Of shumai and Old Sinitic reconstructions

« previous post | next post »

It's no secret that I'm a great fan of the AHD:

"The American Heritage Dictionary of the English Language, 5th edition " (11/14/12)

My devotion to AHD stems not just from its unparalleled inclusion of Indo-European and Semitic roots, but from its outstanding coverage of terms relating to Chinese languages and linguistics.  It was already strong in the latter respect in the earlier editions, but, with the 5th edition (2011), there was a noticeable improvement, such that the treatment of Chinese in AHD cannot be matched by any other dictionary of English.

Not content to rest on its laurels, AHD continues to strengthen its entries pertaining to Chinese.  The latest addition is a completely new entry on shumai.  Those who frequent dim sum eateries will know that shumai (also spelled shaomai, shui mai, shu mai, sui mai, shui mei, siu mai, shao mai, siew mai, and siomai) is a delicate, mouthwateringly delicious type of steamed pork dumpling.

Simpl. 烧卖 / trad. 燒賣

MSM shāomài

Cant. siu1maai6*2

Hokkien sio-māi

Hakka sehw mai

For the variant Sinographical forms of 烧卖 / 燒賣 and alternative names, see the Chinese Wikipedia entry here.

Here's the new entry on shumai from AHD:

shu·mai (shmī)

n. pl. shumai

1. A steamed or fried dumpling that contains a pastelike filling of minced ingredients, as pork, shrimp, ginger, and onion, usually seasoned with soy sauce and often only partially wrapped, exposing part of the filling.
2. A dish consisting of a number of these dumplings.

[Japanese shūmai, from Cantonese siu1 maai6, Cantonese reading (with siu1, to burn, roast, and maai2, variant in composition of maai6, to sell) of Mandarin shāomài (as written with the characters shāo, to burn, roast, and mài, to sell, since shumai are frequently sold in small shops and as street food), rewriting of earlier shāomài : shāo, a little, variant of shào (from Middle Chinese ᶊaw`, from Old Chinese *sr(i)âuh) + mài, wheat (perhaps with reference to the thin wheaten wrapper, from Middle Chinese maːjk, from Old Chinese *mrək, perhaps of Indo-European origin; see melə in the Appendix of Indo-European roots).]

The American Heritage ® Dictionary of the English Language, Fifth Edition copyright ©2016 by Houghton Mifflin Harcourt Publishing Company.

The American Heritage Dictionary website is at ahdictionary.com

VHM:  If some of the special characters and formatting are lost in transmission, I would encourage readers to go directly to the link for the AHD entry on shumai provided above.

A tremendous amount of research went into the writing of this etymology, and the editors have been extraordinarily generous in allowing such a lengthy note.

Most exciting to me is that the AHD etymology for shumai raises the possibility of a link to IE for the last syllable.  Two decades ago, I wrote a very long and detailed proposal for considering the Sinitic word mài 麥 ("wheat") as having been derived from an Indo-European source. This is on pp. 36b-38a of "Language and Script: Biology, Archaeology, and (Pre)History," International Review of Chinese Linguistics, 1.1 (1996), 31a-41b.  The linguistics for the trans-Eurasian spread of wheat are complemented by the genetic and archeological evidence.

In addition to the evidence cited at the end of the AHD etymology, we may also adduce Tocharian B mlutk-, to grind, crush, and Latin mola, emmer groats < molere 'to grind' (= Gothic malan, Hittite 3sg. mallai '(s)he crushes', etc.)


Zhao Zhijun, "Eastward Spread of Wheat into China — New Data and New Issues".  Chinese Archaeology, 9 (2009), 1-9.


…Sometime between 5000 and 3000 BP, a cultural package containing wheat, sheep and bronze industry spread from their homeland of the West Asia into the vast Eurasian Steppe, where distributed many small Bronze Age cultures linked to each others, and played like a cultural channel to transfer this package further east to the Northern Zone in China, and then the eastward spread direction turned to south, and the package eventually reached to the core area of ancient Chinese civilization, i.e., the Central Plains and Haidai regions.  Nevertheless, it is most likely that the Eurasian Steppe had been a main route connecting the West and the East long before the use of the well-known Silk Road.

Another possibility of the eastward spread of wheat is the route along the Silk Road of the historical time, i.e., spread from West Asia to Central Asia, continuously transferred along the northern part of Xinjiang, then passed through the Hexi Corridor, and eventually reached to the Northwest region of China


Chunxiang Li,  Diane L. Lister, Hongjie Li, Yue Xu, Yinqiu Cui, Mim A. Bowe, Martin K. Jones, and Hui Zhou.

