Water chestnuts are not horse hooves

« previous post | next post »

One of my favorite ingredients in Chinese cooking is the crunchy water chestnut, but it always puzzled me that the name for this item is mǎtí 马蹄 / 馬蹄.  Although technically it's not a nut (it's the corm of an aquatic vegetable) and doesn't really look like a horse hoof, I tried to convince myself that maybe there was some sort of resemblance between the two after all.

It turns out that, while on the one hand mǎtí 马蹄 / 馬蹄 really does mean "horse hoof" and just happens to be the title of a chapter [the 9th] in my favorite early Chinese book (Zhuang Zi / Chuang Tzu / Wandering on the Way), on the other hand it also has a completely different etymology when applied to the water chestnut.  Namely, it is borrowed into Mandarin and other Sinitic topolects from Cantonese maa5 tai4-2, maa5 tai4, where it is the transcription of a Kra-Dai substrate word (Li, 2012) (compare Zhuang makdaez).  Source.  I became even more hopelessly confused when I learned the derived Cantonese expression maa5 tai2 fan2 馬蹄粉 and thought that, well, this must be some sort of gelatin made from horse hooves (but that's just an urban legend anyway), when in truth it's simply water chestnut starch.  This is but one example of how Chinese characters frequently lead us seriously astray when it comes to understanding the derivation and meanings of Sinitic words.

In various parts of China, the water chestnut has many different names (see here).  And here is an amazing dialect map showing the locations of 31 different names for this plant, all dutifully, but misleadingly, written in Chinese characters.  They all seem to be quite different names when written in Chinese characters, but when you look beneath and behind the characters to their sounds, probably more than three quarters of these names derive (at least in part) from the Cantonese borrowing of the Kra-Dai substrate term.  That's why I always tell my students, including those who are learning Literary Sinitic (LS) / Classical Chinese (CC), that one needs to keep the sounds of Sinitic words uppermost in mind when reading texts, including very old ones, and not get too hung up on the characters.

I also knew the water chestnut as bíqí (some people pronounce it as bíqi, bíjì, bóqí, etc.), though I didn't know how to write it in Chinese characters.  Now I find that it is written 荸荠.  An etymological note for this term in Wiktionary reads as follows:

From 鳧茈 (MC bɨo d͡ziᴇ). According to the Compendium of Materia Medica:

爾雅鳧茈鳧茨荸薺[▼ expand/hide]

Mallards like to eat it, so Erya names it fúcí [lit. "mallard gromwell"], which later corrupted to fúcí and further corrupted into bíqi.

Alternatively, it may be derived from 脖臍 (MC buət̚ d͡zei, "navel"), perhaps because its corm resembles a navel or because its stolon resembles an umbilical cord (Wei, 2004).

In the face of such wild speculation, I suspect that all of these variants are ultimately cognate with the Cantonese < Kra-Dai word for the plant.

It would be easy to write a whole long paper on the numerous Sinitic words for water chestnut, but I think the main principles of their origins have already been laid out above.  It would be even more confusing if we got into words for the water caltrop (see also here), some of which are shared with those for water chestnut, though the plants belong to completely different species and genera.

I was finally disabused of my half-century of confusion about the feeble explanation that a water chestnut "looks like a horse hoof" yesterday when I was reading a book by the Polish Sinologist and linguist, Mieczysław Jerzy Künstler, The Sinitic Languages:  A Contribution to Sinological Linguistics, tr. Mieczysław Jerzy Künstler and Alfred Franciszek Majewicz, Collectanea Serica, New Series, 1 (Sankt Augustin:  Monumenta Serica Institute, 2019), where, on p. 31, he gives the true etymology of the word by suggesting that it comes from Cantonese, which got it from Thai or perhaps Zhuang.  Künstler is not responsible for all that I've written in the preceding paragraphs, but I gratefully acknowledge that it was his suggestion that sent me scrambling to do the research that resulted in what I've presented above.

