Dungan: a Sinitic language written with the Cyrillic alphabet

« previous post | next post »

The Dungan people are a group of Sinitic speakers whose Muslim ancestors fled to Central Asia (mainly in parts of what are now Kyrgyzstan and Kazakhstan) over a century ago when the Qing (Manchu) government suppressed their revolt (1862-1877), one of many Muslim uprisings in the course of Chinese history since Islam arrived in East Asia during the Middle Ages.

When they came to Central Asia, the Dungans were mostly illiterate peasants from northwest China who spoke a series of topolects from Shaanxi, Gansu, and other areas.  From 1927 to 1928, they wrote their language with the Arabic alphabet, and from 1928-1932 they used the Latin alphabet.  In 1952-53, the Soviet government created for the Dungans a writing system based on the Cyrillic alphabet, which they continue to use till today.

The Dungans write stories, poems, and plays using their script, and they have also published newspapers and magazines written in it.  Ever since the Russian Sinologist, Boris Riftin, and the Dungan author, M. Sushanlo, came to Penn in the early 80s, I have been intensely interested in Dungan as living proof that Sinitic languages can be written with an alphabetic script, and in 1990 I wrote an article detailing what I was able to learn about Dungan writing as of that date: "Implications of the Soviet Dungan Script for Chinese Language Reform".

Now, nearly a quarter of a century later, I find myself in Tokmok, Kyrgyzstan at the International University of Central Asia attending a conference on Dungan language and culture that has brought scholars from many nations, including Norway, France, Russia, China, the United States, Kazakhstan, and Kyrgyzstan.  We have enjoyed two days of presentations on a wide variety of topics followed by lively discussions.  Eventually, perhaps within a year, we will publish a volume of papers growing out of this conference, and I'm certain that it will advance the state of the field far beyond what was known about the subject when I wrote my 1990 paper.

Until the book comes out and a more thorough account can be given, I will say only that one of the most important aspects of Dungan writing from my perspective is how easily it has enabled Dungan to borrow words directly from other languages (including Arabic, Persian, and Russian), e.g., Russian traktor instead of Mandarin tuōlājī 拖拉机.  Naturally, it is vastly easier to learn to read and write with the Dungan script than is possible with Chinese characters.

As an added note, I was stunned to learn that the Boston Marathon bomber brothers, Tamerlan and Dzhokhar Tsarnaev, were born in this very city, Tokmok, in the center of Asia, to which their parents had been sent by Joseph Stalin, himself from Georgia, a state in the Caucasus, the same as Chechnya, the homeland of the Tsarnaev brothers.


  1. Avinor said,

    April 20, 2013 @ 5:38 pm

    Interesting that the Cyrillic hard and soft signs are used to mark the tones (if tones are marked at all). An unneeded/unused feature of the original alphabet is used for something completely different when adapting it to an unrelated language. Is this unique? Another possible example I can think of is the Yiddish vowels.

  2. julie lee said,

    April 20, 2013 @ 8:42 pm

    Have you met with any Dungan speakers over there in Kyrghyzstan and can you understand their Chinese speech, which is based on the topolects of Shanxi, Gansu, etc.?
    Can they understand your Mandarin (assuming they have not previously learned Beijing Mandarin) ? Aside from borrowings from Russian, Turkish, etc., does Dungan still sound pretty much the same as the present-day speech of Shanxi, Gansu, etc. ?

  3. Victor Mair said,

    April 20, 2013 @ 9:14 pm

    @julie lee

    There were many Dungan speakers from Kyrgyzstan and Kazakhstan at our conference. I can understand over 80% of what they say, partly because I've had a lot of experience in northwest China and am used to their topolects, plus I also know Russian and some Uyghur, etc., so that helps too. On the other hand, they have more difficulty understanding my Modern Standard Mandarin (Putonghua / Guoyu). The phonology of Dungan overall is still pretty much that of northwest Chinese.

    William S.-Y. Wang, the Berkeley linguist, came out here with one of his graduate students after reading my 1990 article and complained to me that he couldn't understand much of what the Dungans were saying.

  4. Julian Bradfield said,

    April 21, 2013 @ 3:43 am

    "Boston Marathon bomber brothers"? I must have missed the trial that happened overnight.

