Turkic kaymak and Sinitic sū: a dairy product and a food texture

« previous post | next post »

From Jacob Reed:

Inspired by Miss Gao's 小高姐’s latest video, I've been trying to track down how 酥 acquired its present, seemingly contradictory connotations of "crispy" and "soft / relaxed". Paul Kroll's Classical / Medieval dictionary lists that it originally comes from the Persian for kaymak / clotted cream. 汉语大词典* indicates that this meaning is first attested during the Tang period.  Neither provide any indication of how we got from kaymak / clotted cream to "crispy" (the use of butterfat in pastry?).

In any case, I'm now curious if there's a more general trend of Sinitic dairy terms (like horse-related terms) coming from Central Asia, which would only make sense.

[VHM:  *Hànyǔ dà cídiǎn 漢語大詞典 (Unabridged dictionary of Sinitic)]

So, what is this sū 酥 that is crispy, flaky and soft, satiny, limp at the same time?  And what does it have to do with the Persian word for Turkic kaymak, which is like clotted cream?

There's a distinctive Chinese food texture called by this word sū 酥.  It signifies something edible that is flaky, crispy, maybe even crumbly — like shortbread, though good pie crust probably exemplifies it most purely and essentially.  Another variety of this texture is the butter, flour, and egg covering around the much-loved Taiwanese pastry called fènglí sū 鳳梨酥 / 凤梨酥 ("pineapple cake / shortcake").

On the other hand, dictionaries say that sū 酥 also means soft, satiny, limp, which seems to go at cross purposes to the sū 酥 crispiness I so treasure.

What intrigues me most of all, though, is that Paul Kroll's dictionary connects sū 酥 with Central Asian Turkic kaymak, a dairy product, and an Iranian term for fat.  Here's the entry from Kroll's dictionary (p. 432b):

    1. (med[ieval]) kaymak, clotted cream; butterfat (< Iran. fšu-, fat, fatty matter, as in fšutā-, cheese).

Here are the linguistic portions of the entry on "kaymak" from Wikipedia:

Kaymak is a creamy dairy product similar to clotted cream, made from the milk of water buffalos, cows, sheep, or goats in Central Asia, some Balkan countries, some Caucasus countries, Turkic regions, Iran and Afghanistan.

The word kaymak has Central Asian Turkic origins, possibly formed from the verb kaymak, which means melt and molding of metal in Turkic. The first written records of the word kaymak is in the well-known book of Mahmud al-Kashgari, Dīwān Lughāt al-Turk. The word remains as kaylgmak in Mongolian, and with small variations in Turkic languages as qaymaq in Azerbaijani, qaymoq in Uzbek, қаймақ in Kazakh and Shor, каймак in Kyrgyz, kaymak in Turkish, gaýmak in Turkmen, კაიმაღი (kaimaghi) in Georgian, and καϊμάκι (kaïmáki) in Greek.


From Jonathan Skaff:

On sū 酥 as kaymak, see Edward Shafer’s chapter in K. C. Chang’s Food in Chinese Culture and Keith Knapp’s article in the most recent volume of Early Medieval China. You also can obtain a feel for its “soft, satiny and limp” texture by watching this Turkish woman’s YouTube lesson on kaymak preparation.

From Jinyi Cai, an excellent pastry chef:

As for the term “sū 酥”, it is certainly very interesting. When I think of it in terms of baking or cooking, it does refer to the flaky, crispy, and crumbly texture specifically. Pie is the typical Western example of sū 酥. Traditional Chinese desserts, such as liúlián sū 榴莲酥 ("durian dessert / crisp"), sū pí dàntà 酥皮蛋挞 ("crispy egg tart"), qiān céng sū 千层酥 ("mille feuille; millefeuille; puff pastry; thousand layers crispy dessert"), are all very popular even nowadays. These all have the flaky and crispy texture. The process of making sū 酥 is interesting too and it correlates to the Central Asian Turkic term, since you need to layer fat (animal fat, butter mostly) on top of the flour layer, then fold it, then repeat many times in order to get the many layers of crispy texture. It definitely makes sense for the term “kaymak” to refer to cream or butterfat.