Reading Künstler's book also gave me great comfort when I found that he naturally and unselfconsciously uses the word "dimidiation" to describe the splitting of an initial consonant cluster by adding an intercalary vowel between two consonants of the cluster to create a disyllabic unit (p. 80 and passim).  He gives the classic example of this process:  kūlóng 窟窿 ("cave; cavity; hole") < k'long (Wiktionary refers to this as the disyllabic form of (OC *kʰloːŋʔ, "hole; aperture") and offers a number of other instances of this process.  I mention this because, when I used the term "dimidiation" in earlier Language Log posts (it's easy to find them by doing a Google search), some readers commented that they had never heard of it.

Among the many other delights garnered from reading Künstler's The Sinitic Languages, I will mention here only two:


In a shop in Peking, I once happened to see jidan in the intended meaning of eggs, written with characters meaning "some, few" (ji) and "dawn" (dan), which is evident nonsense.

[VHM:  That reminds me of John DeFrancis's famous demonstration of a Ph.D. forgetting how to write the characters for "egg" and many other simple terms; see "Dumpling ingredients and character amnesia" (10/18/14).  BTW, Künstler does not use a single Chinese character in the entire 275 pages of his comprehensive treatment of all aspects of the synchronic and diachronic development of Sinitic languages.  I warmly applaud his decision to omit them, since it was his intention to focus on the languages, not the writing systems used to record them.]


"We are Chinese."

In Modern Standard Chinese it is pronounced:

Women shi Zhongguoren.

In Cantonese it goes as follows:

Ngutei hai tongyan.

The phrases share only one common character, namely the last one:  Both MSC ren and Cantonese yan mean "man" and are etymologically (i.e., genetically) the same.  All other characters are different.  This proves that is is not true that dialects differ only in phonetics, as Chinese researchers have often suggested.

Künstler delivered the lectures on which The Sinitic Languages is based in Polish at the University of Warsaw in the early 1990s, so he would not have known about the word "topolect" which I coined during the 80s and began to popularize during the early 90s.  Judging from his strong emphasis on the varieties of modern Sinitic as constituting separate languages, not just dialects, as well as his profound learning concerning hundreds of fāngyán 方言 across the length and breadth of China, I'm confident that he would have both understood and heartily approved of what I meant to achieve by creating the word "topolect" as a precise English equivalent of the Sinitic term.


  1. John Rohsenow said,

    June 29, 2019 @ 6:09 pm

    Keeping in mind your praise of Kunstler for not using characters, could
    I nevertheless inquire as to the characters used to write "Ngutei hai tongyan" ('We are Chinese') in Cantonese? Are they easily accessible in a standard character database? Are there any Mandarin to Cantonese or
    English to Cantonese conversion programs using either Cantonese romanization and/or Cantonese characters?

  2. Chris Button said,

    June 29, 2019 @ 7:05 pm

    It would be even more confusing if we got into words for the water caltrop (see also here), some of which are shared with those for water chestnut, though the plants belong to completely different species and genera

    We do incidentally have a depiction of a water caltrop (菱) in the oracle-bone inscriptions. All that's left of it in the modern script is its merged form with 土 as the top component of 夌 (i.e. without the 夂 which was added later) and minus the 儿 (人) beneath it.

  3. Victor Mair said,

    June 29, 2019 @ 9:02 pm

    The Cantonese word for "Chinatown" is Tong4 jan4 gaai1 (MSM Tángrénjiē) 唐人街 — "Street of the Tang People" (not "Han People").

  4. Anthony said,

    June 29, 2019 @ 9:02 pm

    @John Rohsenow
    The phrase "We are Chinese [people]" in Mandarin:
    Traditional) 我們是中國人
    Simplified) 我们是中国人
    Pinyin: Women shi zhongguoren.

    The same in cantonese:
    Ngutei hai tongyan.
    (Wikipedia spells the first word as ngo5 dei6. What romanization is used in that book?)