  5. dw said,

    April 21, 2013 @ 9:32 am

    An unneeded/unused feature of the original alphabet is used for something completely different when adapting it to an unrelated language. Is this unique?

    The Greeks converted unused consonants from the Phoenecian alphabet into vowel signs, e.g. aleph (glottal stop) became alpha.

  6. Mr Punch said,

    April 21, 2013 @ 9:32 am

    Historical fact, Julian, not legal decision. John Wilkes Booth and Adolf Hitler never went to trial.

  7. minus273 said,

    April 21, 2013 @ 9:51 am

    @dw: This is almost certainly an extension of mater lectionis already existing in the system.

  8. julie lee said,

    April 21, 2013 @ 9:56 am

    @Victor Mair

    It must be thrilling to understand 80% of the Chinese spoken by Dungan speakers. When I was in London, I had the hardest time understanding the English spoken by the long-distance operators who had Scottish and Irish accents. I understood about 30-50% of their speech. Had to strain very hard. Very frustrating.

  9. Gorobay said,

    April 21, 2013 @ 12:18 pm


    Javanese, a Brahmic script, does not need all the Brahmic letters, so it recycles the extras as respectful variants of standard letters. It is similar to our capitalization of proper nouns. For example, ꦧ and ꦨ were originally ba and bha; now they are both ba but the latter denotes respect.

  10. Keith Ivey said,

    April 21, 2013 @ 12:24 pm

    Avinor, it seems similar to the use of otherwise-useless consonant letters to represent tones in Hmoob/Hmong.

  11. Vladimir Menkov said,

    April 21, 2013 @ 12:54 pm

    While I am aware that a workable Latin and later Cyrillic orthography
    was created for the Dungan language, and I’ve seen some of the
    published using it during the USSR era (dictionaries, fiction by
    writers such as Yasir Shivaza, school textbooks), I am quite curious
    to which extent this writing system is actually in use today (or have
    been in use previously). That is, on an average day, would an average
    Dungan speaker in Kyrgyzstan or Kazakhstan encounter any writing in
    Cyrillic Dungan? On an occasion that he has to write anything (a
    letter, a note, an email, a shop sign), what are the chances he’d use
    Dungan (as opposed to Russian, Kyrgyz, or Kazakh)?

    There seems to be a fairly active internet Dungan forum (
    http://dungane.kz/forum/ ), but at least the post headers seem to be
    all in Russian.

    I spent some time traveling in Kyrgyzstan in 2007, and even thought I
    passed through some areas where Dungan ought to be spoken, I remember
    *not* seeing a single sign in that language. (Perhaps this sign was
    the only exception:
    ; can anyone read it? ). While Dungan is obviously is a minority
    language, the situation is very different from e.g. that of Yiddish in
    Williamsburg (Brooklyn), where you see Yiddish signs all around.

  12. Pete said,

    April 22, 2013 @ 3:18 am

    @Avinor: When the Phoenician alphabet was adapted for use with Greek, it didn't contain any dedicated vowel letters because it was used for a Semitic language that allowed the vowels to be inferred from the context, like Arabic or Hebrew today.

    That wouldn't work in an Indo-European language like Greek, so the letters for unneeded consonant sounds (aleph ʔ, he h, heth ħ, yod j and ayin ʕ) were used to represent vowels (alpha a, epsilon e, eta ɛː, iota i and omicron o). In the Latin alphabet eta is a consonant again because of regional variations in Greek usage.

  13. Stephen Zweig said,

    April 22, 2013 @ 7:06 pm

    @Victor Mair

    While you were in Central Asia, did you see any dictionary of the Dungan language? According to several paper I've read, the Dungan orthography omit the /n/ final after erhua if there's no /n/ sound in the pronunciation. Second, some paper have even suggested evidence in the use of neutral tones in personal names. Last, can you confirm the phonology in http://www.uusikaupunki.fi/~olsalmi/dungan/Dungan%20as%20Chinese%20Dialect.html? Are the three statements below accurate?

    1. "If the у [u] is medial, the initial consonant usually has no labialization. The labialization is occasionally a bilabial trill (Zav'jalova 1979: 44–45)."

