose roots and affinities we have just today begun to look at seriously.

The other probable Xiongnu word that I can cite off the top of my head is the one that ultimately yielded English "horde" (though most etymological dictionaries only trace it back to Turkic languages).  In a Language Log comment nearly three years ago, I wrote about it thus:

Positing a steppe origin for the word "horde" makes a lot of sense, since it ultimately goes back through Mongolian orda, a sociopolitical and military organizational structure, to a Xiongnu (parallel to, but not identical with "Hun") word referring to the camp or headquarters of a chieftain.

Aside from yielding the word "Urdu" ("language of the military camp"), this very ancient steppe term is also evident in the name "Ordos" (a desert in Inner Mongolia).

A few more strictly phonological notes.

Axel Schuessler has written a magisterial review of Baxter and Sagart in Diachronica, 32.4 (2015), 571-598.

For another review of Baxter and Sagart, see Ho Dah-an “Such errors could have been avoided:  Review of Old Chinese:  A New Reconstruction”, Journal of Chinese Linguistics, 44.1 (January, 2016), 175-230.

Finally, Chris Button, who has his own reconstruction of Old Sinitic (see here, here [pdf], and here [pdf]), kindly contributed the following note on jìnglù 徑路

This one is in Bailey's Khotanese Texts volume. The Old Chinese reconstructions would be 徑 *kjáŋs and 路 *ráks . During the course of the Han period the -s final (written in superscript) would shift to -h with the concomitant loss of the stop coda -k in 路. The -ja- in 徑 would often have been articulated closer to -e- (as the front vowel hypothesis reconstructs it) but underlyingly it would have been -ja- and it is this -j- glide that is carried over to the coda in Middle Chinese. Interestingly it is the inability of non -velar codas to support a -j- glide that accounts for their convergence in rhyming categories such that Baxter artificially separates -am and -jam into two rhymes -am and -em where traditionally there was just one, whereas -aŋ and -jaŋ (Baxter's -aŋ and -eŋ) have always been separated traditionally.

One of my next posts will use language to document Eurasian connections via hats, but in a quite different way from these two on ancient swords and Old Sinitic reconstructions.

[Thanks to Edward Shaughnessy, Tsu-Lin Mei, and Michael Carr]


  1. julie lee said,

    March 12, 2016 @ 8:26 pm

    Fascinating post, thanks.
    A bit off-topic: As you mention the Huns (Xiongnu 匈奴 in Mandarin, read like English Shiung-nu) I thought of a recent encounter with a young Turk, a visiting scholar from Turkey. He told me the Turks were descended from the Huns. I'd always pronounced "Huns" to rhyme with English "puns". But he pronounced "Huns" as English "hoons". Suddenly I realized the "hoons" must be the Shiung-nu (Xiongnu) of Chinese history, because this word (匈奴)is pronounced Hoong-nu in Cantonese, which has a more ancient pronunciation than Mandarin. I think it was Professor Pulleyblank who first proposed the Xiongnu to be the Huns (rhyming with "puns"), but I wondered about it because the sounds were so different. (Though years before Pulleyblank I'd heard a Chinese historians say that the Xiongnu were responsible for the Fall of the Roman Empire.) And English-language history books still tell us that the Huns "may" have been the Xiongnu of Chinese history. Obviously "Hoon" and "Hoong-nu" are similar in sound. Since the Chinese script is made up of syllables, not vowels and consonants, and there was probably no "hoon" syllable in the Chinese language used by the ancient Chinese transcriber, he would have used the syllable "-nu" to suggest a syllable ending in "-n". Similiarly, Chinese newspapers today transcribe the name "Trump" as 川普 Chuan-pu, pronounced like English Trwan-pu, since there is no syllable ending in "-m" or "-mp" in the Mandarin syllabary.

  2. JS said,

    March 13, 2016 @ 12:24 am

    Just to be clear let me shamefacedly point out that I don't have any Old Chinese reconstructions to speak of — I offered an off-the-cuff form for Moye 莫邪 last time that included *z-, bad as (1) this would apply if anything to the homograph xie2 邪 'evil; crooked', not to ye2; and (2) even there it is an outdated suggestion for the Old Chinese stage. Both xie2 and ye2 are now often seen as from *l-; some (like Baxter and Sagart) would add *ɢ- as a source of the latter initial.

    Nothing really substantial separates the reconstructions for jing4lu4 徑路 above except for Schuessler's main vowel *-e- in jing4 as suggested by the "front vowel hypothesis" versus Button's (phonemic) *-ja-; I am interested to know why he has chosen this older direction.