Journal of Archaeological Science, 38 (2011), 115-119.


Wheat has been one of the most important crop[s] in Eurasia since the Neolithic period. Understanding the spread of wheat cultivation is crucial to understanding the spread of agriculture as a whole and the interactions between prehistoric populations across the Eurasian continent. However, the routes by which wheat cultivation spread eastwards have been poorly understood to date, due to the scarcity of plant remains recovered from archaeological sites. Desiccated wheat grains excavated from the Xiaohe cemetery in Xinjiang, and dated to the early Bronze Age, show excellent DNA preservation. Here we present an ancient DNA (aDNA) analysis of wheat (Triticum sp.) grains excavated from Xiaohe and provide the first definitive evidence for bread wheat in China during the Bronze Age. The nuclear ribosomal DNA internal transcribed spacer regions (ITS1 and ITS2) and the intergenic spacer region (IGS) were amplified. The IGS region within the D genome of wheat has a 71 bp insertion that is absent from corresponding regions in the A and B genomes. The results showed that the Xiaohe wheat showed most sequence similarity to hexaploid bread wheat (Triticum aestivum), including the characteristic insertion into the D genome. The presence of bread wheat at the Xiaohe cemetery is discussed in relation to it having spread into Xinjiang by the Bronze Age, providing new insight into the origins of bread wheat in East Asia.

[Thanks to Li Liu]

Notes on Tocharian

From Douglas Adams:

As an equation mlutk- is sort of all right, but distant.  Latin mola must be *molh2-ah2- or *melh2-ah2-.  (In Latin *-e- > -o- before an -l- so the root vowel could be, in PIE terms, either.)  PIE *molh2- 'grind' is found in Tch as mely- 'crush' (this is the word for crushing grapes among other uses).  The present mely- reflects *molh2-ye/o-, like Gothic ga-malwjan and ON mo/lva 'shiver, break into fragments.'  Tch mlutk- 'crush' (or 'hull'?) is ultimately  related.  The -t- reflects PIE -ske/o-, a present-forming suffix, which has been extended throughout the paradigm.  Otherwise we have *m(e)l-eu-T-.  The *mel- is common to both; the -h2- and -eu- are two different "enlargements" (remnants of an old layer of derivational morphology whose semantics are totally obscure)–the *-T- is yet a third.

Thus the better cognate, or at any rate the closer cognate, for Latin mola would be TchB mely- (TchA malyw-).

From Donald Ringe:

There are lots of Tocharian verb stems in -tk- (often -tka-, though the dictionaries don't always say so), and they don't all necessarily reflect the same process; but some have been extracted from inherited presents in *-ske- ~ *-sko- that were constructed to roots ending in dental stops, the *-s- being dropped between the stops by regular sound change.  (That was seen more or less simultaneously by Craig Melchert and Jay Jasanoff in the mid-1970's.)  A good example is Toch. B /nətka-/ 'to prompt' (a transparent attested form is the past participle nätkau), which appears to reflect *nud-ske/o- (cf. Vedic Skt. nud- 'push') with the thematic vowel lopped off; the underlying /-a-/ of the TB root can be a later addition.

If something similar happened to 'crush' we should be looking for a root something like *mlewd(h)-; if that's not what happened, we have a lot more explaining to do.

Actually the first thing to ask is:  are there possibly related Tocharian words that are not just transparent derivatives of this Tocharian root?  If there are, they might help us figure out how this weird-looking thing got to be like it is.


  1. leoboiko said,

    July 19, 2016 @ 1:58 pm

    Is AHD using Schuessler's Minimal Old Chinese (from the ABC Etymological Dictionary) as the reference for Old Chinese reconstructions? That's cool. Most people use Baxter-Sargat (perhaps because they've been kind enough to share their reconstruction online), but I like the cautious approach of Schuessler's variation (which is derived from Baxter's, but restrict itself to the less controversial parts).

    For comparison, shāo-mài < *sr(i)âuh-mrək (Schuessler) would be *stewʔ-mərˤək in Baxter-Sagart. (Incidentally, I notice *sr(i)âuh, used by AHD, is listed by Schuessler under 稍 shào “little”; but 少 shāo “little” gets *hjauʔ? instead.)

  2. Chris Button said,

    July 19, 2016 @ 3:29 pm

    The curious thing about 麥 "wheat" is the foot component at its base that would be expected to be found at the base of its phonetic 來 "come" instead. It is almost as if the usages have been reversed and this is attested even back in the oracle-bones.