    I hope my conversion into characters is correct. I did this by searching the relevant bits of the mandarin version through Wikidictionary. I then looked up their cantonese equivalents in the dropdown table in each entry called "Dialectal[!] synonyms of (character)".

    For good measure here is the "Egg" thing = Jidan.
    The correct way to write it :(S) 鸡蛋(T)雞蛋
    The incorrect way: (S)几旦 (T)幾旦

    It is rather amusing to think the Traditional version of the 'misspelling' is even more complex than the 'proper' method of writing it.

  5. Clyde Law said,

    June 29, 2019 @ 9:02 pm

    That Cantonese sentence actually has two characters in common, not one. Rendered in characters and Jyutping romanization, it would be:
    我哋係唐人。 Ngo5dei6 hai6 Tong4jan4.

    Both the first and the last characters are cognate with the Mandarin rendition: 我們是中國人。

    哋 in 我哋 is arguably originally derived from 等, which Cantonese uses as its pronoun pluralizer instead of Mandarin's 們 (derived from 每人). 係 is Cantonese's general copula instead of 是. Finally, southern Chinese people often call themselves 唐人 (Tang people), since southern China was Sinicized during the Tang Dynasty.

  6. cameron said,

    June 29, 2019 @ 9:26 pm

    How would that example sentence "We are Chinese" be pronounced and written in Literary Sinitic/Classical Chinese?

  7. Michael Watts said,

    June 30, 2019 @ 2:37 am

    The Cantonese word for "Chinatown" is Tong4 jan4 gaai1 (MSM Tángrénjiē) 唐人街 — "Street of the Tang People" (not "Han People").

    That's not specific to Cantonese; it's also the Mandarin word.

  8. Victor Mair said,

    June 30, 2019 @ 8:12 am

    Specialist notes From Diana S. Zhang:

    1) Looked a bit more into the Shuowen jiezi 說文解字 for this sweet beloved rhizome — and found 芍 that represented 鳧茈. It seems that 鳧茈 was a commonly known word with one way of writing (let me know if different evidence comes up) before 許慎's time (ca. 100CE). An attestation in historical documents is 後漢書 劉玄傳: "At the end of Wang Mang's reign (ca. 20AD), there was a great famine in the South. Almost all people flocked into the wild marshlands and dug water chestnuts to eat. (王莽末, 南方飢饉, 人庶羣入野澤, 掘鳧茈而食之.)" 李賢 (654-684)'s commentary to this sentence: "Guo Pu (276-324) says (in his Erya that this plant) 'grew in the lower farmlands, with thin sprout resembling dragon's whiskers, root resembling top of the finger; it was black and edible.' (郭璞曰'生下田中, 苗似龍鬚而細, 根如指頭, 黑色,可食.')"

    2) According to the 本草綱目 (as you showed in the post), by the Ming Dynasty, 鳧茨 was already another way of writing 鳧茈. I conjecture that this alteration should have already existed in the mid-Tang time when the 支 rime (which 此 is in) and 脂 rime (which 次 is in) merged. However, the two major Five-Dynaties commentators to the Shuowen jiezi, 徐鉉 (916-991) and 徐鍇 (920-974) did not present this alternate way in their commentaries. Wonder why 鳧茨 did not appear in the 說文解字繫傳. Perhaps out of dialectal reasons?

    For your interest, pages from three versions of 說文 commentaries:
    a. 大徐 (徐鉉): "芍,鳧茈也。從艸。勺聲。胡了切。"

    大徐.JPG [clear facsimile with highlighting available from VHM for those who are interested]

    b. 小徐 (徐鍇): "芍,鳧茈也。從艸。勺聲。臣鍇曰,今人所食鳧茈也。堅烏反。"

