« previous post | next post »

In China (and around the world among China watchers), everybody's talking about this ungainly syllable.  "Duang" surfaced less than a week ago, but already it has been used millions and millions of times.

"The Word That Broke the Chinese Internet" (2/27/15) by Bethany Allen-Ebrahimian

"'Duang' is Everywhere on the Chinese Internets, Here’s What It Means" (2/27/15) by Charles Liu

"Chinese netizens just invented a new word, and it's going insanely viral" (2/28/15) by Ryan Kilpatrick (English text part of the way down the page)

We know exactly how the word arose and developed, every step of the way.  It was first used by the famous actor, Jackie Chan, in a shampoo commercial that was made in 2004.

From the Liu article, quoting Jackie in the ad:

I don’t want to say it, but when commercials are shot, they can add special effects afterwards, and the hair will be like “DUANG”! So rich! So shiny! So soft! (But if) I were to do that, the audience would scold me for not actually having this hair. …And after making the commercial, the hair is still mine. You don’t have to add any special effects.

Ryan Kilpatrick says that "duang" can be roughly rendered in English as "boing".  Fair enough — that's a pretty clever translation, but we'll dig deeper into the semantics of duang later on.  For the moment, it's enough to note that duang is an onomatopoeic expression for describing the fluffiness and bounciness of someone's hair.

Before analyzing the phonology and morphology of duang, let's watch a video of Jackie delivering the commercial.  Naturally, this version has been doctored to emphasize the effect of Jackie telling us how "duang" his hair became after using Bawang Anti-hair Fall Shampoo.

The doctored version is hilarious, and there are also cute gifs of the magic moment when Jackie says "duang" floating around the internet (some are embedded in the articles cited above in this post).

You can watch the original 2004 commercial here (in Kilpatrick's article), about halfway down the page.  Note that there are two videos:  1. the heavily manipulated one that has been sweeping the internet, 2. a. the original commercial, b. a slightly modified version where Jackie's words in praise of the naturalness and purity of Bawang shampoo are cleverly subverted by voice-over to say exactly the opposite:  the shampoo is full of artificial ingredients (not Chinese herbs) and the ad-makers have used special effects to emphasize the bounciness of Jackie's hair, etc..

So here's the complete derivation of "duang".  At first it was just a spontaneous sound in Jackie Chan's mouth.  Then somebody wrote it down in romanization for the subtitles of the doctored version of the video (the subtitles for the original commercial and the slightly modified version just skipped over "duang", omitting it altogether).  At this early stage in the evolution of "duang", it didn't have a tone attached to it, just "duang".  Later, the pedants decided that it should have a first tone, so now people who are fussy and take the time to do it add the first tone diacritical:  duāng.  But that's still not good enough.  Since Jackie Chan is Chinese and he spoke "duang" as part of a Chinese sentence, surely there has to be a Chinese character for this powerful meme, and this is it (you have to imagine that the two characters stacked on top of each other are actually a single character written within the square occupied by one of the characters):



The problem is that the sinographic form of "duang" doesn't yet exist in any electronic font, much less any dictionary, whether in printed or online form.  It might as well have been invented by Xu Bing as part of his "A Book from the Sky".  To the extent that it is circulating around the web, it is doing so as an image (i.e., it is not yet typable and transmittable via Unicode).  The vast majority of people are still just writing "duang", and I expect that will be the case for quite a while yet.  Thus, what is probably the most popular Chinese word on the planet right now — in its written form — exists almost entirely in romanization.

Where did they dig up this new character — chéng 成 + lóng 龍/龙 (lit., "become dragon")?  Simple — that's Jackie Chan's name in Chinese:

陳成龍 (traditional) / 陈成龙 (simplified)

Cantonese Can4 Sing4 Lung4 / Mandarin Chén Chénglóng

Phonologically, there's no connection between the two characters of Jackie Chan's name and "duang".  The sound "duang" was just arbitrarily assigned to the new character.

The most that can be said phonologically in defense of the new character is that "duang" and Lung4/-lóng both end in voiced velar nasals.

"Duang" is not even a legitimate syllable in Modern Standard Mandarin, although d- is a possible initial and -uang is a possible final.  Nor do I know of any Sinitic language, topolect, or dialect that has the syllable "duang"

This reminds me of one of David Moser's classic papers, "Some Things Chinese Characters Can’t Do-Be-Do-Be-Do" (2001), where he gives examples of "scat singing" in Chinese that couldn't be represented without pinyin.

He shows how Pinyin allows you to break through the one-syllable chunk that the characters tie you to, and allows you to create syllables that combine the consonants and vowel segments of Mandarin into new chunks like the following:

g as in “gan” + ing as in “ding” = ging
shu as in “shuang” + ing as in “ding” = shuing
p as in “ping” + uan as in “guan” = puan
d as in “dong” + ua as in “gua” = dua
ku as in “kuan” + en as in “men” = kuen

"Duang" would be another example of this sort.  You can't convey this level of morphology with the characters, which code for a static, unbreakable syllable.

As for semantics, one could argue that, when it was first uttered, "duang" really had no meaning at all, but was just a funny sound effect.  Already, however, in less than a week, "duang" has been given various definitions, such as jiā tèxiào 加特效 ("add special effects").

Where do Chinese words come from?  Sometimes they come from nowhere — just out of thin air.  Where do Chinese characters come from?  Sometimes people just make them up out of other characters, or out of their own minds.

"'Book from the Ground'" (12/5/12)

"The unpredictability of Chinese character formation and pronunciation" (2/6/12)

"Weird characters" (7/7/13)

"New radicals in an old writing system" (8/29/12)

In recent years, Chinese artists have become fond of inventing not just new radicals and new characters, but whole new writing systems that incorporate Roman letters and English words, as Petya Andreeva has shown in a paper written for my "Language, Script, and Society in China" course, which I hope to publish soon in Sino-Platonic Papers.

[Thanks to David Moser, Maddie Wilcox, South Coblin, Axel Schuessler, Nathan Hopson, Wendy Fuglestad, and BJS]


  1. Lugubert said,

    March 1, 2015 @ 2:00 pm

    For "inventing … characters", I think Chao Yuen Ren's rendering of Jabberwocky deserves mentioning.

  2. Michael Watts said,

    March 1, 2015 @ 2:05 pm

    This reminds me of one of David Moser's classic papers, "Some Things Chinese Characters Can’t Do-Be-Do-Be-Do" (2001), where he gives examples of "scat singing" in Chinese that couldn't be represented without pinyin.