    Further evidence that 麥 "wheat" was a loanword possibly comes from the instability of the velar coda: 麥 is *mrək but 來 is *rə. Interestingly, 來 *(m)rə is also the phonetic in 每 *məʔ "every" where it appears in abbreviated form above the 母 component.

  3. Victor Mair said,

    July 19, 2016 @ 4:06 pm

    From a Berkeley colleague:


    …[W]hen I went thru Boodberg's collected writings I noticed a reconstruction for 'rice' which struck me as possibly compatible with Proto-Indo-Iranian *wrinz'hi-.


    I doubt that this could be mǐ 米, the usual Chinese word for "rice". All the Old Sinitic reconstuctions of that that I know of begin with a bilabial nasal. Could Boodberg have been discussing some other word for rice or a word for a type of rice?

  4. leoboiko said,

    July 19, 2016 @ 4:30 pm

    @Chris: Based on Pulleyblank's etymology (apud Schuessler's entry for 麥, p. 374), perhaps it went something like this?:

    Stage 1: 來 is originally a pictograph for "wheat", *mrək > mɛk > mài

    Stage 2: 來 is borrowed by sound (as a rebus) to express "to come", *rək > *rə > lə > lái (related to a number of Sino-Tibetan etymons).

    Stage 3: 麥 is designed to represent the latter unambiguously.

    Stage 4: But 麥 didn't take; "to come" is a much more frequent word than "wheat", and 來 was already well-set as its representation (cf. 原/源, or 官/管; there are many such "failed disambiguations").

    Stage 5: Finally, the newer graph ends up being used reversedly for the less common word, "wheat", leaving 來 for "to come". Cf. other graphs where the borrowed but more important meaning "took over" the original pictograph, like 北 "back" → "north".

    So 麥 would be unusual in that it ended up with the wrong semantic component; but not unusual in the confusion between original and derived character, and in the loss of original meaning for the original pictograph.

  5. Victor Mair said,

    July 19, 2016 @ 5:55 pm

    From Ed Vajda:

    Here's what I have of the 'grains vocabulary' in Yeniseian languages. Of course, all these words are borrowed, but I don't always have a specific source.

    arba, Yen. ‘barley’ ⇽ Turk. (T 1986:76 cites Altai, Kum. arba, Kirghiz arpa id.; St 2004:194 cites MTurk. jarpa id.) ⇾ Arin (M, W, Kl) arba, Assan (M, W, Kl) arpá, Arin, Kott (H) arba id.
    areš, ‘Yen. names of various grains’ ⇽ Russ. rožʲ id., possibly through Turk. (T 1986:76; St 2004:192) ⇾ Kott (C) âreš (pl âretn), Kott (W) árɨš, Assan (W) arɨš, Kott (H) arɨš, Pump. (VW) oros, Koibal (C) âreš ‘rye’; Kott (M, W, Kl) arɨš ‘barley’; Kott (M, W, VW, Kl) árɨš ‘oats’, ‘wheat’; Arin (H) rožok id. ⇽ Russ. rožj ‘rye’ + ok (loanword suff.)
    bukdaj, Arin ‘wheat’, Arin (M, W, VW, Kl) bugdaj, (H) bugdoek (loanword suff. -ok), Kott (H) butai; Assan tútaj (M), bútaj (VW, W), búgaj (Kl) id.; T 1986:76, St. 2004:193 ⇽ Turk. (Altai buudaj, Khakas puʁdaj, Kirghiz buudaj, Kum. bugdaj id.)
    čarba, Arin (H) ‘grain’, ‘rye’ ⇽ Turk. čarba ‘groats’
    kaŋdáca, Pump. (W) ‘grain’, (M) kaŋdaza ‘wheat’ (origin unclear)
    kokaj, Arin (W) ‘rye’, Arin (M, W) koqáj id. (origin unclear)
    kulpá, Ket ‘groats (gruel)’, ‘barley’; Yugh kurpá id. ⇽ Russ. krupa id.
    sulu ~ suli, Yen. ‘oats’ (⇽ Turk.) ⇾ Kott (H) šuli, (C) šuli ~ šulje ~ šulji (pl šulaŋ), Assan (M, W, VW, Kl) šulí, Arin (M, W, Kl) sulú, (VW, H) sulu, Koibal (C) šulu id.; T 1978/3:10 ⇽ Khakas sula, Tuvan sula, Kum. sulɨ ~ sula id.

  6. Jim Breen said,

    July 19, 2016 @ 6:13 pm

    Apropos of "Simpl. 烧卖 / trad. 燒賣", I note that in Japan the usual kanji form is 焼売. 燒賣 is not unknown, but accounts for < 1% of the kanji usage. Most of the time it is written しゅうまい or シューマイ.

