FOOD & BGVERAGGS, with a focus on naan / nang

« previous post | next post »

The following three items might well have been included in the previous post on Chinglish, but that one got to be rather long and unwieldy, so I'm treating these separately.  In any event, I think that they merit the special treatment they are receiving here.


Just two characters, but they got so much out of them!


The first character is pronounced ròu and means "meat; flesh".

The second character is not very well known.  It has two pronunciations:

a. náng — "a kind of flatbread"

b. nǎng — "stuff food into one's mouth"

Usually náng doesn't have any filling, but this one has a meat filling.

Since I love Indian food and travel in Central Asia quite a lot, the resemblance between Chinese náng, South Asian naan, and Central Asian flatbreads (e.g., among Uyghurs and Kazakhs) with similar sounding names long ago became evident to me.  In my archeological investigations in Eastern Central Asia (Xinjiang / Uyghurstan / East Turkestan), moreover, I encountered well-preserved breads of this type dating to the first millennium BC and first millennium AD, so naan has been around in Central Asia for a very long time, and it's conceivable that it might even have been in Tocharian.

I decided to take advantage of the occasion of writing this post to try to determine the origin of the widespread name of this far-flung flatbread.  With the help of many friends, the history of the word has now become much clearer.

I list the discussion as it developed in several groups, so that the results become more precise, detailed, and extensive as they go on. The discussion on naan / nang is also somewhat interwoven with comments on another widespread Central Asian and East Asian food (dumplings or buns) made with wheat flour called manti, manty, mantu, mantou, etc.

Gene Anderson:

Manti is pretty clearly Turkic, not Chinese.  It appears suddenly in China around Tang (maybe a bit earlier) with a folk etymology.  No Chinese word circulated that widely that early with so little change–manty shows up early in Greek and so on.

Nan, as far as i know, is the Iranian pronunciation of pan, the standard IE bread word, but I don't know.  I'm not sure.  I'd be interested in hearing more.

Stefan Georg:

It doesn’t seem to have a good Proto-Iranian or Indo-Iranian etymology as far as I can tell, but it can be traced to Soghdian, where it is naƔn, so it has been around for a while, and you are certainly right about its origin in that direction.

Marcel Erdal:

Naan is 'bread' already in Pahlavi, and nagan in Eastern Iranian languages including Soghdian. See also J· Harmatta: 'Three Iranian Words for 'Bread", AOH 3,1953, 245-283.

Mehmet Olmez:

Even mantou is originally a Chinese word, but it is not borrowed direct from Chinese. I guess it was borrowed from Mongolian after 13th century. I have a short paper about it (4 years ago I read it somewhere in Turkey and at Arxan northeast Nei Menggu [Inner Mongolia]). Mantou (in Turkish mantı) is a famous local food from Kayseri. Kayseri and around Kayseri was the main basement of Mongolian army. (If I am not wrong) the word man (may be men) occurs for the first time in Uigurica I (F. W. K. Müller, 1. Die Anbeutung der Magier).

Brian Spooner:

I've always assumed that the word was Persian, even if the form of bread may be older.  I have not seen any earlier words for it.

Maria Fasolo:

According to the etymological research I've done, the term "naan", meaning flat bread, goes back to PIE "neogw", having the sense of "naked, unclothed", via Old Persian *nagna, also meaning "unclothed". Naan bread was apparently so called because of its beige skinlike colour and due to the probability that, at least originally, it was eaten as is, without any kind of "dressing": plain naked bread, so to speak. Another explanation of the name is that it was baked uncovered, without ashes being placed on it.

Compare OP "nagna" with Avestan "magna", Sanskrit "nagnà", all meaning "naked, unadorned".

The Hittite form of the word for bread was "ninda".

Nicholas Sims-Williams:

Bailey writes as if *naγna- is somehow derived from *nikana-, but this would contradict everything we know about the historical phonology of the languages concerned. So far as I can see, forms such as Armenian nkan and Sogdian nγn- cannot be connected (except perhaps as a result of some secondary association). Is it possible that *naγna- 'bread' is a specialised use of *naγna- 'naked'? Maybe that's what Harmatta suggested — I don't remember exactly and don't have the article to hand.