    He shows how Pinyin allows you to break through the one-syllable chunk that the characters tie you to, and allows you to create syllables that combine the consonants and vowel segments of Mandarin into new chunks like the following:
    g as in “gan” + ing as in “ding” = ging
    shu as in “shuang” + ing as in “ding” = shuing
    p as in “ping” + uan as in “guan” = puan
    d as in “dong” + ua as in “gua” = dua
    ku as in “kuan” + en as in “men” = kuen

    "Duang" would be another example of this sort. You can't convey this level of morphology with the characters, which code for a static, unbreakable syllable.

    Two questions:

    1. Why couldn't you use fanqie to represent these syllables? As I understand it, that works for all of them except "shuing". Why can't I represent duang in the traditional way, say as 动双切?

    2. On a related note, I've read that the traditional view is that the beginning of shuang is sh-, not shu-, and therefore the rime is wang, not ang (a rime of -ang would get us shang, a very conventional MSM syllable). According to that same view, the syllable 困 kun is an onset of k- plus a rime of -wen, a perfect rhyme, discounting the w, with 门 men. How does that differ from the proposed syllable kuen?

  3. Meng Xiangzhong said,

    March 1, 2015 @ 5:29 pm

    Forgive the childishness, but I can't help but wondering (a) if "shuing" is supposed to sound like "schwing" and (b) if so why haven't we come up with a good character for THAT sound ;)

  4. Rubrick said,

    March 1, 2015 @ 6:40 pm

    Duang or no duang, Chan's hair is totally fleek.

  5. Lew Perin said,

    March 1, 2015 @ 10:26 pm

    Ignoring characters for a moment, why isn’t there a duang *syllable* in Mandarin? I’m no phonologist, but a Mandarin speaker’s mouth wouldn’t need to work very hard to reach duang’s final from its initial, would it? Mandarin is famously phoneme-poor, so adding the duang syllable could be a measurable improvement in the information theoretic efficiency of Mandarin communication.

    Here duang can stand in for many pinyin-representable but unused syllables. If lots of those unused syllables shouldered part of the Mandarin semantic load, speaking and hearing Mandarin could become significantly easier.

    I’ve read Moser’s fine article on what Chinese characters can’t do, but I don’t think writing is to blame — are Chinese illiterates really using a wider range of syllables than gaokao victors?

  6. Eli Nelson said,

    March 1, 2015 @ 11:12 pm

    @Lew Perin: My understanding is that the "unused" syllables represent combinations of sounds that didn't happen to arise in Chinese words through regular sound change from previous forms. You could compare it to English words starting with /z/: in Old English, /z/ was an allophone of /s/ intervocallically within a word, so both were written with the letter "s". Word-intial /z/ simply never developed in Old English, so there was no convention for spelling it. Words that start with /z/ now are the result of contractions, imported words or invented words like onomatopoeia. People have no difficulty pronouncing them, but they're rare.

  7. Michael Watts said,

    March 1, 2015 @ 11:23 pm

    Was word-initial /s/ prevented from being realized as /z/ when it occurred intervocalically in old english? That can't have been all that rare.

  8. Wentao said,

    March 2, 2015 @ 12:35 am

    On a related note to Eli Nelson's comment above – I have always been unsure regarding where to draw the line between systematic and accidental phonological gaps. I once had a long argument with a friend about what kind of gap is the absence of bén in Standard Mandarin. It is assumed to be accidental because we have all three other tones (奔, 本, 笨), but I argued that it is in fact not "accidental" because they are made impossible by systematic, although not obvious, historical constraints.

    Similarly, should the lack of duang be accounted to a systematic gap, i.e. the rime -uang can only come after retroflex affricates? I would appreciate it greatly if linguistics experts on Language Log can enlighten me on this matter!

  9. Michael Watts said,

    March 2, 2015 @ 1:00 am

    wentao –

    I have absolutely no knowledge of historical chinese phonology, but I'm interested in whatever you have to say about it. I don't think, though, that modern chinese people feel particularly constrained by the historical development of their language; most of them don't know about it. Contractions are an active phenomenon in MSM; we see xuyao -> xiao; buyong -> beng (I don't really get that one); and zheyang -> jiang. Do you think the "target space" for contractions is restricted by the set of syllables that currently exist? I tend to suspect that it's driven more by raw phonological considerations of the language-as-it-is-now. Otherwise, how could Jackie Chan have come up with duang at all?

    I've been told that pejorative words like 笨 often started in some other tone and migrated to the fourth tone because fourth tone is more pejorative or more agressive or whatever. Assuming that's true, would it mean that when a syllable doesn't exist in fourth tone, that's always accidental, even though the absence of another tone might be systematic?

  10. flow said,

    March 2, 2015 @ 7:01 am

    @Wentao very interesting. I believe the key point here is to distinguish between a diachronic (historical) and a synchronic (ahistorical) point of view. Now, while the diachronic view is very important, it will hardly matter in the minds of native speakers growing up. On that count, i think we can safely say that *ben2 is an accidental gap since it cannot be linked to a pattern of gaps within the phonology of *present-day* Mandarin.

    @Michael Watts I think a further important difference has to be made, in phonology, between 'things not permitted by the phonology of their mother tongue' and 'things people find hard or impossible to utter as a speech sound in fluent speech'. This difference may perhaps be best illustrated by clicks; there's perhaps good reason that almost no language in the world has clicks as regular speech sounds, yet it's not impossible, as the few languages with clicks attest. Now a native English speaker will typically find at least some clicks easy to utter and will use them without giving much thought, BUT not inside a regular word: try to say "ts-ts-ts" [|||] (sound of disapproval) and then "Ba|avia" (instead of "Batavia")—it's hard to get right!

    Now we should think that when speakers find things genuinely hard to utter, those sounds, sound combinations or some sounds in some positions will be systematically missing from a language. As for Mandarin, i guess this is the case for syllable-final laterals as i have observed how easy Chinese speakers find to say "lo" and how hard it is for them to say "ol" (with a clear [l]).

    But, remarkably, there are also systematic gaps in languages that people do *not* find hard to say, and i guess as for Chinese, "king", "ging" are such cases—i.e. now they may be, but in past centuries, *maybe* Mandarin speakers found velar stops preceding a palatal vowel hard to pronounce (and maybe it was just a fad—i heard the Castilian [θ] started as a fad).

    As for David Moser's remark on how you can re-combine the letters of Pinyin to arrive at plausible new syllables that speakers can say but characters can not render—ging, shuing, puan, dua, kuen—well, i think most of his examples are systematic gaps in Mandarin!

    1) There's no 'gi-' as velars and non-palatal sibilants are in complementary distribution with palatal consonants before palatal vowels;

    2) 'shuing' as a syllable is 'too crowded' as it were; we do have lun, lin, ling, long (= lung), luan, guai, jiao, jian, jiang etc but no *win, *wing; there's guan, guang, gao (= gau) but no *gaun, *gaung. The pertinent points are that a) the coda can not be filled with more than one segment, and b) you can only have "one i and one u" within one syllable (and where they coincide they come typically out as 'ü' [y]).