    2. "Aspiration is velar: [px] [tʂx] etc. (Zav'jalova 1979:43)."

    3. "в [v], ж [ʐ] and й [ʝ] are strongly fricated. They are originally medials with a zero initial. In Dungan they are probably best regarded as initials. "

  14. Victor Mair said,

    April 22, 2013 @ 11:38 pm

    @Vladimir Menkov, Stephen Zweig, and others

    Ivo Spira, Hai Feng, Taras Ivchenko, and other specialists on the Dungan sound system will answer you in due course. Please be patient for a few days, however, since they are currently doing fieldwork or are travelling home from the conference and without good internet connections.

  15. anonymous said,

    April 23, 2013 @ 1:20 am

    @Stephen Zweig

    To the best of my knowledge, Zavyalova is the expert in this area, and Salmi (that site in Finland) is very reliable as well, but why not just listen for yourself? A film dubbed in Dungan, with different speakers using somewhat different accents: http://jesusfilmmedia.org/video/1_17820-jf-0-0/dungan/jesus . (If this stalls, look for it on youtube, though the audio quality may not be as good.)

    As for personal names, a list of them appears in the back of this dictionary along with canonical tone per syllable: http://ibt.org.ru/russian/bible/dng/DNg_15.pdf . Bear in mind, though, that tone roman numeral 'I' may actually be either of two different pitches — this first made known by Zavyalova, I believe. It seems to me though that the question should be re-phrased, into: 'Is neutral tone any more prevalent in names than in other Dungan words?' Someone in the know will have to answer.

  16. JS said,

    April 23, 2013 @ 8:19 am

    ^ or see this CCTV 13 segment from not so long ago on the Dungan.

  17. An Hussey dungan said,

    April 23, 2013 @ 1:29 pm

    г.Токмак имеет оченб древнюю историю – в начале нашей эры здесь процветал древний Суяб: столица Западно-тюрского каганата.
    Tokmak ochenb has an ancient history – the beginning of our era there flourished the ancient Suyab: the capital of the Western Turkic Khaganate.

  18. Eric Vinyl said,

    April 23, 2013 @ 5:51 pm

    Hooray! Victor Mair with a post on one of my favourite subjects ever! (a Mandarin dialect written in Cyrillic.)

    Professor Mair, you rule! (once again.)

  19. haiyun said,

    April 24, 2013 @ 7:55 am

    I am a Hui from NW China and I met several Dungans who preferred to be called as Huizu, which surprised me.
    I can communicate with them in NW dialects (Gansu-Shaanxi) except some Russian words. Their old Chinese words seem to be very old-fashioned.

  20. Julien said,

    April 27, 2013 @ 1:05 am

    "a conference on Dungan language and culture that has brought scholars from many nations, including Norway, France, Russia, China, the United States, Kazakhstan, and Kyrgyzstan."

    Really, who in particular?

  21. Victor Mair said,

    April 27, 2013 @ 5:55 am


    For starters, participants included Christoph Harbsmeier and Ivo Spira from Norway / Germany, Françoise Bottero from France, Taras Ivchenko from Russia, Mukhame Khusezovich Imazov from Kyrgyzstan, Hai Feng from China, Hussain Daurov from Kazakhstan, and VHM from USA. There were many others, but this should give you an idea of the number and quality of those in attendance.

  22. Jonathan Gress-Wright said,

    April 27, 2013 @ 10:43 pm

    I was surprised to read that borrowings into Dungan are more faithful to the original forms than borrowings into Mandarin, and that this is due to the different writing system. I'm used to thinking of loanwords as being constrained by the phonology of the borrowing language, but not by the writing system itself. Are these loanwords specifically mediated through writing, e.g. are they typically the kinds of words that are learned in school rather than in the normal course of acquisition?

  23. Ivo Spira said,

    April 28, 2013 @ 2:18 am

    @julie lee

    I find that mutual intelligibility between Dungan and Modern Standard Mandarin (MSM) depends on a range of different factors.

