Archeological and linguistic evidence for the wheel in East Asia

« previous post | next post »

The domesticated horse, the chariot, and the wheel came to East Asia from the west, and so did horse riding:

Mair, Victor H.  "The Horse in Late Prehistoric China:  Wresting Culture and Control from the 'Barbarians.'"  In Marsha Levine, Colin Renfrew, and Katie Boyle, ed.  Prehistoric steppe adaptation and the horse,  McDonald Institute Monographs.  Cambridge:  McDonald Institute for Archaeological Research, 2003, pp. 163-187.

We now have newly unearthed archeological evidence for the introduction of wheeled vehicles to East Asia.  Here is a link to the recent discovery of the oldest wheel-tracks at the Pingliang Tai site in Huaiyang, Henan Province 河南淮阳平粮台.

The Pingliang Tai tracks are 4,200 years old and were produced by a two-wheeled cart.  In comparison, the oldest wheels in the West are much earlier.

J. P. Mallory and I discussed some of the evidence for the origins of the Chinese chariot in The Tarim Mummies (London:  Thames and Hudson, 2000), pp. 324-326:

If the chariot came from the West, what about the name for the vehicle? Here there is some linguistic evidence to support such a movement. The Chinese word for chariot, the modern Mandarin ch'e, would have been articulated roughly as *kl^yag† during the Shang dynasty, and this word bears a certain resemblance to one of the Proto-Indo-European words for 'wheel' (*k^wék^wlo) which provided the base for the word for vehicle in Tocharian, i.e. Tocharian A kukäl and Tocharian B kokale. Rather than a direct Tocharian source, it has been suggested that the underlying form may have been some form of early Iranian language. This would hardly be surprising in that the Indo-Iranians perfected chariot warfare and either introduced it or, at least, were so proficient in it that they were the acknowledged masters of chariotry in the Near East. It seems probable, then, that Bronze Age Iranians or Tocharians came into contact with peoples of western China in the 2nd millenium BC and introduced the chariot to the Shang. The venue of the meeting of these two worlds was, naturally, the modern province of Xinjiang and the area just to its northeast. Once introduced, the Chinese began to work their own special technological magic on any western loan. But the West was not only supplying them with vehicles, it was also sending something of its own magic into the very courts of the Shang.

[†In this paragraph, ^ indicates that the following letter is a superscript.]

(the quotation is from p. 326)

See also:

Victor H. Mair, "Old Sinitic *Myag, Old Persian Maguš, and English Magician", Early China, 15 (1990), 27–47.  Available on JSTOR here.  See esp. 45-46.

Robert S. Bauer, "Sino-Tibetan *kolo 'Wheel'," Sino-Platonic Papers 47 (Aug. 1994), 1-11. (free pdf)

chē 車 *kɬàɣ ("chariot"), cf. Greek kýklos, Latin cyclus

In general, if a radical new technology or cultural manifestation is borrowed from another civilization, chances are good that the word for that technological innovation or cultural phenomenon may have come along with the object or idea that it represents.


[Thanks to Peter Kupfer and Chris Button]


  1. Christian Weisgerber said,

    March 12, 2020 @ 2:27 pm

    Initial k is problematic for a borrowing from Iranian. After merging the labialized velars into the plain velars (*kʷ → *k), Indo-Iranian shifted the latter to č before front vowels. The reflexes of PIE *kʷékʷlos are Sanskrit cakrá, Avestan caxra.

  2. Chris Button said,

    March 12, 2020 @ 9:06 pm

    I tend to prefer a Tocharian borrowing too as discussed here:

    Regarding the recent discussion around a *t- prefix discussed over on the Tocharian, Turkic, and Old Sinitic "ten thousand" thread, I note that Baxter & Sagart reconstruct 車 as *[t.qʰ](r)A (with *C.q(r)a for its alternate reading). This takes it away from the likely connection with Proto-Indo-European *kʷ(e)kʷlo- and/or Tocharian kokale/kukäl.

    Personally, although I prefer to derive aspiration of obstruents in OC from an *s- prefix as supported elsewhere in Proto-Tibeto-Burman, I do like Pulleyblank's suggestion (following Shafer's 1950 discussion of the Bodic verb morphology) that aspiration may have resulted from reduplication of the onset (e.g. *k-k- becomes *kʰ-) in this case, given the probable loanword status. In the case of 車 we would then have *k-klàɣ giving *kʰlàɣ hence *kɬàɣ and ultimately *kɬàːɣ via the sporadic lengthening that occurred before the -ɣ coda (the alternate reading would simply be *klàɣ without the reduplicated onset and without any later lengthening of the nucleus). This brings it very close to Proto-Indo-European, particularly when one considers that kʷl- as a cluster would have violated a phonotactic constraint in any case.