    3) The bilabial stops and nasals are notoriously ambiguous when it comes to whether or not they may be followed by a bilabial glide. While you can clearly hear that glide in syllables like bo, mo, po [bwo, mwo, pwo] and they sound very much like e.g. guo, nuo, kuo, there still is no opposition to, say, ge, ne, ke [gɤ, nɤ, kɤ] (and yet Taiwanese school teachers insist on [bɤpɤmɤfɤ] instead of the more natural 'bopomofo'—but then *my* teacher uttered strange sounds too when teaching us writing in school). Because of this 'bilabial ambiguity', there are no *bua, *muan, *puai, only ba, man, pai etc in Mandarin. Note how 不用 buyong (= bu-iung) became 甭 (beng), which is identical to / not separable from b+weng ("bung"), and how many speakers realize Pinyin "feng" as [fung]. Still, maybe *puan is a gap waiting to be used.

    4) 'dua' is the one 'acceptable' of Moser's examples in terms of accidental vs systematic gaps; there's duan, duo, tuan, tuo, gua, guan, guo. On the other hand, it's remarkable that there's no *tua, either, and no *duang, *tuang, although there's guang; also, *lua, *nua are missing but there are luan, nuan (and *luang, *nuang are missing again). Maybe all that's left to say is that it is as it is, maybe there's a rule hidden, maybe these gaps are just waiting to be used, like spare parts.

    5) Lastly, 'kuen' is not at all impossible. In fact, it already exists, and you probably know it even if you don't speak Chinese: this sound appears in the name of the famous Kunlun mountain range. It so happens that you can analyze Mandarin using only *two* vowel phonemes for the nuclear position, at least for complex syllables that incorporate a separate glide (the 'simple' syllables a, e, yi (= i), yu (= ü), wu (= u) are a different story, but note how native speakers chose to write three of them with an initial letter that seems to represent a glide when Pinyin was designed).

    In the two-vowel view, the nuclear vowel of le, luo, lou, lei, lie, lüe is always the same (write it 'ɵ': lɵ, luɵ, lɵu, lɵi, liɵ, lüɵ) and contrasts with la, *lua, lao (= lau), lai, lia, *lüa (as for the gaps shown here, they're filled out when using other initials). The nuclear vowel in kun, lun, gun contrasts with a as in kan, luan, guan, in other words, these syllables *are* really / phonemically / ambiguously kuɵn, luɵn, guɵn; in this analysis, therefore, there's simply no free space to squeeze in a kuen as distinct from kun as both are the same. Listening closely you will in fact be able to discern quite clearly a sort of e-ish offglide in the pronunciation of many speakers; "kun" in Mandarin is not so much k+u+n (as the Pinyin spelling would suggest) but, rather, ku-uɵ-ɵn or (k[u){ɵ]n} if you will with overlapping constituents and fuzzy borders. Coincidentally, both Zhuyinfuhao and Fanqie have similar analyses, viz. ㄎㄨㄣ (i.e. k, u, en) and 苦悶 (i.e. k(u), (m)en—one of many possible spellings).

    As for "duang" as used by Jacky Chan, there's a certain similarity with interjections like "aiyo" and so on which are also known to often not obey the stricter rules of the 'main' phonological pattern of languages. Feelingwise, it's certainly much more appealing and less of an outsider when compared to other gaps. As such, i can fully imagine this syllable to become part of the language (mainstream it already is). +1 for adding ⿱成龍 and ⿱成龙 to Unicode v9!!!

  11. ahkow said,

    March 2, 2015 @ 8:17 am

    Professor Mair gives Jackie Chan's name in Chinese as 陳成龍, which is actually incorrect. 成龍 is a stage name – not sure if the surname is ever used in that context. Chan's actual name is 陳港生, and as he tells it the family originally had the surname 房 (fang35 in std. Mandarin, fong11 in Cantonese) – and notably his son Jaycee now goes by 房祖名.

  12. Wentao said,

    March 2, 2015 @ 10:43 am

    Thank you for your detailed and informative response! I think your distinction of diachronic and synchronic approaches is very useful in explaining the gaps. However, I have an argument against your observation that the absence of bén "cannot be linked to a pattern of gaps within the phonology of *present-day* Mandarin". I believe that there is a rule that goes something like this: "The second tone is lacking in syllables that start with a voiceless, unaspirated consonant and end with a nasal coda." The only exceptions are dialectal expressions such as 甭 béng and 咱 zán, which I (personally) don't like to consider as part of MSM. Admittedly that's a rather complicated rule, but it's a rule after all!

    Also, in past centuries Mandarin speakers actually found ging and king *easier* to pronounce then modern-day speakers, because they used to be part of the language. Later they became palatalized to jing and qing, but we can still find traces of them in for example the transliterations Peking and Chungking.

    @Michael Watts
    Your question about whether the target space is restricted is very interesting. My instinctive answer is "yes", but there are instances such as 甭 béng that are the only lexemes of that syllable. I have to think of more examples before reaching an answer.

    I'm skeptical of the theory that the fourth tone being more "aggressive". Can you give more examples? Although I'm not entirely sure about the etymology of 笨, but a search gives me the result that in Middle Chinese it was also 去声 (Mandarin 4th tone), and the definition 大肥 (big and fat) looks like right:

  13. flow said,

    March 2, 2015 @ 12:21 pm

    @Wentao Indeed there's a rule for this! I was perhaps to quick in stating there was none, but how could i ever know—i can't! Just let me say that while a rule is a rule is a rule, another question is whether it is felt / known / observed by native speakers. It does look somewhat arbitrary when looking at it in a purely synchronic fashion, but of course that doesn't have to keep people from finding zan2, beng2 or *ben2 "un-Mandarin" or even slightly yucky, even if they couldn't nail it down.

    As for the ease of saying "king", "ging", i guess that ease or ability to produce a given sound combination is but one factor in sound change. Others have to do with reputation, adaption to a standard, things like that. And some things do come back, like the Old Japanese /pa/ which became /ha/, which in turn opened some space to allow a newly formed /pa/ into the language. IMHO it is conceivable that "king" comes back to Mandarin proper, and maybe words like "King Kong", "Burger King" and so on will be triggers for this.

  14. Rodger C said,

    March 2, 2015 @ 12:39 pm

    Was word-initial /s/ prevented from being realized as /z/ when it occurred intervocalically in old english?

    Sure, as far as we can tell. Old English had the same juncture feature as Modern English or German.