  7. Victor Mair said,

    July 19, 2016 @ 7:51 pm

    I am grateful to Chris Button and leoboiko for raising the graphic side of the linguistics of the word for "wheat" in early Sinitic. In truth, my quest for the possible west Eurasian source of the Sinitic word, founded on the archeological and botanical evidence, was inspired by the premise that it probably began with the consonant cluster ml-. The basis for that hypothesis was that an original ml- word for wheat written with the pictographic character 來 was appropriated to write the abstract l- verb for "come". After dimidiation of the ml- initial, that left m- for the word meaning wheat and l- for the word meaning "come". This is what lead me on the quest for words in west Eurasian languages having to do with wheat that contained m and l.

  8. ~flow said,

    July 20, 2016 @ 5:49 am

    @leoboiko—got to love that "failed disambiguations" expression. FWIW for those interested, there's a recent post here on LL where, inter alia, character families such as 原/源/元 were discussed ("Devin, a newly discovered language" http://languagelog.ldc.upenn.edu/nll/?p=22185).

    My personal favorite of "failed" character evolutions is certainly 网>網>网—originally a simple picture of the thing it signifies, later to be amended with 亡 (a hint to its sound) and 糸 (a hint to its meaning), which sidelined the pictorial (and arguably memorable) 网 which got reduced to 䍏 and fused with 亡, causing 網 to enter onto a millennia-long collision course with the very similar (in sound, meaning, and general usage patterns) 綱 until, finally, the two were disambiguated by the PRC spelling reforms which reconstituted 网 (and turned 綱 to 纲).

    I'm not aware of listings that go specifically through the histories of such character families (網网, 麥來, 原源(元??), 官管, 北背); I should certainly be glad if someone could provide pointers.

  9. V said,

    July 20, 2016 @ 6:37 am

    A local fish store recently opened a fast food freezer. Last week I bought and have half eaten a whole box of shrimp Siu Mai, as it is called on the box, accompanied by 燒賣. It took me a bit to get to the Wikipedia page explaining what it is and teh various ways it's referred to in other languages, and recipes. I just saw dumplings with shrimp on top, uncovered, and though it's a good idea. They seem to be from Vietman, imported here from the Netherlands. In the ingredients, the left column (English, French, Spanish) says "Shiu Mai" and the right column, (Dutch, German, Greek) says "Siu Mai" (the Greek is Σ'ΙΟΥ Μ'ΑΙ).

  10. leoboiko said,

    July 20, 2016 @ 6:40 am

    @~flow: Japan is still stuck on 網 (pun intended); 网 is so much more elegant :( I don't think there's any confusion with 綱, though, with both characters being well in use.

    Thanks for the link, it's of interest to me. I don't know of any listings like that, unfortunately. I've been severely missing a state-of-the-art graphical history of Chinese characters, which would take into consideration recent research in oracle-bones as well as historical linguistics.

  11. Victor Mair said,

    July 20, 2016 @ 6:50 am


    "I've been severely missing a state-of-the-art graphical history of Chinese characters, which would take into consideration recent research in oracle-bones as well as historical linguistics."

    Matt Anderson and Chris Button are working on various aspects of such a project. We'll try to put the results of their work online before too long. David Prager Branner had made great progress on this work (especially the phonological side) some years ago, but I think that he has set it aside, at least for the present.

  12. Victor Mair said,

    July 20, 2016 @ 6:53 am


    It is for reasons such as these that a friend of mine once said that the Chinese character system is yītāhútú 一塌糊塗* (lit., "one collapse muddle", i.e., "a [complete] mess**; in a terrible condition; in an awful state", or, as Baidu fanyi renders it, "All Fall Down"!), yet we (some of us) keep trying our best to make sense of it, as though it were integral and coherent.

    *hútú 糊塗 is also written as 胡突, 胡塗, 糊突, 鶻突, etc.

    **as Donald Trump famously said of Jeb Bush

  13. leoboiko said,

    July 20, 2016 @ 9:58 am

    @Mair; That's very interesting. And being online certainly makes a world of difference; Button's Phonetic Ambiguity in the Chinese Script: A Palaeographical & Phonological Analysis sounds fascinating (I've been curious about "the Boltz question" ever since reading the latter), but I couldn't afford it. There are so many interesting projects coming out; I wish I could study under you people…

  14. Michael said,

    July 20, 2016 @ 11:28 am

    That is spectacular work. I love the AHD and bought it on your recommendation.

    I previously thought that the Chinese word for honey was the only known word of IE origin. Fascinating.