  3. David Marjanović said,

    March 13, 2020 @ 7:35 am

    Baxter & Sagart reconstruct 車 as *[t.qʰ](r)A (with *C.q(r)a for its alternate reading)

    Interesting. A uvular [q(ʰ)] seems hard to explain from Indo-European – why did Baxter & Sagart reconstruct it?

  4. Chau said,

    March 13, 2020 @ 2:51 pm

    The Taiwanese word for ‘wheel’ is lián [see A Dictionary of Southern Min (Taiwanese-English Dictionary), p. 169]. This word belongs to the vernacular stratum of Taiwanese, and there is no proper sinograph to write for it. The above-mentioned dictionary gives 輪 (lún) which is just a Chinese translation. Rev. William Campbell’s A Dictionary of the Amoy Vernacular (廈門音新字典) gives 碾 (lián) for a specific word, 石碾 chio̍h-lián, which is a horizontal, heavy stone wheel for grinding grain.

    What is interesting about lián is that it can be linked to OHG rat ‘wheel’ through 3 sound changes because of the phonological structure of Taiwanese (or Southern Min): (1) r- > l- due to the lack of the r sound in Taiwanese; (2) homorganic nasalization of the final -d: -d > -n; and (3) infix of the glide -i-, a process called “vowel bending” by Schuessler. Thus, the correspondence goes as follows: OHG rad > *lad > *lan > lián ‘wheel’.

    Sinitic languages continually renew themselves, and in the process borrow words from foreign sources. The one discussed in O.P. was probably the oldest one, whereas the ancestral Holó language may have borrowed one much later from a Germanic source related to the Old High German branch.

  5. David Marjanović said,

    March 13, 2020 @ 6:33 pm

    homorganic nasalization of the final -d: -d > -n;

    But High German /d/ is voiceless. Why wouldn't the Taiwanese outcome simply be *lat?

    infix of the glide -i-, a process called "vowel bending" by Schuessler.

    What conditions did Schuessler think this process had?

  6. Chris Button said,

    March 13, 2020 @ 8:56 pm

    One correction to my earlier post (not affecting the PIE connection).

    I believe the alternate reading of 車 *kɬàɣ (later *kɬàːɣ) would be
    *kàɣ without the liquid. Had the liquid been present it would have developed much like 墅 *klàɣʔ instead (related to 野 *làɣʔ, later làːɣʔ) minus the glottal.

    @ David Marjanović

    A uvular [q(ʰ)] seems hard to explain from Indo-European – why did Baxter & Sagart reconstruct it?

    Firstly, they clearly don't think it's related to Indo-European.

    Secondly, they want to relate 輿 *làɣ with 車 and reconstruct a uvular for the former. Stuck with the uvular, they then need to slap a random t- prefix on the latter to fit within the confines of their system.

    Incidentally, I think the actual etymological relationship of 輿 *làɣ is with 以 *lə̀ɣʔ and associated words. However the phonological and semantic proximity with 車 no doubt allowed a "folk-etymology" of sorts to confuse the creators of the 輿 graph that also then lead B&S to take its 車 component literally.

  7. Victor Mair said,

    March 15, 2020 @ 12:47 pm

    From Sasha Lubotsky:

    In my contribution to the "Mummies" conference, I argued that the Chinese word for "chariot" [*kl^yag†] corresponds _not_ to Toch.B kokale (which deviates rather strongly from the Chinese form), but from Toch. B klen̄ke, A klan̄k 'vehicle, Skt. yāna-, vāhana-', Toch. AB klān̄k- 'to ride, travel (by vehicle)', PIE *kleng- (cf. Modern German lenken 'to guide, conduct', Wagenlenker 'charioteer').

    Victor H. Mair, ed., The Bronze Age and Early Iron Age Peoples of Eastern Central Asia (Washington, D.C.: Institute for the Study of Man Inc. in collaboration with the University of Pennsylvania Museum Publications, 1998). 2 vols, p. 382.

  8. Victor Mair said,

    March 16, 2020 @ 9:38 am

    From Peter Kupfer:

    At the museum Schloss Gottorf in the far north of Germany (Rendsburg-Eckernförde), I saw the preserved track of a two-wheeled cart from 3400 BC, supposed to be the world’s oldest, found in a place named Flintbeck

  9. Chris Button said,

    March 16, 2020 @ 1:09 pm

    In my contribution to the "Mummies" conference, I argued that the Chinese word for "chariot" [*kl^yag†] corresponds _not_ to Toch.B kokale (which deviates rather strongly from the Chinese form), but from Toch. B klen̄ke, A klan̄k 'vehicle, Skt. yāna-, vāhana-', Toch. AB klān̄k- 'to ride, travel (by vehicle)', PIE *kleng- (cf. Modern German lenken 'to guide, conduct', Wagenlenker 'charioteer').