It's also worth noting Skt. nagnáhu- 'yeast', which Mayrhofer, EWAiA, II, 6, regards as an Iranian loanword.

Leopold Eisenlohr:

I would be interested to hear how this discussion develops, but I don't think I can contribute much. I looked at some old Persian and Avestan dictionaries, as well as works on Turkic-Iranian contact, and did find some references to naan in its various forms (such as nagan, nagn). I quickly skimmed the Anbetung der Magier in Uigurica but did not see man or men (I was half expecting it to appear as Jesus's body, since it is in the story of the Magi visiting Jesus). I looked in a few other Uyghur lexicons but to no avail.

This probably isn't useful, but there are two entries for 'bread' in Douglas Adams's dictionary of Tocharian B. The possible connection of kanti with 'knead' might mean that it could tend to an initial n- sound like in naan, but that's pure speculation. I don't think the m in mantou could ever become the k in kanti, though.

kanti* (n.) ‘± bread’
[-, -, kanti//] kr[e]nta śwatsanma kanti tänktsi ārwer yāmormeṃ ‘having made ready good things to eat, even bread’ (375b5), [ka]ntiś yikṣye masa o[k] t[o]m‘flour for bread went, eight tom]’ (433a2). ∎Etymology uncertain. VW (187) suggests a connection with PIE *gnedh– ‘press together’ [: Old English cnedan ‘knead,’ OCS gnesti ‘press,’ Old Prussian gnode ‘trough for kneading bread’ (P:371)] but the semantics are hardly compelling.

śro-kanti* (n.) a kind of bread
[-, -, śro-kanti//] śro-kant[i]ś yikṣye ‘flour for śro-kanti’ (433a16). ∎A compound of śro, of unknown meaning, and kanti ‘± bread,’ q.v.

In the Irq Bitig, the word "mäŋ" appears in the sense of "food (generally); game; prey." That is mentioned in one of the entries below. You might be interested to see the Chinese derivation from mian. I am also including Clauson's entry for ban-, 'to dip,' which often occurs with breads. The most common Turkic word for bread is etmek. I did see the form n'n in Manichaean Turkic glossaries as well.

Here are two potentially interesting entries in Clauson, An Etymological Dictionary of pre-thirteenth-century Turkish (Oxford, 1972):

Peter B. Golden:

It has been an interesting discussion. Naan, of course, is completely outside of the Turkic sphere, although a convincing Iranian/Indo-Iranian etymology has yet to be produced.

On Turk. mantı : E. Turkistan (Uyghur and dialects), Central Asia (Uzbek, Qara Qalpaq and Ughur dialects brought to the region, Qazaq, Kyrgyz), Volga-Ural Turkic (Tatar, Bashkir), Oghuz (Turkey Turkish Turkmen, but not Azeri) and one Siberian Turkic language (Tuvan manchı, probably a loanword there from Mongol). There is a good entry on this term in A.V. Dybo (ed.), Etimologicheskii slovar’ tiurkskix iazykov. Obshchetiurkskie i mezhtiurkskie leksicheskie osnovy na bukvy L.M.N.P.S. (Moscow: Vostochnaia literatura RAN, 2003): 30-31, which also points to its presence in Manchu-Tungusic (see V.I. Tsintsius (ed.), Sravnitel’nyi solver’ tunguso-min’chzhurskikh iazykov (Leningrad: Nauka 1975), I:528-529, which mentions a “Proto-Mongol” [Kitan form, perhaps Janhunen’s “Para-Mongolic” would be better] mantan. Clauson (Etym. Dict. Pre-Thirteenth-Century Turk.) does not know it, but has (p. 766) Uyghur mén (min?) < Chin. 麵 (miàn or other characters?).

I recently published an article on “comestibles” in Mahmud al-Kashghari, which notes several types of breads or bread-like foods. [VHM:  I can send the pdf to those who are interested.]

Alexander Vovin:

Russ. mandy is a dialect form, we do not use it in the literary standard. It is a Turkic loan, although the ultimate source is, of course, Chinese. The standard Russian word is pel'meni (nom. plur.), nom. sing is pel'men' (practically never used except in situations like: 'Eat the last pel'men'!'. The word is borrowed either from Komi-Permyak or Udmurt pel'n'an', which is a compound consisting of pel' 'ear' + n'an' < Ir. nan 'bread' that you also cite. This etymology you can find in Vasmer's etymological dictionary of Russian vol. III, p. 230 of the Russian translation, I don't have the German original), or with more details in V. I. Lytkin and E. I. Guliaev's Kratkii etimologicheskii slovar' komi iazyka, Moscow: Nauka, 1970, p. 219.