  15. leoboiko said,

    July 20, 2016 @ 11:49 am

    @Michael: Buddhist terms alone comprise an entire stratum of Indo-European terms layered into Chinese. My favourite is 魔 (Japanese/Korean/Vietnamese ma), “magic/evil/demon” < Sankrit Māra "Buddhist demon king", a causative from the verb mṛ “to die” < PIE *mer > Latin mors > English “mortal”, Spanish muerte etc.

    This implies, among other things, that King Koppa from the Japanese video game Super Mario, as a Great Demon King 大魔王 Daimaou, holds a title etymologically related to "mortal" , muerte, "murder" etc.

  16. Jichang Lulu said,

    July 20, 2016 @ 5:09 pm

    Likely off topic, but since Ed Vajda has mentioned Yeniseian words borrowed from Turkic 'wheat' (modern Turkish buğday): Mongolian has ᠪᠤᠭᠤᠳᠠᠢ buɣudai буудай 'wheat'. Also from Turkic?

    For Korean 밀 mil 'wheat', there seems to be a Middle Korean form milh, remarkably similar to Old Chinese *mrVk. The Naver dictionary gives the 1461 Neung'eom gyoeng oenhae 능엄경언해 楞嚴經諺解, a commentary on the Leng yan jing or Śūraṅgama sūtra, as a source for milh.

  17. Dave Cragin said,

    July 20, 2016 @ 9:57 pm

    A recent article in Science offers a similar example of the spread of crops and language: it links the spread of Asian crops to the Island of Madagascar with that of people from Austronesia. Years ago, I had heard a discussion of Madagascar’s language, Malagasy, and its similarities to other Austronesian languages by John McWhorter. The new evidence of Asian crops fits with this.

    A few examples from McWhorter:
    “stone”: bato (Tagalog); batu (Malay), vatu (Samoan), vato (Malagasy)

    “Eye”: mata (Tagalog), mata (Malay), mata (Samoan), maso (Malagasy)

    For those who don’t know the story: Madagascar is an island off the coast of mainland Africa. Yet, its "language is more closely related to Hawaiian than Bantu."

    The new article is: Culinary frontier’ tracks Madagascar's Asian settlers, Andrew Lawler, Science 03 Jun 2016: Vol. 352, Issue 6290, pp. 1154-1155

  18. Endymion Wilkinson said,

    July 21, 2016 @ 4:15 am

    The American Heritage Dictionary (5th edition, 2016) makes more of an effort than most to include etymologies for Chinese loanwords in English. However it makes no effort to date the first known entry into English of these loanwords (for that you have to turn to the OED). As a result, AHD sometimes suggests unlikely etymologies. For example its entry on "chopstick" (sic) suggests that the word chop came from Pidgin English chop (quick) from Cantonese gap. But the OED cites as the first known usage in English of chopsticks an English traveler in Tongkin in 1699. That was some time before Pidgin English developed in Canton and the other treaty ports in the late (?) eighteenth and nineteenth centuries. It was also some time before the southern dialects switched from zhu to kuaizi for chopsticks.

  19. leoboiko said,

    July 21, 2016 @ 6:01 am

    "chopstick" (sic)

    Is the "(sic)" due to the singular? It seems to be around, and at least since 1838.

  20. Wang Yujiang said,

    July 21, 2016 @ 10:23 am

    @ Victor Mair
    shumai is an English word that is a transcription from spoken Cantonese.
    shumai in Chinese Pinyin is shaomai.
    Although the filling of a shaomai may consist rice, shao means cooking, and mai means selling.
    The Chinese characters in modern Chinese (baihuawen) are phonetic symbols.

  21. Rodger C said,

    July 21, 2016 @ 11:41 am

    Buddhist terms alone comprise an entire stratum of Indo-European terms layered into Chinese

    I thought Michael was referring to words in the earliest stratum of written Chinese, like "honey" and (I was always taught" "horse" and "dog." Now someone will correct me about these.

  22. Victor Mair said,

    July 21, 2016 @ 1:43 pm

    From Peter Golden:

    See András Róna-Tas and Árpád Berta, West Old Turkic, I: 186-188, entry under Hung. búza (< ultimately buɣday).

  23. leoboiko said,

    July 21, 2016 @ 3:17 pm

    @Rodger: I don't know if those would be as old as "wheat", but perhaps one way of hunting for earlier-than-Buddhism loans is to examine disyllabic morphemes. It seems that most of those are ultimately either loans or mimetic words (often animal names, like 蟋蟀 xīshuài "cricket" < *srit-srut or 蝙蝠 biānfú "bat" < *pen-puk). Sproat, in A Computational Theory of Writing Systems, presents some 70 disyllables found computationally in modern Mandarin; he estimates there must be about 100 of them.