  3. Victor Mair said,

    March 13, 2016 @ 1:03 pm

    From Elizabeth J. W. Barber:

    I've long been intrigued by the European medieval tradition of naming fine swords as they came off the forge–as though the sword had received a soul, which could turn the weapon to good or evil in its lifetime. That is, it had an animate-like potency. Then I noticed that other objects received names, like the horn Oliphant and the unbreakable fetter Gleipnir. But name-receiving objects were clearly always things that DO something. (I summarized all of this in our book WHEN THEY SEVERED EARTH FROM SKY, 97-102, but especially p. 100.) Now the most interesting point for you: The earliest representation of an "animate sword" is Hittite, at Yazilikaya (photo on p. 101), with a head wearing a deity-hat sticking out of the top of the hilt! Any trace of sword-souls in your eastern Eurasian material??

  4. languagehat said,

    March 13, 2016 @ 4:18 pm

    Hey, Elizabeth Barber taught me linguistics and was the first to set me on the path to a graduate degree in Indo-European linguistics. Hi, Dr. Barber, fancy meeting you here!

    Also, another fascinating post, and I can't wait to see what connections turn up.

  5. Victor Mair said,

    March 13, 2016 @ 5:52 pm


    Then you must have gone to Occidental College, where Jim Mallory went and C. Scott Littleton taught, not to mention one Barack Obama who was a student there for awhile.

    Incidentally, I will be staying with Betchen this weekend. We have a lot of catching up to do.

  6. Victor Mair said,

    March 13, 2016 @ 5:54 pm

    From Chris Button:

    @ JS

    In response to the question about reconstructing ja rather than e, the short answer is that the “front & rounded vowel hypothesis” reflects a tendency in Shijing rhyming that has unfortunately been foisted upon reconstructions of Old Chinese. To be more specific, the hypothesis has broken up traditional rhyme categories such that a single rhyme group like 元 is now split into three separate rhymes -an, -on, -en. The problem is that -an, -on, -en could still rhyme together in Old Chinese but had a tendency to rhyme separately in the Shijing. The reason for this is because they are fundamentally -an, -wan, -jan which all share a common -an rhyme while differing under the lack or presence of a labial or palatal prosody. There are two pieces of evidence that support this:

    Rhymes with velar codas like ­-ak, -wak, -jak (or -ak, -ok, -ek according to the hypothesis) are actually kept apart in the traditional rhyme groups too. That is to say, unlike -an, -wan, -jan which could rhyme together in Old Chinese, -ak, -wak, -jak could not rhyme together. This is due to the greater ability of velar codas to maintain a palatal or labial articulation than other codas. Pulleyblank actually treated -wak and -jak as -akw and -akj which, without wanting to get too specific, can broadly be considered a transcriptional difference. This is not an option for ­rhymes like -wan and -jan.

    The front & rounded vowel hypothesis causes a proliferation of vowels in related characters written with the same phonetic. This confuses the basic Old Chinese ə/a alternation. For example, 定 -jaŋ is phonetic in 綻 -jəŋ, 舟 -əw is a deformed phonetic in 朝 -aw, and 覃 has two readings -əm and -am. However, the front & rounded vowel hypothesis gives us 定 -eŋ as phonetic in 綻 -in, 舟 -u as phonetic in 朝 -aw, and yet keeps 覃 as -əm and -am.

    What the front & rounded vowel hypothesis has done to the Old Chinese vowel system is broadly similar to what was done to Indo-European. Indo-European was reconstructed down to a single e vowel with a variant o. Several scholars have since pointed out that this is really ə with a variant a. Old Chinese similarly goes back to ə with a variant a. However, since the notion that there is really only one “vowel” ə which sometimes surfaces as a, implies that vowels don’t really exist, people have been rather reluctant to accept such a proposal. Presenting this to most linguists is like standing in front of a room full of biologists and telling them that DNA doesn’t really exist. As a result people have crammed Indo-European and Old Chinese full of other vowels to make them both appear more typologically “normal”. I have discussed this topic quite extensively here:


  7. languagehat said,

    March 13, 2016 @ 6:45 pm

    Then you must have gone to Occidental College

    I did indeed, class of '72 (well before the president's day).

  8. Chris Button said,

    March 13, 2016 @ 8:45 pm

    I should probably add that my comment regarding Pulleyblank technically represents an amalgamation of two of his versions of OC:

    – Pulleyblank (1977-8) had -akw and -ac
    – Pulleyblank (1991) had -akɥ (labiopalatal!) and -akj, with -akw then being reassigned elsewhere.

    Needless to say the labiopalatovelar -kɥ seems a little far-fetched with four kinds of velars (plain, labial, palatal, labiopalatal) being too much for any system.

  9. Victor Mair said,

    March 13, 2016 @ 11:47 pm

    From Marcel Erdal:

    Kâshgharî fol 611 has qïrŋaq 'a broad knife, like a cleaver used to cut meat or dough'. I have no doubt that this is a metathesis of qïŋraq, as a reader, who made a note in the margin of the ms., already thought. Metatheses are frequent in clusters with /r/ and this is a rather difficult one as it is. The -ŋr- is quoted by Clauson from two modern languages and could, I think, be found in further Siberian languages. Note also kiŋgar 'faucille' = sickle in Ordos Mongolic, kiŋgara 'large knife (often with two blades), chopping knife, cleaver; harvesting knife, sickle' in Written Mongolian (Lessing). Ramstedt, Kalmuck dict. p. 234b mentions kîre with, among others, the translations "schlechtes Messer, Dolch" and some Turkic and Mongolic cognates. So this, like ordo, is a word found in both Turkic and Mongolic. The proto-form is sure to be qïŋraq because Mongolic – in view of the Written Mongolian form – often drops word-final K when compared to Turkic.