    As usual, the actual degree of success in communication depends on the ability and willingness of the interlocutors to adapt their speech to that of the other person, and so a Dungan doing business in China will naturally shift his speech towards MSM and/or the relevant local topolect, just as a MSM speaker will have to get used to the peculiarities (from his point of view) of local speech in a Dungan community. Given the power imbalance between Dungan and MSM, it is more and more the case that Dungan speakers shift their speech in the direction of MSM. One can hear anything from the occasional use of words from MSM (D. дянхуа "phone" < MSM diànhuà 電話 is by now widely used) to full acquisition of MSM (students in China). It is much rarer for a MSM speaker to speak Dungan as such, the major exception being speakers of NW Chinese, who, just as @haiyun observes, can easily communicate with the Dungans, in spite of certain obstacles such as Russian words and literary Dungan neologisms.

    The lexical choices made by the speakers greatly influence intelligibility. If the Dungan speaker uses a lot of words from Russian, Turkic, Persian etc. a MSM speaker unfamiliar with with these languages and northwest Chinese will have a hard time. The same goes for late Qing dynasty words that are no longer used in spoken MSM, as well as for neologisms. This means that the topic of conversation is a very important factor: the contrast between a conversation about China and Chinese products and a conversation about religious customs in a Dungan village is significant.

    The degree to which I myself understand non-putonghua'ed Dungan speech varies very much according to the topic of conversation and the distinctness and speed of the speaker's diction. In some cases I cannot really follow the speaker, in other cases I understand practically everything. As someone who speaks Russian and is familiar with a range of Middle Eastern languages, the greatest challenge for me is not non-Sinitic loanwords (such as the common дун-я "world" < Arabic dunyā), but rather NW Chinese dialect words and certain archaisms and neologisms (which category does та тўён "take a photo",corr. to 拓圖樣, fit into?).

    @Vladimir Menkov

    I have a similar impression. So far I have only seen it actively used in works of fiction, in the Dungan newspaper (1-2 pp in Dungan), and in the lessons of Dungan language and literature in schools (e.g. in textbooks, on the blackboard). Signs on house gates announcing the availability of certain products are marked in Russian ("молоко"). Works by Dungan scholars may quote passages or words in Dungan, but they are by now entirely in Russian (this contrasts with J. Janshansin, for example, who wrote several linguistic works in Dungan).
    The restricted use of Dungan writing is certainly related to the decline in Dungan-language education and the ongoing process of language shift towards Russian (and to some extend towards MSM as well) in the Dungan population.

    I'm not sure about how to read the sign, I'll get back to you after consulting a native speaker. The orthography is somewhat peculiar.

    @Stephan Zweig @Victor Mair @anonymous

    I am certainly no specialist on the Dungan sound system, the current authorities are Zavyalova, Salmi, Hai Feng, and Lin Tao, not to mention Imazov himself, although one may not agree with his framework of analysis.

    I am working on this, however, getting ready to transcribe the many hours of audio recordings that Taras Ivchenko and I recently made. So for now I'll just make a few informal observations.

    Correct, the final /n/ is not written when erhua is present. I have not done any phonetic analysis of this phenomenon yet.

    Neutral tones do occur, and I expect that to be the case for personal names too. An investigation of the phonology and phonetics of personal names is already on my list, it is a fascinating topic.

    1. The realization of labialization as something which sounds like a bilabial trill is present in the speech of many speakers.

    2. Aspiration is certainly strong, but my impression is that it is not always velar. I seem to be hearing [pç] in front of [i], for example.

    3. Yes, they are all strongly fricated.

    There are several general dictionaries:

    Вурус—Җун-ян хуадян := Русский-дунганский словарь. Сост. Ю. Яншансин, Л. Шинло. Фрунзе 1959.

    Краткий дунганско-русский словарь := Җеёди хуэйзў-вурус хуадян. Сост. Ю. Яншансин Фрунзе: Академия Наук Киргизской ССР 1968.

    Русско-дунганский словарь. Сост. М. Х. Имазов. Фрунзе 1981.

    Җеёди хуэйзў-вурус хуадян := Краткий дунганско-русский словарь. Сост. Ю. Яншансин, 2. изд., ред. Б. Дуваза, Ф. Машинхаева, А. Фокин, Ф. Хаваза. АН КР, Институт истории и культурного наследия. Центр дунгановедения и китаистики. Москва: ИПБ [Институт перевода Библии] 2009.