    One example is 葡萄 pútao "grape", thought to derive from budāwa, which I don't know whether it's Elamite (Haspelmath/Tadmor (ed.)) or Persian (Schuessler). Another is 獅子 shīzǐ "lion", either < Tocharian A śíśäk, B ṣecake (Pulleyblank) or Persian šer/šē/šī (Haspelmath/Tadmor) (cf. languagehat thread).

  24. Victor Mair said,

    July 21, 2016 @ 7:24 pm

    Kamil Stachowski, Names of Cereals in the Turkic Languages

  25. Jichang Lulu said,

    July 22, 2016 @ 8:38 am

    Stachowski (linked to above) on buğday (p. 89ff.):

    "It seems then, that a great number of Tkc. forms […] can in fact be reduced to one initial shape of *boguda". The final glide would be "a diminutive suffix". [I suppose this would make Mongolian buɣudai more likely to be a Turkic loanword.]

    A Turkic-internal etymology by Tatarincev (from a root *bug/k) "does not seem to be particularly convincing".

  26. Jichang Lulu said,

    July 22, 2016 @ 9:21 am

    In an article on IE words for 'barley', Václav Blažek suggests a connection between a IE root reflected in Proto-Celtic *mraki- (Modern Irish braich 'malt') and 麦. He mentions reflexes in other IE branches incl. Anatolian.

    Here's Blažek's article in Czech (pdf) and English (Scribd).

  27. Victor Mair said,

    July 22, 2016 @ 10:19 am

    From Marcel Erdal:

    It's really impressive, how many animal husbandry and agricultural terms are shared by Turkic and Mongolic: Concerning agriculture, the languages share the terms for apple, barley, rice, wheat, garlic, the verb for sowing and planting, and a word signifying 'crop, grain, cereal, field', derived from this latter. The word for 'onion' is also shared, B U T: Comparing Mongolic songgina 'onion' with Turkic sogon 'onion' we note that the Mongolic term HAS to come from Turkic sogon+gina, incorporating a Turkic diminutive suffix which was not in use in Mongolic. So this Mongolic term has to have been borrowed from Turkic, which might be the case also for the others. Also interestingly, there are a number of names for types of 'wild onion', which are not shared. I once wrote a paper showing, I think, that the Turkic / Mongolic 'apple' term comes from Indo-European.

    Perhaps unsurprisingly (since these people were herders before they were farmers), the domain of shared animal husbandry terms is even wider.

    From András Róna-Tas (7/26/16 7:08 AM):

    To the interesting discussion I would only add: Marcel is right when he points to the fact that the terminology of husbandry and agriculture is to a high proportion common to Mongolic and Turkic, what is perhaps also of interest, that the same terminology appears in Hungarian as loan of West Old Turkic origin. On pp. 1160-1161 of WOT I collected the terminology. What was also striking: practically no terms of horse-keeping, and those terms which pertain to it are partly late Coman terms.
    This shows that there existed a culture of husbandry and agriculture in Central Eurasia with a high degree of Old Turkic origin, which was taken over by the Mongols, but also by such non-Altaic people as Hungarians.

  28. Victor Mair said,

    July 22, 2016 @ 11:26 am

    Marcel's "apple" paper appeared quite a while ago: "Around the Turkic 'apple'’. Journal of Indo-European Studies. 21 (1993), 27–36.

  29. Victor Mair said,

    July 22, 2016 @ 2:50 pm

    From Juha Janhunen:

    Let me just add that my teacher Aulis J. Joki also wrote an early paper on 'apple' under the title "Der wandernde Apfel" (1964, also in Finnish "Omenan vaellus", 1963).

    As I see it, the relationship between Turkic *bugda-y 'wheat' and Mongolic *budaxa/n 'cereals' (> 'grain, wheat, millet, rice') is best explained by a metathesis on either side. The original form would have been either *budaga or *bugada, with regular vowel loss in Turkic and weakening of *g to *x (later > zero) in Mongolic. This is, of course, a Turkic (Pre-Proto-Bulgharic) loanword in Mongolic (Pre-Proto-Mongolic), and it may be noted that the meaning on the Mongolic side is generic, suggesting that the Mongols were not much into agriculture at the time the lexical borrowing took place.

  30. Victor Mair said,

    July 22, 2016 @ 2:51 pm

    From Marcel Erdal:

    In my apple paper, I also quoted Prof. Joki’s paper on this topic.