    I was quite confused by this comment until I took a look at the paper from the conference. The comparison is with 乘, which would have been something like OC *lə̀ɲs (from earlier *ljə̀ŋs), rather than 車.

    The problem with an external association with 乘 is that the meaning "chariot" derives from the original sense of 乘 *lə̀ɲ "mount, ride, ascend". The morphology with the suffixal -s shows the nominal sense to be derived from the verb, and the graphic form depicted in the oracle-bone form leaves us in no doubt that that.

  10. Chau said,

    March 18, 2020 @ 5:55 am

    @David Marjanović
    “But High German /d/ is voiceless. Why wouldn’t the Taiwanese outcome simply be *lat?”

    Thank you for the stimulating question. To answer your question, I spent the last few days combing my notebooks for loanwords from the High German branch with a final -d that gave a final -t as the outcome. (It took me so much time because, I am ashamed to say, I am very poorly organized.)

    As you know, the finals in Taiwanese entering tone (入聲) are all voiceless (-p/t/k). Thus, the Germanic loans originally with a -d final are expected to give a -t in Tw, if they did not undergo further nasalization. For example, OE hȳdan ‘to hide’ (stem hȳd-) > Tw bi̍t 密 ‘hidden, secret, confidential’ as in pó-bi̍t 保密 ‘hide, keep in secret’ [with a sound change of h- > b-]. Therefore, it is natural to expect loans from OHG with a -d final to show up with a -t final in Tw. However, there are two reasons for my long search (besides the one above). (1) Germanic loanwords in Taiwanese pertain mostly to OE, OFris and ON, and the High German branch is a minority. (2) There is a strong tendency for Germanic loanwords with a final -d or -t to undergo nasalization in Tw. For example, OE ġæt/OFris gat ‘gate’ > *gat > Tw koan 關 ‘gate’ such as 嘉峪關 (Tw Ka-io̍k-koan), the gate at the western terminus of the Great Wall. As a consequence, OHG words with a -d final turning up as loans in Tw with a -t final are not easy to find.

    But here is a precious one. OHG parafrid, pfarifrid, pferfrit ‘horse’ (derived from Late Latin paraverēdus ‘extra post horse’), which through MHG pfert gave rise to NHG Pferd ‘horse’, provide a secondary word for ‘horse’ in Sinitic, 匹 . The Taiwanese pronunciation of 匹 is phit, obviously related to the second element of OHG para-frid. (Recall that there was no f sound in OC: 古無輕唇音, so that foreign f > p/ph.) As a secondary word for ‘horse’, it cannot be used alone like 馬, but can be used in a pleonastic as 馬匹 ‘horse’, or as a measure word for horse 一匹馬 ‘one count of horse’, or in a word compound 匹配 ‘to match, e.g., in horse breeding’.

    To the second question regarding “vowel bending” by Schuessler, I went back to his book (ABC Dict. of OC) and cursorily scanned through the book (it has no general index), but failed to find it. I strongly remember I learned the term from the book not long after it was published (pub. date 2007). I am confident it should be in there somewhere. I will report to you when I come across it.

  11. Chris Button said,

    March 19, 2020 @ 1:08 am

    "Vowel bending" is the term Schuessler uses to describe the changes to the vocalism of Old Chinese type-B syllables in their evolution to Middle Chinese. The type-A/B distinction refers to a bifurcation of syllable types across the whole Old Chinese lexicon and is not something that can be applied ad hoc outside of this two-way split. The distinction has come up many times on LLog. Most recently I just brought it up here:

  12. Jonathan Smith said,

    March 19, 2020 @ 2:21 pm

    Rather, for anyone still reading, Schuessler's "vowel bending" or "warping" refers to complementary changes in A *and* B. The key work is in an edited volume from 2006 ("The Qièyùn System ‘Divisions’ as the Result of Vowel Warping") but yes the idea is mentioned in passing in 2007. Note this is a *descriptive* account; thus "the cause of this warping is a matter of speculation" (2007: 121).

  13. Chris Button said,

    March 19, 2020 @ 11:18 pm

    @ Jonathan Smith

    Technically yes. Thanks for the clarification.

    In the case of the above discussion the comment was about the insertion of "i" or, as Pulleyblank treats, it /ɨ/ which fronted to /i/ or rounded to /u/ depending on the conditioning environment (the hallmark of the type-B syllable).

    In terms of the cause of the emergence of /ɨ/ in type-B syllables, I personally believe Pulleyblank is on the right lines proposal with his moraic analysis of syllable weight based on Kuki-Chin languages where it reflects a tense/lax distinction at the phonological level and surface length distinctions (both of the vowel and sonorant consonants) at the phonetic level. The correlation between OC syllable distinctions and Kuki-Chin tense/lax distinctions are not one-to-one for a variety of reasons (not the least the fact, as Matisoff has pointed out, we're dealing with prosody)

RSS feed for comments on this post