  15. Jeffrey Willson said,

    March 2, 2015 @ 2:30 pm

    Flow's comment is a thorough treatment of the phonological issues. It makes perfect sense to accept rules excluding certain syllables from MSM if they cannot arise by regular rules from Middle Chinese (MC); after all, many 方言 fangyan are phonologically much closer to MC than MSM is, and many Chinese people speak MSM as a style adapted from their own fangyan rather than regarding it as a completely independent language. Also, MSM is highly dependent on a formal standard, and reference books preferentially sanction forms that are regular reflexes of MC words. There are words in the language that do not fit in the MC system; some, like 甭 béng, are contractions. I'd have to guess how the tone of 咱 zán arose, but might be the result of dialect mixing: a dialect word that was 陰平 or 上聲 in some fangyan (with rising contour) being borrowed into Mandarin without adapting its tone contour. In any case, I do not share Wentao's reservations about considering them part of MSM. Another interesting example is 媽 mā, which according to the rules of MC -> MSM reflexes can only result from an entering-tone 入聲 syllable, but almost certainly is not entering-tone.

    The strange thing about duāng is that it appears to be possible in the MSM system, and also the reflex of a possible syllable in Middle Chinese (twang in Baxter's system; 端光切 or possibly 端庚合口, not 动双切, which is daewng in Baxter's system), and yet it hasn't become a word earlier. It could conceivably arise as a contraction for something like 得旺 dé wàng. I guess the absence of this word says something about the resistance of the standard language to new morphemes, as well as a historical lack of interest in cataloguing the vocabulary of living speech into reference works.

  16. Victor Mair said,

    March 2, 2015 @ 5:03 pm

    "Millions share new Chinese character" (3/2/15)


  17. gaoxiaen said,

    March 2, 2015 @ 6:12 pm

    When I asked students, they all said that there was no word pronounced as biang2. Maybe that's why this word was invented.


  18. Jesse Tseng said,

    March 2, 2015 @ 8:04 pm

    The word that Jackie Chan says in the commercial doesn't sound like duāng ㄉㄨㄤ [twɑŋ] to me, but [twɑ̃w]. I don't know if there is a Portuguese word "doão", but what he says sounds to me like that. It does not rhyme with any Mandarin pronunciation of 双 庄 窗 etc. that I am familiar with, and that is not how Jackie Chan pronounces other -ang words like 当 and 广 elsewhere in the commercial.

    But I guess the word has now taken on a life of its own, and so it needs to be retrofitted with standard Mandarin phonemes, and agreed-upon pinyin and hanzi forms.

  19. Victor Mair said,

    March 2, 2015 @ 8:35 pm

    @Jesse Tseng

    I commend you! Millions of people are out there saying and writing "duang", but you record what Jackie Chan really said.

  20. Victor Mair said,

    March 2, 2015 @ 10:09 pm

    Some relevant Wikipedia entries:




  21. Chuck Winter said,

    March 3, 2015 @ 2:18 am

    Mr Chan is a great actor/comedian, who has been known (in his films) to have a go at karaoke and singing. I always considered his "duang, duang" as a rendition of the 1958 The Chiffons song:"He's so fine (do-lang, do-lang), I wish he was mine (do-lang, do-lang)…". Fits the bill perfectly: looking in the mirror, grooming, self-admiring…

  22. Chuck Winter said,

    March 3, 2015 @ 2:25 am

    The Chiffons song @youtube

  23. Uncle Hanzi said,

    March 3, 2015 @ 8:06 am

    I figured I would find out all about it here.

  24. Rachel said,

    March 3, 2015 @ 5:49 pm

    Michael Watts said: "Do you think the "target space" for contractions is restricted by the set of syllables that currently exist?"

    I have possible exception to this. In Taiwan, when I was there (as recently as 2004), it was popular in Mandarin to call something "biang". This seems to have been a borrowing from Taiwanese, where biang is a perfectly legitimate word. (Forgive my use of Pinyin for Taiwanese; I don't know romanization for Taiwanese, nor am I fluent enough in IPA.)

    Also interesting is that folks invented a portmanteau character for this, composed of the elements 不一样。 Someone has uploaded a hacked together version of this to Wikipedia: http://commons.wikimedia.org/wiki/File:Biang.svg

    Not sure how this relates to the famously stroke-laden character for biangbiangmian, either. http://news.takungpao.com/taiwan/liangan/2014-02/2288922.html

  25. David Moser said,

    March 4, 2015 @ 1:36 am

    Hi Flow:

    Thanks for this, I'm sure you're right that most of these are systematic gaps in Chinese. Interesting that the coda can't be filled with more than one segment — I never noticed that.

    I just have one question about "kuen", though. Leaving aside the question of whether Chinese would naturally make the distinction, the difference between "gun" and "gen" are quite clear, and I, at least have little problem pronouncing the distinction between the naturally occuring "kun" and my hypothetical "kuen." So I would deny that the syllable already exists in Kunlun", the distinction is out there to be made.

    A related question, though. Why the difference in the vowel sound of "wen" and "ren", and all the other "en" pinyin examples like "gen", "weng", "en", "men", etc. This has always seemed like a glitch in pinyin to me. What's going on?

  26. flow said,

    March 4, 2015 @ 11:36 am

    @David Moser See, in my dialect of MSC i find it hard and somewhat unnatural to make a difference between "kun" and "kuen". I have no such qualms with "biang" or "duang". Then again, i'm not a native speaker, and native speakers may not all feel the same.

    As for your second point, the thing is that Chinese pronunciation has been somewhat shoe-horned into what expressive means the Latin alphabet has to offer as long as you don't depart from sequences of the 26 basic letters (plus "ü", which the orthography tries to avoid very much, producing such—entirely readable, but still unnecessarily irregular—forms as "yu" vs "nü", where the nucleus is undeniably the same).

    As is the case with (mosquitos vs elephants) vs (one seagull vs another seagull, both belonging to the various arctic ringspecies)—where we are very clear that we do have two different species in the first case but our mental tool suffers a breakdown in the second, so it is the case with the sounds of Chinese (and certainly other languages as well) when trying to neatly segmentize them. Segmentation is quite easily done in cases such as "ke", "ling", "san". It is much more difficult to do in an unambiguous fashion for the syllable codas of, say, "yong", "wen", "yun", "que". Part of the difficulty comes from the fact that, apart from language games, er-suffixing, and maybe rhyming literature, there's very little going on *inside* a Chinese syllable in terms of phonological processes. In a language like German, you can e.g. "inspect" the final sound of [ra:t] (which might be "Rad" (wheel) or "Rat" (counselor)) by submitting it to a morphophonemic process as occurs for example when forming the plurals, for what you then get is [re:—dɐ] vs [re:—tə]. In these forms, the final consonants have swapped positions from the syllable coda to the syllable onset, and the neutralizing effect of obstruent devoicing has been thus cancelled.