Although I don't have any substantial Tocharian lexicographical materials at home, like Adams' dictionary of Tocharian B, I think that Marcel is right: the word nan must have an Iranian origin. Certainly it cannot be Turkic because of the initial n-, and cannot be Chinese, either, for obvious reasons.

Francesco Brighenti:

The Middle Persian word nān ‘bread’ and its Iranian cognates do not seem to have been originally borrowed from Tocharian. Bailey, in his Dictionary of Khotan Saka (p. 179), lists a word nāṃji (hapax) which he translates as ‘bread’. Bailey derives both Khotanese Saka nāṃji (< *nānači-) and Middle Persian nān from Old Iranian *naγna– < *nak(a)na-, older *nikana-, from ni– and kan– ‘to put down (into the ashes)’. If there had been any possibility to derive the Khotanese Saka word from the neighboring Tocharian languages as a loan, I think Bailey would have mentioned it.

J. Harmatta, “Three Iranian Words for ‘Bread’”, Acta Orientalia Academiae Scientiarum Hungaricae 3 (1953): 245-283, available here, has a lengthy discussion of this Iranian etymon. He concluded there might be two different Old Iranian words, *nikana-/*nikāna– ‘covered bread’ and *nagna– ‘naked bread’, which would refer to two different early methods of baking. Here is the summary of Harmatta’s linguistic hypothesis found in the article ‘Bread’ (by H. Desmet-Grégoire) in the Encyclopædia Iranica (see here):

In the Iranian languages the words for ‘bread’ inherited from Old Iranian seem to reflect two different early methods of baking (Harmatta). Harmatta suggests that the practice of baking bread ‘covered’, that is, in ashes, is reflected in the word naγan, found especially in the eastern Iranian languages (Sogdian, Baluchi, Pashto, etc.) and Armenian nkan, a loanword from Iranian (Parthian), which must be derived from Old Iranian *nikana– (lit., that which is buried or covered) […]. On the other hand, the practice of baking bread ‘uncovered’, in an oven, seems to be reflected in the common Persian word nān (Mid., NPers., and western Iranian dialects), probably derived from Old Iranian *naγna– ‘naked’.

P.O. Skjærvø, “An Account Tablet from Eighth-Century Khotan”, Bulletin of the Asia Institute 15 (2001): 1-8, available here (see pp. 3-4), considers it more likely that all the words for ‘bread’ discussed by Harmatta derive from a Middle Iranian form *nigan-, “which was borrowed widely in various local forms – an Iranian Wanderwort.”

A further complication is represented by the existence of an Old Indo-Aryan word, nagnahu ‘yeast, ferment’, which has long been regarded as an Old Iranian loan in Vedic Sanskrit. The old theory, according to which the Sanskrit word was borrowed from Iranian *nagnaxvād– ‘bread seasoning’ (see Mayrhofer, KEWA II: 126 and EWAia II: 6), is questioned by A. Lubotsky in his article “The Indo-Iranian Substratum”, in C. Carpelan, A. Parpola & P. Koskikallio (eds), Early Contacts between Uralic and Indo-European: Linguistic and Archaeological Considerations (Helsinki: Societé Finno-Ougrienne, 2001), 301-317, available here (see pp. 3, 6, 7 and 10 in the pdf).

Lubotsky hypothesizes the existence of a Proto-Indo-Iranian word, *nagna– ‘yeast, bread’ from which both Sanskrit nagnahu (< *nagnajhu-, with unusual suffixation) ‘yeast, ferment’ and Proto-Iranian *nagna– ‘bread’ would have derived. According to him, the Proto-Indo-Iranian word in question would be an item of agriculture borrowed from the BMAC* language(s) of southern Central Asia in the late third – early second millennium BCE. Thus, an origin of this word in (southern) Central Asia is not ruled out even in Lubotsky’s linguistic hypothesis.