    小徐.JPG [clear facsimile with highlighting available from VHM for those who are interested]

    c. 段玉裁注: "鳧茈也。見釋艸。今人謂之葧臍,卽鳧茈之轉語。郭璞云:苗似龍須。根可食。黑色。是也。廣雅云:葃姑、水芋、烏芋也。名醫別錄云:烏芋一名藉姑,一名水萍。藉與葃同音,萍必芋之誤。此專謂茨菇,不必因烏字牽合鳧茈也。茈徂咨切。從艸。勺聲。胡了切。二部。古勺聲與弱聲同,芍之可食者、其蒻也。"

    段.JPG [clear facsimile with highlighting available from VHM for those who are interested]

    From the 段注 above we already see his mentioning of an alternate way of writing: 葧臍. Thus Wikitionary's conjecture that 荸薺 may derive from 脖臍 should be right? Although, I haven't yet looked into how much such a derivation is purely phonetic. In any ways, what grabbed my eyes are two things:

    3) Duan Yucai 段玉裁(1735-1815)'s difference from his contemporary Zhao Yi 趙翼 (1727-1814), both of whom were from the modern 江蘇常州 area — 段玉裁,just like the 大小徐, did not mention 鳧茨 at all in his commentary. But 趙翼, in his poem 曉東小巖香遠邀我神仙館午飯 writes: "君不見古來饑荒載篇牘, 水擷鳧茨野采蔌."

    4) According to Duan Yucai, 廣雅 (by 張揖, ca. 227CE) defines the same plant as 葃姑 — which becomes 茨菇, 茨菰 or 慈姑 (Sagittaria sagittifolia) as written nowadays. These are two different plants (荸薺 is Eleocharis dulcis)! 段玉裁 noticed this and pointed it out on purpose "此專謂茨菇, 不必因烏字牽合鳧茈也". It may just be interesting to look into edible plants from the Han to the Wei time — throughout all the riots, turbulence, immigration, and wars — what did people eat? How was common people's recipe transitioned, with different names of edible plants mixed up, forgotten, and re-established?

  9. Victor Mair said,

    June 30, 2019 @ 8:40 am

    There are excellent articles on the terms for and history of Chinatowns in Wikpedia (English and Chinese).

    For the history of the expression Tong4 jan4 (MSM Tángrén) 唐人, see this Wikipedia article.

    When Chinatowns of the Sinitic diaspora were populated mainly by Cantonese and speakers of other southern topolects, they were usually referred to as "Tang People's Street", but in recent times, as more and more persons from other parts of China have begun to frequent them, the terminology has also been changing, with northerners preferring "Zhōngguóchéng 中國城 / 中国城. Huábù 華埠 / 华埠 (where bù 埠 / 埠 literally means "lpier, wharf, dock; port city; commercial or trading port") is also often heard.

  10. Rodger C said,

    June 30, 2019 @ 9:04 am

    I thought the water chestnut was called "horse's hoof" because its spiky receptacle resembles a caltrop, a medieval weapon designed to trip up charging horses. This entry mentions "caltrop" without invoking the original meaning.

    David Hawkes' translation of the Dream of the Red Chamber includes a servant called Caltrop. I always supposed her name could also be rendered Water Chestnut.

  11. Philip Taylor said,

    June 30, 2019 @ 11:27 am

    "trip up" charging horses surely qualifies as the understatement of the year. The caltrop was intended to inflict agonising injury by penetrating the horse's sole, rendering it instantly lame and therefore hors de combat.

  12. Chris Button said,

    June 30, 2019 @ 2:44 pm

    I thought the water chestnut was called "horse's hoof" because its spiky receptacle resembles a caltrop…

    The original semantics in Chinese seem to have developed similarly. Compare 菱 with 馮 (whose phonetic 仌 seems to be being used for 夌 with a bilabial element) and 凌 via a similar association of caltrop with Latin calcāre "trample, cross (on foot)" from whence Portuguese calcar "trample, subjugate".