    In Chinese, there are no such morphophonemic processes, and i know not of any test that could demonstrate the "real", "hidden" nature (that elusive beast phonologists have been chasing for a hundred years, only to assert its multi-faceted nature) of the syllabic nucleus of "gen" vs that of "wen". However, what we can do is go and try to arrange known smallest utterances—for example, the 400-odd syllables of regular standard Mandarin (discounting tones and putting interjections aside)—in a regular system. What we then find is that whatever may constitute a phonetic difference between the codas of "gen" and "wen", it is not used to make a systematic difference. In other words, while we certainly *could* write "gɜn" for the first and "wɷn" for the second, we never find a 'regular' speaker who uses and systematically keeps apart "gɜn", "wɷn", "wɜn", although "gɜn" vs "gɷn" *is* a regular opposition.

    The thing is, our letters trick us into believing something strange is going on, or that there *should* be "wɷn" != "wɜn" because there *is* "gɜn" != "gɷn", but the fact is that the space that allows us to think so was opened by our too-permissive analysis (which relies on sequences of phonetically unambiguous symbols), *not* by the observable facts of the language. In a Chinese syllable, there is no space in the phonological floorplan to accommodate for the surmised opposition. Put another way, because MSC phonology only provides for a two-way distinction for the syllabic nuclei of complex syllables, there can't be "wɷn" != "wɜn" because there's already "wen" vs. "wan". You cannot have "wan" != "wen" != "wön" != "wün" and so on in this system, there's only a nucleus with "a" and the other one. How that "other one" comes out may greatly vary depending on the phonotactic vicinity found in the syllable: "guo", "gei", "que" all have the "same" nucleus in this view, the "not 'a'" nucleus. It is as though this "not 'a'" nucleus was very much mixed with the medial and finals and "smeared" across the syllable final, a bit like an electron in the quantum dynamic view of the atom.

    Indeed, we do find that (1) the "i", "u" and "ü" components in MSC syllables do interact *across a distance*, because there can't be *wau, *wou, *iei (only wai, yao, wei, you), and the singular and special yai (as in 天涯 tianyai) has been ousted from Putonghua (during the 20th c, it would seem), although it persists in Taiwan, and we also find that (2) the distinction that Pinyin makes between e.g. "an" and "ang" (i.e. /an/, /aŋ/) comes out as [aⁿ], [ɑᵑ] (a front ("light") "a", gravitating towards palatal nasalization, vs a back ("dark") "a", gravitating towards velar nasalization—sorry for the rather imprecise terms here). The second fact is much less surprising than the first, but both show that the Firthian school of thought may be in a better position to deliver an acceptable model for the MSC syllable codas than the segments-only school of thought.

    I hope these observations can help to make it plausible that there cannot be "kun" != "kuen" in MSC.

    BTW your "weng" does not strictly belong into the "en" group of syllable finals, unless what you meant was "eN", "N" standing in for "nasal (coda)" (of which there are two in Mandarin).

  27. Michael Watts said,

    March 4, 2015 @ 3:46 pm

    You can't blame wen, ren, gen, men, en, etc all being written with the same vowel on romanization; zhuyin represents them all as ending the same way (with ㄣ). It also represents kun, gun, chun, jun, xun, etc the same way. They're conceived of as all using the same vowel. If I knew chinese songs I'd look for rhymes between -wen syllables and -en syllables.

  28. flow said,

    March 4, 2015 @ 5:59 pm

    @Michael Watts Exactly. What i'm saying is that without resorting to "unusual" Latin letters (the authors of Pinyin did consider to use e.g. Ч, Ц at some point but later reverted that) Latin offers but a basic stock of vowels so you quickly run out of options when designing an orthography for a language that is rich in vowels. A sound-based spelling / orthography always gives away hints what the authors' mental model of the phonological system might have looked like; in this case, however, the evidence is relatively poor because building a Latin orthography means dealing with too few letters and an established tradition among foreigners. Interestingly, ㄛ "o" and ㄜ "e" (as in "ge", not as in "gei") were not differentiated in early drafts of the Zhuyinfuhao.

    And yes, finding 'good' "gun"/"wen"/"men"/"en" rhymes (maybe by soliciting native judgements) would corroborate the view that "gun" is, underlyingly, already "guen".

  29. Michael Watts said,

    March 4, 2015 @ 7:03 pm

    flow –

    under the model you've described where chinese syllables have one of two nuclear vowels, how is the contrast between sha / she / shi explained? I know shi is traditionally viewed as having a zero rime, but I don't really believe in the concept of a zero rime, and I notice that in a song the vowel of shi may be held.

  30. Victor Mair said,

    March 4, 2015 @ 8:09 pm

    "China goes ga-ga over new Chinese character 'duang'"


  31. Zev Handel said,

    March 4, 2015 @ 8:47 pm

    In discussing the issues that @flow, @David Moser, @Michael Watts and others have raised here (phonotactic constraints on Mandarin syllables; historical sources of systematic gaps; phonemic analysis, etc.) it is helpful to recall that the pinyin romanization system really consists of two parts: a set of standardized representations for all the initials and finals of Mandarin (which we might call "base pinyin"), and a set of spelling rules that modify the base pinyin forms when initials and finals are combined. These spelling rules affect only written representation, they do not indicate a change in pronunciation.

    The set of spelling rules is quite complicated, the degree of complexity often comes as quite a surprise to those familiar with pinyin usage.

    Let me give a few examples to illustrated the difference between "base pinyin" from "spelled pinyin".

    1) There is a Mandarin final with base pinyin form "-uen". (I will mark base pinyin forms with "-" to distinguish them from spelled forms.) It consists of a main vowel represented be "e", a preceding on-glide represented by "u", and a nasal coda represented by "n".

    There are two spelling rules related to this base form:
    1a) When no initial consonant is present, change "u" to "w" and spell as "wen" (e.g. wén 文).
    1b) When an initial consonant is present, remove the "e" and spell as "un" (e.g. kùn 困).
    The spelled finals "wen" and "un" look quite different, but they are just different representations of underlying "uen".
    It is interesting to note that because of the spelling rules, the base form "-uen" never actually appears in spelled pinyin.

    2) There is a Mandarin final with base pinyin form "-ün". It consists of a main vowel represented be "ü", and a nasal coda represented by "n".