    Русско-дунганский словарь := Вурус-хуэйзў хуадян. Сост. Ю. Яншансин, 2. изд. Москва, ИПБ [Институт перевода Библии], 200x.

    @Jonathan Gress-Wright
    I think that the "faithfulness" of the borrowings is not primarily due to the writing system, but to the fact that knowledge of Russian among Dungans has grown dramatically over the years, to a point where practically all Dungans have native or native-like Russian (I am less sure of the situation in Osh and Uzbekistan, though, where Uzbek seems to be more important). Having a similar alphabet certainly helps, but this fact cannot alone account for the pronunciation.

  24. Victor Mair said,

    April 28, 2013 @ 5:48 am

    @Jonathan Gress-Wright

    I'm very glad that Ivo Spira has weighed in, since he is a true specialist on Dungan language. In fact, his lateness in joining the commenters to this post was due to his having been in the field, recording samples of Dungan speech among communities who live near Issyk Kul.

    For the moment, I would add only that loanwords are not entirely "constrained by the phonology of the borrowing language", but that sometimes borrowed names and terms enlarge the phonemic inventory, e.g., Mozart and tsetse fly. Naturally, this mainly occurs in the oral realm, but the availability of a writing system that permits the recording of such changes (e.g., consonant clusters, as with the Dungans using Russian "traktor" instead of Mandarin "tuōlājī 拖拉机") does make a huge difference in sanctioning such borrowings in the standard form of a given language.

  25. Jonathan Gress-Wright said,

    April 28, 2013 @ 9:13 am

    @Ivo and @Victor:

    Thank you for your replies! I found this paper by Vendelin and Peperkamp (2006) that suggests orthography can influence the adaptation of loanwords. They were focusing on adaptation of English non-words by late French-English bilinguals and found that orthography does affect degree of adaptation. This makes in terms of what Ivo said about Dungan speakers' competency in Russian, showing that competence in the source language influences adaptation of words from that language. This also fits with what Victor said about the effect of Chinese orthography. We would predict that illiterate Mandarin speakers might adapt Russian loanwords differently.

  26. Jonathan Gress-Wright said,

    April 28, 2013 @ 9:13 am

    Sorry here is the paper I cited:


  27. Ivo Spira said,

    May 14, 2013 @ 9:46 am

    Now that I've had a chance of consulting some native speakers, here is a provisional explanation of the sign mentioned by @Vladimir Menkov :



    Normalized version:

    К[рестьянское]/Х.[озяайство] Губеза
    Сый кан фу, ги та мәю кухуан.


    K[rest'janskoe]/Kh.[ozjajstvo] Gubeza
    Syj kan fu, gi ta meju kukhuan.

    /sɨj²⁴ kʰan⁵¹ fɯ⁴⁴, ki⁵¹ tʰa⁵¹ mə²⁴ju⁵¹ kʰu⁵¹xuan/

    The last line can be rendered in Chinese characters as follows:
    Theoretical reading in Standard Mandarin:
    Shéi kǎn shù, gěi tā méi yǒu kǒuhuan.


    "Gubeza [a Surname][Russian:] Farm. [Dungan:] Whoever cuts down trees [here], for him/her there is no forgiveness."

    "кухуан" is a key concept with religious overtones. Its basic meaning is "permission", but it also shades over into "blessing" and has the additional senses of "forgiveness" and "will" (as in "God's will"). I am somewhat uncertain about the character version 口唤. The message is quite strong: if you cut down trees here, you will never be forgiven, perhaps not even in the life hereafter.

  28. Ivo Spira said,

    May 14, 2013 @ 9:48 pm

    Seems I was a little bit too fast, the sign reads КУХАН. But my informants still said "кухуан".

  29. Mark Mandel said,

    July 16, 2014 @ 9:49 pm

    @Ivo: You are, ahh, forgiven. The Х, cut off by the hole in the sign, looks very much like an У. That, together with your informants' pronunciation, could easily have left you thinking you'd seen КУХУАН. (The "А"s permit me to view the text as all-caps.)

RSS feed for comments on this post