As I think of it more thoroughly, I wonder what Chinese people used to make sū 酥 in ancient times, since butter wouldn’t have been available back then. I looked it up and couldn’t find any recipe to make sū 酥 without using butter or animal fat. This makes me wonder. Did people use something else to make sū 酥, maybe alcohol and grains? The Chinese character sū 酥 simply can be divided into two parts: the left yǒu 酉 and the right hé 禾. 酉 is alcohol* and 禾 is grain. But then I also think of the regional food / drink sūyóuchá 酥油茶 (buttered tea) that people in Xinjiang and Tibet drink. I am not sure of the history and origin of sūyóuchá 酥油茶 and whether it came before Chinese people started making sū 酥 desserts and food. I don’t think the making of sūyóuchá 酥油茶 buttered tea) involves alcohol. It’s a very interesting topic for sure!

[*VHM:  a pictogram showing an ancient vessel used in making and storing fermented millet alcoholic drink (not wine, which is made from grapes and other fruit).  Cf. the cognate jiǔ 酒 ("alcohol; liquor; brew").

As for the other definition of the Chinese character sū 酥: soft, satiny, limp. I think this is mostly used to refer to people. It’s like a metaphor or analogy: the person is limp / soft / weak / feeble, so that he/she is almost like a crispy and flaky sū 酥 pastry that cannot be touched or disturbed.



Selected readings

In "Galactic glimmers: of milk and Old Sinitic reconstructions" (1/8/19), I discuss several Eurasian milk-related terms.


  1. BobW said,

    March 27, 2020 @ 7:52 pm

    Must make it hard to order tacos.

  2. Chris Button said,

    March 27, 2020 @ 10:36 pm

    When prepared properly, clotted cream does have a crust on the top

  3. Jonathan Smith said,

    March 28, 2020 @ 1:38 am

    re: "contradictory connotations of "crispy" and "soft / relaxed"
    IMO "crispy" is off; crispy foods are not called su1 酥. "Soft/light/relaxed/un-solid" seems closer for both the word su1 酥 generally (su1ruan3 酥软, su1ma2 酥麻, tui3 dou1 su1 le 腿都酥了, etc.) and for the specific application to pastry-ish textures.

  4. David Dettmann said,

    March 28, 2020 @ 7:14 am

    I am a big fan of crispy and flaky, and one of my favorite 酥 is the fried sliced chiles (often with sesame seeds and peanuts) that I started noticing a few years ago in North China, 香辣酥. It is an addictive snack.

    酥油's potential link to Turko-Mongolian süt/süü is intriguing. In a footnote in their analysis of 饮膳正要, P. Buell and E. Anderson cite entries for "Mongolian tea" 穌簽 and "butter" 穌油 with variants of that su character (p. 111 of A Soup for the Qan). Maybe the occasional 酥 link to milk/cream in Chinese came from sources like the YSZY?

  5. Chris Button said,

    March 28, 2020 @ 10:39 am

    Iran. fšu-, fat, fatty matter, as in fšutā-, cheese).

    It's an interesting comparison. Having now looked around a bit, are we sure the meaning is not "cattle, livestock, sheep" rather than "fat, fatty matter" per se?

    a pictogram showing an ancient vessel used in making and storing fermented millet alcoholic drink (not wine, which is made from grapes and other fruit

    酥 seems to have been treated as having 穌 as its abbreviated base. Somewhat tangential to the discussion here, I have a problem with how 穌 is often reconstructed with *sŋ- as its onset (on the basis of its apparent phonetic 魚 with *ŋ-) as if the Middle Chinese *s- resulted from the *ŋ component just being rather arbitrarily preempted.

    I treat pre-OC *sŋ- as regularly giving the rather unstable OC *ʰŋ- as the source of Middle Chinese *x-. I would therefore favor reconstructing 穌 with a lateral fricative *ɬ- (a far better source of Middle Chinese *s-) and assume that it was close enough phonetically with the shift of unstable *ʰŋ- to *x- to allow for an interchange of velar nasals and laterals in xiesheng (phonetic) series that extended to their voiced counterparts. There's good supporting evidence for this elsewhere (e.g., 牙 *ŋ- being phonetic in 邪 *l- whose lateral onset is supported by its etymological associations).

  6. Bathrobe said,

    March 28, 2020 @ 7:08 pm

    Mongolian хайлмаг (khailmag) is a very nice snack, firm but soft. From the Internet I found the following description:

    khailmag is … basically caramelized clotted cream. The cream is heated in a pan and sugar and flour (and sometimes raisins) are combined into the mix (once any liquid fat has been scooped away). Simply the best ingredient for an afternoon tea.