    Turkic bugday actually has two Mongolic cognates. Beside the one you mention, there is also *buudai (or in what other way you want to spell it) attested in practically all Mongolic languages including the ones in China and the Middle Mongolic Muqaddimatu ‘l-Adab and Hua-Yi yiyu (but apparently excluding Dagur). This is closer to Turkic both in meaning and shape and thus likelier to be dismissed as a recent loan. But it is still remarkable how absolutely widely spread this word also is both in Turkic and Mongolic; the absence of this one in the Secret History and the presence of the one you have referred to in that source may just be a coincidence. This double Turkic-Mongolic connection is in itself interesting; can you think of other such cases? Concerning ‘your’ term, a buudan ~ budaan metathesis is understandable, but the °y ~ °n alternation deserves more attention: IF °y > °n replacement could in this case have served a collective meaning, it could explain the more general content of budaan. There is also the interesting fact of Turkic tarı-g (and its later shapes) signifying either ‘wheat’ or ‘barley’, depending what was the central cereal of the society in question.

  31. Victor Mair said,

    July 22, 2016 @ 3:04 pm

    From Peter Golden:

    A bibliographical side-note: Turk. alma (Mod. Turkish elma) “apple” < "Ancient European" *amlu > ablu ~ aplu was accepted by Gamkrelidze and Ivanov, Indoevropejskij jazyk i indoevropejcy (Tbilisi, 1984), II: 639), Eng. version (the one available to me at present): Indo-European and the Indo-Europeans (Berlin-NY: Mouton de Gruyter, 1995), I: 550.

  32. Victor Mair said,

    July 23, 2016 @ 6:50 am

    From Jichang Lulu:

    Erdal's comments are very interesting indeed. If I understood him and Stachowski correctly, the situation would be similar for 'onion' and 'wheat': similar Turkic and Mongolic terms, but embedded (possible) Turkic morphology, suggesting a Turkic > (para-?)Mongolic borrowing. At least in the case of wheat, that would seem to agree with the presumed direction of the spread of the crop itself.

    He also brings up wild Allium which is a whole nother tin of annelids.

    At any rate the Turkic and Mongolic (and per Róna-Tas also Hungarian) words don't seem to be related to the Chinese, Korean and possibly IE etyma. On that topic I wanted to bring to your attention the elephant in the room, which I didn't want to comment about out loud. I mentioned Middle Korean milh, where the consonants point towards a Chinese loan. That in itself would be quite interesting, because the borrowing would have to be quite early indeed for the OC *-r- to be preserved. I wonder if there are any other Korean words indicating such an early borrowing. There's Korean 붓 bus 'writing brush' from 笔, which is also an early borrowing but not so blatantly Old.

    (Marc Miyake on 'writing brush':

    The aforementioned 'elephant' is that the Moscow-school Altaicists see the Korean 'wheat' not as a Chinese borrowing, but as a reflex of an 'Altaic' proto-word. That's in Starostin et al.'s Altaic etymological dictionary.

    Those Altaic reconstructions are, to put it mildly, controversial. Yet some of the attested Tungusic words S. et al cite (and I haven't checked the sources) also suggest a Wanderwort with MRG-like consonantism. I was wondering if you could marshal the forces of Tungusica to provide an informed opinion. Alexander Vovin has reviewed the Altaic dictionary very critically, and is also an expert on Old Japanese; it would be great to hear his opinion on the Tungusic wheat words, as well as on ancestors of Japanese 麦 mugi.

    As for Blažek's article, again I think his *mrk requires a second opinion. On the face of it, a crop-word *mrk attested from Celtic to Anatolian to possibly Indo-aryan surely induces a Yay!-moment. But if you read his attested words, I can't see too much crop-related about them so getting to a 'barley' meaning for the root (indeed even for the Proto-Celtic root) needs some arguing for. And of course barley isn't wheat as any fule no. So deep in my heart I'm sceptical, and brought up Blažek's work to lure some Indoeuropeanist into assessing the *mrk claims.

  33. Victor Mair said,

    July 23, 2016 @ 6:55 am

    From Marcel Erdal:

    The original Turkic form of the word was alïmla; alma is one of the secondary variants (as I pointed out in that paper of mine).