    There are two spelling rules related to this form.
    2a) When no initial consonant is present, change "ü" to "yu" and spell as "yun" (e.g. yún 雲)
    2b) When the initial consonant is "j-", "q-", or "x-", then "-ün" is spelled "un" without the umlaut (e.g. xùn 訓)

    Note that the different base forms (1) "-uen" and (2) "-ün", representing two very different finals, can end up spelled exactly the same: "un". But because they always appear in different contexts, we can always recover the base form from the spelled form, and the intended pronunciation is unambiguous.

    One result of these spelling rues is that pinyin words that look like rhyming words sometimes aren't, and ones that look like they don't rhyme sometimes do. Take a look again at the four example characters from above:

    1a) 文 wén (base: 0 + -uen), 3 sounds
    1b) 困 kùn (base: k + -uen), 4 sounds
    1c) 雲 yún (base: 0 + -ün), 2 sounds
    1d) 訓 xùn (base: x + -ün), 3 sounds

    Based on surface spelling, it looks like the last three should rhyme together. But they don't. If you want to know what's going on with the actual sound components of Mandarin syllables, you have to first undo the pinyin spelling rules to get back to the base forms.

  32. Zev Handel said,

    March 4, 2015 @ 8:51 pm

    @Victor, the assignment of first tone to duang is part of a very interesting general pattern for Mandarin onomatopoeia: they are all first tone. (Perhaps there are some exceptions, but I'm not aware of any. If anyone knows of any, do tell!) Think of the Mandarin equivalents of "ding-dong", "clink-clank", "swish-swoosh", "glug-glug", "splish-splash", "boom-boom", etc. All first tone (going back to MC píng 平 tone).

    (“Duang" is also clearly a first-tone pronunciation by Jackie Chan in the original video. This might be because that pattern is already built into the language. But it's interesting to speculate on what would have happened if Chan had pronounced it with a distinct falling tone. Might it have acquired first-tone anyway, because of its status as onomatopoeia?)

  33. Zev Handel said,

    March 4, 2015 @ 8:56 pm

    @Jeffrey Willson, the lack of syllables like /twaŋ/, /thwaŋ/, /dwaŋ/ in Middle Chinese is a systematic gap resulting from sound changes from Old Chinese. There is no OC source that could yield such MC syllables. This gap has persisted right down into present-day Mandarin.

    Now that the "duang" has worked its way into that gap, and has already shifted from being onomatopoetic to a non-onomatopoetic adverb, it will be interesting to see if it has opened up the possibility for more "duang"/"tuang"-type syllables to enter the language.

  34. Victor Mair said,

    March 4, 2015 @ 9:33 pm

    Zev, it's true that most Mandarin onomatopoeic words are in the first tone, which is indeed an interesting phenomenon, and one well worth studying, but there are some that are in other tones (including the neutral tone), e.g.,

    da 咑 (dah-dy-rrr) sound made to urge donkey forward

    gézhi 咯吱 (adj.) descriptive of creaking sound (of dry branches, etc)

    gūji 咕唧 (v.) murmur or mumble to oneself or each other

    jíjí 唧唧 (n.) 1. the humming of insects, 2. sighing of persons, 3. whispering

    tà 嗒 to despair

    These examples are taken from this list:


    See also here for animal sounds, many of which are in sounds other than first tone:


    Incidentally, I should note that duāng is far from being the only onomatopoeic word in Mandarin / Chinese that is an "illegal" syllable according to the usual phonological rules. One that I must have heard my wife and her friends say hundreds of times is "pia", which is the sound of SMACKing. Furthermore, as was true of duāng until somebody dreamed up the crazy character of [成+龙], there's no way to write "pia" in characters.

  35. Zev Handel said,

    March 4, 2015 @ 10:23 pm

    Thanks for those, Victor. As exceptions to the general rule they are worth looking into further. (Interestingly, the prevalence of first-tone onomatopoeia even in words attested back to Middle Chinese shows that in MC these words not only had píng 平 tone, but also had voiceless initials. I would've thought there'd be more sound-imitative situations in which a voiced or breathy initial would be appropriate.) Some of the examples you listed are rù 入 tone, which makes sense in cases where you want an abrupt or sharp sound, like the "kak" (咯) of a snapping branch.

  36. Zev Handel said,

    March 4, 2015 @ 10:26 pm

    English, of course, also has onomatopoetic words that violate English phonological rules and are therefore difficult or impossible to represent in our writing system. These include the "raspberry" (bilabial trill) and the clicking of disapproval (already mentioned in this comment thread). The latter is often written as "tsk tsk", which is so far from an accurate representation according to the normal rules of English writing that as a boy I thought that the word written as "tisk tisk".

  37. Victor Mair said,

    March 4, 2015 @ 10:30 pm

    From the time I began to learn Mandarin half a century ago, and especially after I got involved in script reform efforts soon thereafter, I always wondered what would happen to the range of permissible syllables if a digraphia (characters plus Romanization) developed in China. I predicted that the number of "legal" syllables would increase, and even that the whole system of syllables might begin to break down.

    Now that practically everyone in China seems to be familiar with the alphabet (because of Roman letter inputting, primary school reading and writing with Pinyin, widespread learning of English and other foreign languages, etc.), it is possible for a word like "duāng" to be written out in Roman letters and used by hundreds of millions of Chinese. And there are many others, such as "biang" — not the one for the type of Shaanxi noodles that has a weird, extremely complex character (some say it has 57 strokes, some say 62 strokes…) assigned to it — which is quite popular nowadays (and somebody also invented a character for it too, though I don't know anybody who actually writes that untypable character [it's just a curiosity]), and which people simply write out in Romanization, as they are doing with "duang". I'll have to write a separate post on that and related matters, probably after I come back from a busy trip to Chicago.

    Suffice it to say for the moment that, with "duang", I think the floodgates to the introduction of new sounds and syllables have begun to open.

  38. Victor Mair said,

    March 4, 2015 @ 10:35 pm

    "The Story Behind ‘Duang,’ a Meaningless Word That Went Super-Viral in China"


  39. Zev Handel said,

    March 5, 2015 @ 2:01 am

    Victor, I hope you don't mind that I disagree completely with what your idea that permissible syllables are connected to the writing system. Throughout the history of Chinese (as in the history of every other language), extant syllables and phonotactics have changed over time. Various processes (borrowing, dialect mixture, contraction, sound change, onomatopoeia) have brought new syllables and new syllable types into the language. Quite a few have been mentioned in this comment thread already, including zán, béng and biáng. We can add to them liǎ 倆, hā 哈 (originally for transcribing Sanskrit), and the neutral-tone reduced syllables like de 的. Having a syllable-based rather than phoneme-based representational system does not affect this process, does not encourage or inhibit it. Whenever new sounds have come into any one of the Chinese languages, new characters have simply arisen to represent them. This process can happen quite easily in the absence of alphabetic writing. "Duang" is not a new phenomenon, and I seriously doubt if it represents a watershed moment of any sort. Gaps in the language that native speakers feel are pronounceable can get filled, and the writing system has nothing to do with it.