VHM:  *Bactria–Margiana Archaeological Complex

See also the following post for foods made of wheat (noodles, breads, dumplings) in Central Asia and their analogs in China:

"Pulled noodles: Uyghur läghmän and Mandarin lāmiàn" (8/8/14) (the comments are particularly valuable)

A note on mántou 饅頭 / 馒 头 ("steamed bun / bread / roll").  As late as the Ming period (1368–1644), this term still referred to a meat-filled dumpling.  See, for example, the 27th chapter of the novel Shuǐhǔ zhuàn 水滸傳 (Water Margin; All Men Are Brothers).  It already had this meaning as early as the Tang period (618-907), where we find it being used as a metaphor for a tumulus in the vernacular poetry of Wáng Fànzhì 王梵志 (Brahmacarin Wang).  The word mántou 饅頭 also occurs in the Dunhuang vernacular story misleadingly given the title "Hán Qínhǔ huàběn 韓擒虎話本" ("Prompt book for the Tale of 'Catch-tiger' Han") by modern scholars (see Victor H. Mair, T'ang Transformation Texts (Harvard, 1989). The second syllable of the word, –tou (lit., "head"), is a vernacular noun-forming suffix which would probably not have been present in Sinitic much before the time of Wang Fanzhi, at least not in written sources.

Late addition:

Already in the Song (960-1279) and Yuan (1271-1368) periods, we have the word bāozi 包子.  As mántou 饅頭 over the centuries evolved from being a dumpling with meat filling to merely being steamed bread, bāozi 包子 has come to serve as the word indicating a steamed stuffed bun.


gānzhà yángpái 干炸羊排 ("crispy [deep-]fried lamb chops")

They mistakenly got "Do" from reading 干 as gàn instead of as gān ("dry").

They got the notion of "blowing up" from zhà 炸, which can mean "fry; bomb; explode; blow up; burst; blast; flare up".  Gànzhà 干炸 literally means "dry-fry", but — as we know from previous encounters — the simplified character gān / gàn 干 (which collapses the three traditional characters gān 干, gān 乾, and gàn 幹 into one) can mean so many different things that often get mixed up:  "dry; stem; do[ing]; trunk of a tree; dried food; shield; village; manage; work; have to do with; be implicated in; oppose; offend [against]; bank of a river; lay; put; concern; be concerned with; empty; hollow; able; f*ck"; etc., etc. — see also here).  At least they didn't say "f*ck-explode".

Yáng 羊 is "sheep", "goat", "lamb", so they couldn't mess that up too badly.

Pái 排 is "row; line; platoon; series; discharge; expel; arrange; exclude; rehearse; put in order" and part of the words for "spareribs; pork / lamb chops; steak").


The Chinese says simply:

yǐnshí shēnghuó 飲食生活 ("food and beverage life")

It's hard to imagine how or why "beverages" threw them for such a loop, especially since the sign is so well made and the woodwork is so nice.  You'd think they'd be more careful with the lettering too.

[Thanks to Fangyi Cheng, Maiheng Dietrich, Rebecca Fu, and Erika Gilson]


  1. john burke said,

    February 13, 2016 @ 12:09 am

    Possibly mixing up upper-case G with lower-case e?

  2. Keith Rhodes said,

    February 13, 2016 @ 8:57 am

    Keema naan is well known in Britain; it's a naan with minced meat inside.

    As an aside, I have often seen in Indian restaurants in the UK the term "staff" or "staffed" used to qualify the noun "paratha"… I've always assumed it was a mistake for "stuffed".

  3. Gene Anderson said,

    February 13, 2016 @ 11:32 am

    I think mantou is borrowed from Turkic manty/mantu rather than the other way round. It's fairly late in Chinese (known from around Tang times). It's folk-etymologized ("barbarian heads," with folktales to explain that), unusual for originally-Chinese words. And it's all over Eurasia (borrowed into Greek, for instance) from pretty early, which is not at all usual for Chinese words. It's in most (if not all) the Turkic languages, with expected sound changes for a word of considerable antiquity.