  34. Victor Mair said,

    July 23, 2016 @ 6:57 am

    From Juha Janhunen:

    Yes, Mongolic buudai must be a much later borrowing from Turkic, and in the specific meaning of 'wheat'. I could even think that buudai, written either "buudai" or "buq(h)udai", reflects directly the Turkic secondary bisyllabic shape bugdai, and not a trisyllabic shape like *buxudai. In any case, the -y in Turkic must be some sort of derivative suffix, while in Mongolic the unstable n in *budaxa/n is also a secondary element. The Turkic word tarï-g also appears in Mongolic with an unstable n as *tarï-ga/n > *tari-xa/n > tariya/n 'grain, field', which formally is simply a deverbal noun from tari- 'to sow, to plant, to cultivate'. (I think that Turkic tarï-g goes back to *tarï-ga, and that the Turkic deverbal noun suffix -gA : -gAn > -xA/n was borrowed into Mongolian not only in the composition of lexicalized words like this, but also in the grammatical function of imperfective participle – but this is not relevant for the present discussion.)

  35. Victor Mair said,

    July 23, 2016 @ 6:59 am

    From Marcel Erdal:

    Interestingly, there was no special term for 'rye' in Turkic; it was borrowed from Persian or from Russian or called 'black wheat'.
    I was wrong about the 'other' meaning of tarig being 'barley'; it was/is 'millet'. 'millet' is another cereal name shared by Turkic and Mongolic which I didn't mention before: The original term for 'millet' is qonaq / qonoq in both language groups.

    yarma / jarma 'groats' is still another Turkic – Mongolic cereal term!! In case ya:r- 'to split' is Turkic but not Mongolic (in spite of Mo. jarim), this would also be a clear case of borrowing.

  36. Victor Mair said,

    July 23, 2016 @ 10:51 am

    From Jichang Lulu:

    "I could even think that buudai, written either or , reflects directly the Turkic secondary bisyllabic shape bugdai"

    I think Prof. Janhunen meant to give Mongolian spellings for buudai 'wheat' but they didn't make it past WordPress. The Cyrillic orthography is буудай and the Mongol script ᠪᠤᠭᠤᠳᠠᠢ, transcribed buɣudai in a traditional system, or (I think) buqhudai in BJR (Balk-Janhunen Romanisation).

    [VHM: The problem is fixed now in Juha's comment — the second comment above this one.]

    This is as good an occasion as any to promote the BJR system. Unlike the system customarily used in Mongolian studies, which requires the 'transcriber' to disambiguate distinctions the Mongol script ignores, BJR is close to a true transliteration. Despite an emerging ability to input and display text in the Mongol script, a faithful transcription system is still necessary and I think BJR deserves to do for Mongolian what Wylie does for Tibetan.

    BJR is described here by its creators. Example transcriptions of running text: Balk's transcription of a few stanzas from the Mongolian Udānavarga 法句经 (with extensive explanations, Tibetan parallel text and German translation); and my own attempt, involving a tale from the Sutra of the Wise and the Foolish 贤愚经 in Mongol script, BJR, traditional romanisation and Cyrillic.

  37. Victor Mair said,

    July 23, 2016 @ 3:04 pm

    From Jichang Lulu:

    In Blažek's article (at least the Czech version – I haven't managed to access the English) there's a quote of an article by Qing Zhaorong 庆昭蓉
    where she suggests (according to Blažek) that a Tocharian B word tsänkana could be a loan from Chinese 青稞 'highland/Tibetan barley' (whose flour is often used to make tsampa). Blažek mentions a competing etymology and offers more discussion. Here's the word in a Tocharian dictionary:

    I know no Tocharian and have nothing to add, but thought you might find that interesting.

    Source as quoted by Blažek:
    Ching Chao-jung. 2008. On the names of cereals in Tocharian B. Paper presented at the International Conference for the Centenary of Tocharian Studies (Moscow, August 2008).

  38. Jongseong Park said,

    July 26, 2016 @ 4:45 pm

    I'm late to this party, but thanks to Jichang Lulu for bringing up Middle Korean 밇 milh.

    Possible very early Sinitic borrowings in Korean? It has been suggested that 바람 baram "wind" is a cognate of Old Chinese 風 *prəm or *priəm (Mandarin fēng)—see the etymology section in Wiktionary. See also 빗 bit "comb", purportedly from Old Chinese 篦 "before the tonogenesis process that converted the final *-s into qùshēng." according to Wiktionary.

    You might enjoy the discussions in this thread. I was guilty of derailing it with a rather wild speculation about a possible connection between Korean 곰 gom "bear" and Old Chinese 熊 *wum ~ *ɢʷum (Mandarin xióng). Well, the earlier Korean form was 고마 goma, certainly cognate with Japanese くま kuma, but people disagreed on possible connections with the Old Chinese form.

    On another note, thanks for bringing BJR to my attention.

RSS feed for comments on this post