    On the flip side, no matter that you can spell a made-up syllable like "zhing" or "chüe" using pinyin conventions, I am very skeptical that Mandarin speakers will start combining retroflex initials with high front vowels any time soon (i.e. in the next several generations). Alphabetic representation or no, these violate pretty fundamental constraints on what sounds can co-occur for Mandarin speakers.

    This really shouldn't surprise us. All the languages we know that have long histories of being written alphabetically still retain idiosyncratic phonotactic and syllable-structure restrictions which change only slowly (if at all) over time. English speakers are not going to start pronouncing "sr" clusters at the beginning of words even if we can trivially spell them: give "Srebrenica" to an English speaker and you'll get a "shr" back. We don't do "sr" even though we do "shr", "tr", "pr", "gr", "dr", "cr", etc. Give "Kyoto" or "Tokyo" to an English speaker and you'll get three syllables back (toe-kee-yo), even though we have no trouble with tautosyllabic "ky" if the following vowel is "u" (e.g. "cute"). [Likewise with Hyundai. It starts with the same sound as at the beginning of "huge", but we can't handle the "hy" before other vowels, so the "y" sound gets dropped.] Give "tsunami" to an English speaker and you'll get "soonami" back, even though we have no problem with non-initial "ts" (e.g. "cats").

    If pinyin didn't exist, the word duāng would still be spreading in its character-written form.

  40. Michael Watts said,

    March 5, 2015 @ 7:36 am

    Zev Handel –

    why are you claiming that pinyin yun and its derivatives differ (at the end of the syllable) from wen and its derivatives?

    The zhuyin model of these is very clear:

    There is a sound that occurs in syllable final position, en. The zhuyin character for this sound is ㄣ. Obviously, it consists of a vowel followed by a nasal, but they are treated together.

    The en sound may be preceded by one of two glides: -u- (represented in zhuyin as ㄨ, which is also the (full) representation of the syllable wu), producing the -wen series of syllables, and -ü- (which zhuyin represents as ㄩ, which is also the representation of the syllable yu), producing the -yun series of syllables. The -yun series may be preceded by palatal consonants, and the -wen series may be preceded by other consonants, so zhuyin represents the syllable jun as the sequence ㄐㄩㄣ, j-ü-en, and the syllable shun as the sequence ㄕㄨㄣ, sh-u-en. There is no option to follow the vowel ü directly with a consonant; it must be followed by another vowel. And this is true in spoken Chinese (disclosure: my exposure to spoken chinese mainly comes from high school and college students in shanghai, as well as a few tutors with various backgrounds) — wikipedia, in its page http://en.wikipedia.org/wiki/Comparison_of_Chinese_romanization_systems , lists the IPA pronunciation of xun as [ɕyn], but that is obviously incorrect; there is another vowel between the [y] and the [n]. (To me, with my native-AmE ears, it's the reduced vowel… such is life.)

    I don't see any support for the idea that yun has only two sounds where wen has three; the native speaker judgment, and my own ears, are both solidly on the side of three and three.

    On a totally unrelated note, to me a raspberry is not a bilabial trill; a raspberry requires that you protrude your tongue from between your lips and then do the trill.

  41. Michael Watts said,

    March 5, 2015 @ 7:38 am

    postscript: obviously, -ü can be followed by a consonant if it terminates the syllable and the next syllable begins with a consonant. But within a syllable, it can't be.

  42. Michael Watts said,

    March 5, 2015 @ 7:43 am

    second, important, postscript: as I peruse the wikipedia comparison of chinese transcription systems, I notice that the syllable xin is also represented with ㄣ, as x-i-en (ditto, mutatis mutandis, yin and the rest of that series). To me, this is an "error"; there is nothing between the /i/ and the /n/. So I'll have to fall back on the fact that yun, xun, and the rest of that series are in fact pronounced with a glide-vowel sequence; the zhuyin usage evidence is not so strong.

  43. Victor Mair said,

    March 5, 2015 @ 7:53 am

    "If pinyin didn't exist, the word duāng would still be spreading in its character-written form."

    Would it? By the hundreds of millions of occurrences within a week? Somebody made up a funny character for it, but outside of limited handwriting (which people don't do much of nowadays) and a restricted amount of image copying, who is using that character? It is treated as an oddity, but "duang" continues to spread like wildfire.

    And who is writing "biang" (not the noodle one) in characters, even though it too is fairly common?

    Zev, I knew that you would "disagree completely" with me, because we've had this conversation before. But the fact of the matter is — for the reasons I adduced in a previous comment — digraphia IS developing, and it is enabling the writing down of unusual sound configurations.

    Throughout history, there was only one common writing system in China, and that was the characters, whether there were 2,000 or 10,000, or 100,000 of them. Yet, as you say, and as I certainly have said on numerous occasions, interesting combinations of sounds not covered by the characters have continued to develop IN SPEECH. Many of these sounds existed only in speech but never got written down, though for some new characters were devised. This disparity between what is spoken and what can be written in characters is especially the case for languages like Cantonese and Taiwanese where there are many morphemes for which there are no characters. I'm not a believer in the faith of běnzì 本字 ("original characters") — see below for some references.

    BTW, when I was a lad growing up in Ohio, I knew how to pronounce "tsk-tsk", and I also knew how to write it. My wife and her friends all knew how to say "pia" ("smack"), but there's no way to write it in characters, and it's not even a permissible syllable according to the traditional sound system. Now, with the rise of digraphia, it's easy to write "pia", and that's a perfectly adequate way to represent the word that my wife and her friends said hundreds of times quite naturally.

    [VHM added 3/15/15: When I was a lad growing up in rural Ohio, I not only knew how to say and spell "tsk-tsk", I also knew how to speak and spell "tsetse fly". I got to know about tsetse flies from the owner of a gas station on my paper route who had contracted sleeping sickness in Africa during WWII.]


    "Biscriptal juxtaposition in Chinese" (8/17/14)


    "Kiss kiss / BER: Chinese photoshop victim" (7/22/14)


    "How to Forget Your Mother Tongue and Remember Your National Language"


  44. Zev Handel said,

    March 5, 2015 @ 8:01 am

    @Michael Watts, the zhuyin and pinyin transcription systems are based on two different (and equally valid) phonemic analyses of Mandarin pronunciation. I was describing the pinyin transcription system, which is based on a treatment of the finals of 音 and 暈 as consisting of a single main vowel (IPA [i] and [y] respectively) followed by coda [n]. This is because most of the earlier discussion had been mediated through pinyin representations.