  4. julie lee said,

    February 13, 2016 @ 12:48 pm

    Just a few days ago here (my neighborhood market in the San Francisco Bay Area) they were handing out tasting morsels of cooked Afghan "mantu" , stuffed dumplings, which are called "jiaozi" in Mandarin and "gao" in Cantonese. The mantu are sold in frozen packages. The Chinese "mantou", as Victor Mair has pointed out, has no filling.

  5. Victor Mair said,

    February 13, 2016 @ 1:15 pm

    From Craig Melchert:

    NINDA is merely the Sumerian word; we have so many names for different kinds of bread and cake in Hittite (Harry Hoffner wrote an entire book* discussing them) that we cannot tell what the generic word was., since it is hidden behind the Sumerogram. The idea of naan from 'naked' is interesting.

    *Harry A. Hoffner, Alimenta hethaeorum, New Haven: AOS 1974.

  6. Victor Mair said,

    February 13, 2016 @ 1:47 pm

    From Thomas Allsen:

    Like others, I always assumed naan was Persian. Had no idea about the spread of this word and its cognates across Eurasia, which, of course, argues for its antiquity among a sedentary people growing wheat. As a guess, I would exclude the nomads and others who grew millet. The latter has many attributes and uses (fodder, gruel and beer), but to the best of my knowledge it is never used in bread products.

  7. Endymion Wilkinson said,

    February 13, 2016 @ 8:35 pm

    Gene, mantou first appears in Chinese 300 years before the Tang (in the Ode to Pasta, ca. 300 CE). Were there other Turkic loans at this early date?

  8. Endymion Wilkinson said,

    February 13, 2016 @ 8:44 pm

    Victor, I don't think the man in mantou was written 饅 until the ninth century. Prior to that it first appeared as 曼 (later 漫 and finally 饅).

  9. Victor Mair said,

    February 13, 2016 @ 9:38 pm

    From Peter B. Golden:

    Mantou (Turk. mantï) as a borrowing from Turkic into Chinese, as Gene Anderson suggests, seems a bit problematic to me – not impossible, but problematic. There are no roots to which one can point, or derived terms.

    Words with initial m- in Old Turkic are relatively rare (see Clauson’s Etym. Dict. of pre-thirteenth century Turkic . It is more common in Modern Turkic languages, there is a regular b~m shift that occurred historically, e.g. Old Turkic ben “I” has become men in almost all Turkic languages (except, interestingly enough, the Turkish of Turkey, but in eastern Turkish -the language progressively become more like Azeri and in neighboring Azeri it is men). What was interesting for me in Kasghghari was the large number of terms for different kinds of breads, and grain-based products. Kashghari does not spell out whether he was describing the diet of the elite, or urban-based population, or foods common to all. The nomads, I suspect, were more dairy and meat-oriented.

  10. Rubrick said,

    February 13, 2016 @ 10:33 pm

    This probably isn't useful, but there are two entries for 'bread' in Douglas Adams's dictionary of Tocharian B.

    Being only familiar with the other Douglas Adams, and "Tocharian B" sounding like a plausible Hitchhikerism, I at first thought this sentence was truly stretching the bounds of "probably".

  11. Lana said,

    February 13, 2016 @ 10:38 pm

    There's been at least one case of 包子 being translated as "boson":

  12. Victor Mair said,

    February 13, 2016 @ 11:00 pm

    Taking into account the latest comments by Gene Anderson, Endymion Wilkinson, and Peter Golden, together with my remarks in the original post about medieval vernacular forms and meanings of the word mantou, as well as the sum total of accumulated comments thus far, I'd like to suggest that they can be reconciled by recognizing the following:

    1. Although Chinese mantou may not have been a borrowing from Turkic mantï, it did not originate in Sinitic, but rather it came from some other language into Chinese.

    2. Where precisely Turkic mantï came from is still up in the air. Conceivably it could have come from Chinese mantou, but it could also have come from some other language which first borrowed it from Chinese and then passed it on to Turkic, OR, it could have come from the same or collateral source from which Chinese borrowed it.

    3. The fact that the man– syllable of mantou was first written with two different characters that have nothing to do with food, but only in the 9th century acquired the food radical, strongly indicates that the earliest forms were transcriptions of some foreign word. Although we don't yet know precisely what that word might have been or what language it may have been a part of, I suspect that it was related to an Indo-European word meaning "wheat", for which see around the middle of this post.