    Altay Turkic has the noun kïŋrak denoting a 'steel knife used for working leather'. Note that South Siberian has lots of Mongolic loan words but that this cannot be one, as the Mongolic contact languages had all lost the coda

  10. Victor Mair said,

    March 13, 2016 @ 11:48 pm

    From Jens Østergaard Petersen:

    Desultory notes made after skimming a bunch of articles.

    I have checked to see whether there have been similar thoughts before, mainly by searching on NKI.net. No IE connections seem to be postulated by Chinese scholars, but an interesting line is developed that connects our famed swords to the fabled 貘. This beast had a predilection for licking iron, and in stories associated with it, the refined iron from its gall and liver served as material to forge Moye and Ganjiang (one is here reminded of the story of the Wu queen who embraced a pillar and gave birth to the iron that was used to fashion the swords: here iron is obviously talked about as well, not bronze). Exactly which animal is referred to is uncertain, but perhaps some sort of bear or panda (or leopard) is at play. The idea that 貘 is dimidiated as 莫邪 is expressed and some possible ST cognates are mentioned.

    The consensus seems to be that 干將 means 吳匠 (干 being similar to 勾 in 勾吳), just as 歐冶 means 越冶, so no cognates should be sought for them.

    As far as I can see from Google Books, Donald Wagner does not discuss these problems in his Iron and Steel in Ancient China (1993), nor in his The Language of the Ancient State of Wu (1990).

    戴月舟, 干将莫邪研究. MA thesis, 南京师范大学 2008.
    刘付靖, 汉藏语言与 《山海经》 的若干怪兽名称考释; 广西民族研究 2002 年第2 期, 54-57.
    李道和, 干将莫邪传说的演变, wenku.baidu.com/view/e63abcec551810a6f524863e.html.

    I have just skimmed these. I will continue to read about this topic and alert you if I find more of interest.

    You will find Blust here. Christoph Harbsmeier just sent it to me and I have only started reading it.

    The 干將莫邪 articles you can find here.

    Donald Wagner book is here.

    The article he refers to as 1990b you can find here. The details are Donald B. Wagner, "The Language of the Ancient State of Wu”, in The Master said: To study and … : To Søren Egerod on the Occasion of His Sixty-Seventh Birthday, ed. by Birthe Arendrup et al. (Copenhagen: East Asian Institute, 1990), 161-176.

    I guess I should have written 句吳, not 勾吳.

  11. Victor Mair said,

    March 13, 2016 @ 11:49 pm

    From Étienne de La Vaissière:

    As regards Kenglu / xiŋgār/ qïŋïraq/ Akinakès : Pulleyblank, 1962 p. 222.

    As regards the link between this word and Khingila : Tremblay 1999, p. 182-4.

    As regards the common cult of the Sword in the Hunnish and Xiongnu world : my article Is there a nationality of the Hephtalites p. 129 (available at:
    https://www.academia.edu/1476531/Is_There_a_Nationality_of_the_Hephtalites_ )

    From Mehmet Olmez:

    Hirth and others' materials, notes collected also by Talat Bey (in Turkish)

    Hunların Dili (T → The Language of the Huns), Doruk yayınevi. 33, Ankara 1993.

    Kanjis can not be read so well but it must be : 徑路 jinglu

    Boodberg also has the same (1936) characters

  12. Victor Mair said,

    March 13, 2016 @ 11:51 pm

    From Martin Schwartz:

    Re a Turkic form of the shape *kingra-, I would suggest it perhaps not go with Sogd. 'kynK vel sim. = akînak, Gr. akinakEs, but rather with Sogd. xVnG(a)r, (G = gamma) Perso-Arab. xanjar 'dagger'.

    As for OPers. magu-, while this, via Gr. mágos, mageîa, is the etymon
    of magic, magu- is not magician, but a member of a priestly tribe (phúlE) as per Herodotus. Benveniste, in his little book on les mages, got it right in showing there was an Avestan word mouGu- < */magu-/ 'tribe', so that West Iranian magu- was the tribe par excellence, cf. Kohen-s and Levites.The question then becomes what the etym. of *magu- as tribe is. In Fs. Boyce is suggest maybe kinship (via marraige?) as a gift-exchange exchange institution, or somehow other gifting.

  13. JS said,

    March 13, 2016 @ 11:55 pm

    @Chris Button
    Thanks for the detailed notes; I am enjoying reading the manuscript!

RSS feed for comments on this post