    A slightly more detailed description is here: https://www.mongolfood.info/en/recipes/khailmag.html

    But as far as I know (which is not a lot), Mongolian nomads don't combine this with any kind of flour or dough to make pastries. Is it not possible, then, that 酥 is historically a marriage of delicious nomadic fats with the flour-based bakery of the sedentary peoples? That could account for the contradictory meanings of the term.

  7. Kakurady said,

    March 28, 2020 @ 8:12 pm

    酥 describes croissants very well.

  8. Victor Mair said,

    March 28, 2020 @ 9:47 pm

    From Diana Shuheng Zhang:

    "Cold kaymak" 寒酥 was an allusion to snow in Chinese poems! This allusion seems to be late, though, in the Ming Dynasty — yet still fun to take a look at:

    一行分向朱門屋,誤落寒酥點羊肉。—— 《謔雪》

    朝來試看青枝上,幾朵寒酥未肯消。—— 《梨花》

    Both lines are by Xu Wei 徐渭 (1521-1593). Doesn't it feel nice to have "cold kaymak" along with mutton?

  9. Victor Mair said,

    March 28, 2020 @ 10:17 pm

    From David Dettmann:

    It might he tough to get some Chinese 香辣酥 in the near future, but I could make some for you. FYI, I did a post about that topic last year where you can see an image of a container I bought in Ulaanbaatar:

    I should have paid more attention to the question of that post aiming at a Tang Dynasty origin. I've been a bit Mongolia-focused recently and I've been spending a lot of time with that Bruell/Anderson book; it was fresh in my mind. Many, if not most of the entries in that text seem to compiled with a Turkic sensibility, and I like their understanding of how the ingredients were listed in the text. In many entries, first character(s) is a phonetic stand-in for a Turkic/Mongolian word, followed by a classifier in Chinese. In the case of 酥油, that could be read as "milk oil", which could be fairly easily understood as butter. Many of the text's food words are like this.

    I consulted one of my favorite go-to reference books, the 中国烹辞典. For the entries 苏,酥,酪苏,白酥油 (all cited as synonyms for 酥油), they cite each text from which those terms came (respectively): 本草经集注 and 齐民要术,名医别录,唐本草,饮膳正要. Perhaps the answer might be hidden in one of those earlier texts?

    Certainly Tang Dynasty also had plenty of influences from Turkic and other steppe peoples. Maybe butter became fashionable then via something like 酥油并?

  10. Noel Hunt said,

    March 29, 2020 @ 7:01 pm

    There is a wonderful example of the use of 酥 in reference to a person in Lu You's (陆游) beautiful poem,:钗头凤 chāi tóu fèng (The Phoenix Hairpin?): 红酥手,黄藤酒, hóng sū shǒu,
    huáng téng jiǔ, lit. 'red su hand, yellow wisteria wine' (commentators say that this is just the name of a famous alcoholic drink).

  11. R. Fenwick said,

    March 30, 2020 @ 3:11 am

    @Chris Button:
    Having now looked around a bit, are we sure the meaning is not "cattle, livestock, sheep" rather than "fat, fatty matter" per se?

    I'm no expert on the intricacies of Indo-Iranian, but after a bit of my own poking around it appears you're probably right. Kroll's etymology appears to come from an overly simplistic understanding of the Avestan material; after a bit of investigation in Bartholomae's Altiranisches Wörterbuch, it appears Avestan does have the verb fšav- "to fatten, to allow to become fat" (fšuya- in the present), but the sense of the term remains most fundamentally that of fattening livestock (compare, among others, afšuyant- "not raising/fattening livestock", fšaonay- "fat, ample, teeming (of herds)".

    The question is then whether fšutā– arises from the same root, and if so, what its origin is. It could plausibly be from PIE *pḱu-t-eh₂, formally a feminine/collective passive participle, but PIE *péḱu is basically a root noun AFAIK and I don't know that it can be subject to this sort of verbal formation.

  12. nbm said,

    April 4, 2020 @ 7:02 pm

    A parallel, I submit, is the English usage of "short" in cooking, as in shortbread, short pastry, or shortening. "Short" describes a crumbly, friable texture, a texture achieved with a generous proportion of, what else, (solid) fat to the flour and sugar. So the soft & smooth and the crunchy & crisp are related functionally, one might say.

RSS feed for comments on this post