    I suggested there that the Chinese word for "flour" (miàn 麵) may have come — through nasalization — from the same source. The suffixation of the nounal –tou syllable was both to turn the transcriptional man– into a Sinitic vernacular binom, mantou, and to convey the notion of a lump-like object. The still later addition of the food radical (semantophore) 食 to the man 曼 phonophore was to make what was originally transcriptional seem more like an indigenous morpheme.

    Consequently, I agree with Gene that mantou did not originate in Sinitic, but I also agree with Peter and Endymion that it probably came from some pre-Turkic source.

  13. Victor Mair said,

    February 13, 2016 @ 11:36 pm


    That's really funny.

    What makes it even funnier is that the menu actually says qí bāozi 奇包子 ("odd steamed, stuffed buns"), which, if they wanted to go with "bosons" for bāozi 包子, they should have called "odd / strange bosons". Perhaps they were prompted to think of strange particles in quantum mechanics because of the qí 奇 ("odd; strange") and the nounal suffix -zi 子 at the end, which is often diminutive. But the Chinese for "boson" is bōsèzi 玻色子 (note that it has a final -zi 子), not bāozi 包子, which means — as we have seen — "steamed, stuffed bun".

  14. January First-of-May said,

    February 14, 2016 @ 11:35 am

    Of course, in modern Russian, manty (манты) is a separate word for a (subtly different) meat dumpling product. Very tasty.
    The word sounds plural (it has a typical Russian plural ending), and grammatically it kind of is, but if your Turkic versions are correct, it was probably borrowed directly that way. Which might explain the confusion of modern Russians as to what the singular form is, when it comes up (the closest to standard is мант, I believe; my father says мантышка, which is clearly humorous).

  15. Cyberiagirl said,

    February 15, 2016 @ 12:20 am

    I don't understand the mentions of Mongolian. I only know modern Khalkha Mongolian, but the only time I've heard 'mantou' is when describing a steamed bun (бууз/buuz) as being Chinese in style. The word I've heard used for flatbread in Mongolian is бин/bin.

  16. Eidolon said,

    February 16, 2016 @ 3:45 pm

    "I suggested there that the Chinese word for "flour" (miàn 麵) may have come — through nasalization — from the same source. The suffixation of the nounal –tou syllable was both to turn the transcriptional man– into a Sinitic vernacular binom, mantou, and to convey the notion of a lump-like object. The still later addition of the food radical (semantophore) 食 to the man 曼 phonophore was to make what was originally transcriptional seem more like an indigenous morpheme."

    The Chinese root for flour being a heavily nasalized version of the Indo-European word for wheat, if true, still does not make *mantou* an Indo-European word, however. If -tou was indeed added within the context of forming a Sinitic vernacular binom, then the word is at best a Sinitic word with a heavily nasalized Indo-European root, as opposed to a word that actually existed in Indo-European. This is significant because it affects how we view the spread of the word into other languages. If it's via Chinese *mantou*, then we'd expect those languages to reflect the actual structure of the Chinese word, with attendant modifications to fit their phonology and morphology. If, however, it's via an Indo-European word cognate to Sinitic *mantou*, then we don't actually know what that word is, but can perhaps reconstruct it from the borrowings in other languages.

    This is a historically tenable problem, not a hypothetical, as if the term is Indo-European, then besides Greek there must be other Indo-European examples of similar foods both in an etymological and a culinary sense. Perhaps Vovin's *mandy* could be an example, unless its Turkic etymology is proven? My own feeling on the matter is that, unless we have an Indo-European word similar to *mantou/manti*, the borrowing from Indo-European does not make much sense, as no known root for wheat in Indo-European sounds similar to *man* while the similarity between Turkic *manti*, Para-Mongolic *mantan*, and Chinese *mantou* is evident.

  17. Peter Taylor said,

    February 16, 2016 @ 5:10 pm

    At some point in the various references to words beginning man my thoughts turned to the Spanish manjar (meaning food in general, and delicacies in particular), the Catalan menjar (to eat), and the French manger (likewise). But these are false cognates of the man words discussed here; although some claim that the Spanish manjar comes from manna, the majority opinion seems to be that it's from Latin manducare.

RSS feed for comments on this post