  13. Victor Mair said,

    July 1, 2019 @ 7:14 am

    From Diana Shuheng Zhang:

    There are two more small points on this water chestnut post:

    1) About 脖/葧臍 (脖 and 勃 have always been homophonous) versus 鳧茈. 脖/勃 and 鳧 sounded very close to each other twice in the history of literary Sinitic sounds. The first time is the late stage of Old Chinese around the Western Han. In this stage: 脖/勃 *b'ut (type A) VS 鳧 *bu (type B). However, I doubt if there were any attestations of their 對轉 at such an early stage. Then, from around the Eastern Han on, these two sounds grew farther apart, into 脖/勃 bot VS 鳧 bu > vu. Then, with the loss of 入 tone in the Yuan time (yes, this second time came so late!), they sounded similar again — 脖/勃 bo VS 鳧 vu > fu. Therefore, regarding these late-imperial attestations mentioned in my last comment, maybe the 譌變假借 that both 本草綱目 (Ming) and 說文段注 (Qing) quoted, was what occurred in the second time.

    2) Water chestnut has a fun name in the Southern Xiang (xiangnan 湘南) area: 菩栗子. 菩 (MC. bu) should derive from 鳧. Just found this 菩栗子 name quite interesting!

  14. Terry Hunt said,

    July 1, 2019 @ 8:07 am

    As Prof. Mair probably knows but did not mention, 'dimidiation' is a standard term in Heraldry used to describe the practice, dating back to at least the 13th century, of combining two coats of arms — such as those of a husband and wife, or of a town that has become one of the Cinque Ports — by joining the dexter half of one to the sinister half of the other.

    This can result in some amusing charges (pictured items) on the resulting dimidiated shield, such as that of Great Yarmouth showing three half lions-half herrings ("Per pale Gules and Azure three Lions passant guardant in pale Or dimidiated with as many Herrings naiant in pale Argent.") first recorded in 1563.

  15. Michèle Sharik Pituley said,

    July 1, 2019 @ 10:27 am

    Philip Taylor: rendering it instantly lame and therefore hors de combat.

    I see what you did there. Lol

  16. Rodger C said,

    July 1, 2019 @ 10:27 am

    @Philip Taylor: I'd thought of writing "stop charging horses in their bloody tracks," then spotted an ambiguity.

  17. Chris Button said,

    July 2, 2019 @ 9:14 pm

    Philip Taylor: rendering it instantly lame and therefore hors de combat.

    I see what you did there. Lol

    ha ha – yes that is a good one!

  18. B.Ma said,

    July 3, 2019 @ 6:16 am

    It is cromulent to say 我哋係中國人 in Cantonese, but it can mean something different from 唐人. I think someone who lives in Guangzhou and is from there would not hesistate to say 我哋係中國人.

    I am a 唐人 but whether I am a 中國人 depends on who I am talking to and their understanding of what 中國 means.

    In my view, 唐人 are people who are ethnically and/or culturally Chinese. 中國人 can mean the same thing, or it can mean someone from a country that calls itself 中國. The latter can refer to citizens of those countries, or people associated with those countries (usually ex-citizens). I do not think most people from Taiwan really consider themselves 中國人 any more as the word has become linked to the PRC. Hong Kong people who hold HKSAR passports are citizens of the PRC and thus 中國人, but in political terms if they are generally opposed to the Communist Party they do not consider themselves 中國人, while those who support the CCP do.

    I had a conversation about this with a tourist from the PRC on the tube in London recently. He approached me asking for directions, but the first thing he said was 你是中国人吗? I said no, I am a 英国人 and 唐人, but not 中国人. He said that I looked Chinese. I said that I spent a lot of my life in Hong Kong but most of it was when it was not part of the PRC (= 中国), neither of my parents were born in a country that could be called 中国, and if I want to visit the PRC, I need to spend a lot of money on a visa which doesn't let me work or own property or open an Alipay account, so in what sense can I consider myself from 中国?

    Incidentally, 中國 is also a region of Japan, and Japanese people seem to be more understanding that I can be of Chinese heritage while not being Chinese (it may be hard for people not of Japanese heritage to become Japanese citizens, but it happens, while it is close to impossible for the PRC).

RSS feed for comments on this post