    The zhuyin transcription system is based on a treatment of these same two finals as consisting of an onglide (IPA [j] and [ɥ] respectively, i.e. short versions of [i] and [y]) followed by main vowel [ə] and then coda [n].

    Both are valid phonemicizations. You can't decide between them based on what you hear. Even if you agree that 音 is pronounced [in], you can still treat this phonemically as /iən/ -> [in].

    One possible way of deciding between the phonemicizations is by looking at rhyming. The problem with this is that native speakers don't always agree, in part because rhyming practice is based not just on phonology but also on culture, including historically valid rhymes. Rhyming judgments can also be influenced by spelling, so judgments of zhuyin users and pinyin users might differ. Still, you can ask some native speakers which subsets of the following four characters rhyme:

    a) 恩
    b) 音
    c) 暈
    d) 溫

    According to the zhuyin representation, all four have the same main vowel and ending (ㄣ) so they should all rhyme. According to the (base) pinyin representation (a) and (d) should rhyme with each other, and nothing else should rhyme.

    It would be an interesting exercise to try to survey people you know (without telling them why you are asking, as it may influence their responses) to get their responses.

  45. Zev Handel said,

    March 5, 2015 @ 8:03 am

    As for the raspberry, that's interesting. I always called both sounds raspberries — that is, with or without the tongue sticking out and serving as an articulator. I didn't know there were some English speakers for whom there was a distinction!

  46. Zev Handel said,

    March 5, 2015 @ 8:17 am

    @Victor, and I knew you would disagree with my disagreement! That's fine, of course. I do agree with you that digraphia is developing and that pinyin is now an alternative method for representing new pronunciations, especially in the computer-encoded age. I don't dispute the effect that pinyin is having on written Chinese. I think where we disagree is that I don't see that pinyin is enabling changes to the spoken language that were previously impossible — I don't see Mandarin suddenly acquiring all kinds of crazy non-Mandarin-like syllables, any more than I see English suddenly acquiring all kinds of crazy non-English-like syllables. I would make four points:

    1) There is nothing in the Chinese character writing system that precludes the invention and spread of a character for "duang". Indeed, we have already seen that happen. Right now that spread is done by means of graphic image files instead of Unicode. But the Unicode folks are right now working hard to get the character into Extension F! It will be soon.

    2) "lia" was until not long ago as illegal a syllable as "pia" (note there is no tia, dia, or nia). That did not preclude the invention of a character, 倆, to represent it. That character has since had a successful life. Should people have wished to write down "pia" in the pre-pinyin era, they could have done so by inventing a character. (For all I know, "pia" has been written by character at some time and some place. There's no reason it couldn't have been.)

    3) Chinese has been represented in alphabetic scripts of one sort or another for many hundreds of years. The option of writing the sounds of novel syllables by means of an alphabet, instead of by inventing new characters, has been available to Chinese speakers long before the modern era. But there was no need for it. Colloquial speech could be recorded by means of newly-invented characters, and often was.

    4) Cantonese can be, and has been, written in Chinese characters for centuries. The system works so well that it is picked up by native speakers without the need for any formal instruction, and is widely used (as you know!) in advertisements, comics, subtitles, scripts, etc. etc. For the last several centuries, Hong Kong Cantonese speakers have been intimately acquainted with the Roman alphabet and with mechanisms for transcribing Cantonese using that alphabet. But Cantonese speakers still don't say "lift", they say "lip". They still don't say "tart", they say "taat". The basic phonology of a language is not so easily transformed!

  47. Zev Handel said,

    March 5, 2015 @ 8:27 am

    @Michael, here is another test you can use. In a Mandarin syllable containing both an on-glide and a main vowel, it should always be the main vowel that lengthens if the whole syllable is lengthened. The short on-glide remains short. (Same with a short off-glide after the main vowel.)

    So, for example, if you ask a native speaker to hold the syllable jiāng for four or five beats, you will get jiaaaaaaaaaaaaang, not jiiiiiiiiiiiiiiiiiang. For duān, you will get duaaaaaaaan, never duuuuuuuuuuuuuan. For guai: guaaaaaaaaaaai. For mei: meeeeeeeeeeeeeeei, never meiiiiiiiiii.

    So you can ask a native speaker to stretch out a syllable like jūn 君. If you get jüüüüüüüüüüüüüün, then the ü is probably a main vowel, as pinyin supposes. If you get jüeeeeeeeeeeeeen, then you've probably got an onglide plus main vowel [ə], as zhuyin supposes.

    You can try the same thing for chūn 春. Do you get chuuuuuuuuuuuuun, or do you get chueeeeeeeeeeeen?

    (What I think you hear in the transition from the vowel to [n] in a pinyin syllable like xun or jun is not a main vowel schwa. It's a brief [i] sound that is not a target pronunciation, but the result of unrounding the lips, a necessary transition from the lip-rounded vowel [y] to the unrounded consonant [n]. As the lips unround during the movement from [y], a short [i] sound results.)

  48. flow said,

    March 5, 2015 @ 1:57 pm

    @Zev Handel that is an interesting test. But what is the outcome for, say, the syllable "go" as in "and if you should goooo…" sung by a native singer of An. or Brit. English? Do you hear it / produce it with a long [o], a long [u], or a [ou] with a dramatically slowed-down transition time? As for "yuin", my feeling is that a prolonged version should both stretch the "ü" and the "i" part ("yuin" is [jüin] for me, not [jüen]).

  49. flow said,

    March 5, 2015 @ 2:40 pm

    @Zev Handel concerning rhyme classes, i've uploaded a sketch of MSC rhyme classes here: https://github.com/loveencounterflow/phonologie-auslaute-chinesisch#rhyme-classes; i'd really enjoy criticism of that schema.

  50. flow said,

    March 5, 2015 @ 2:51 pm

    @Michael Watts concerning the vowels of la, le, li, lu, lü etc and the "two vowels" model, the short answer to me is that these syllables have a simpler structure that syllables like lai, lei, lou, lüe and so on.

    The effect is that in the simpler syllables, you have more options to put into the single available position (in Mandarin, at least five), while in the complex syllables, what you gain in complexity by adding more positions to the syllable final is in part counteracted by a reduction in what each position can accommodate. In complex finals, the medial position can only have i, u, ü; the code can only have i, u, n, or ng, and the nucleus one of *two* possibilities: a and "not a".

    Two considerations corroborate these findings: (1) we find that when we look at consonant clusters across languages, those clusters do enlarge the overall range of possible utterances, but the price to pay for the added complexity is that not all consonants can partake in all clusters; rather, the number of clusters is typically quite reduced. This is in analogy to what we find in MSC finals. (2) When we compare moderately complex with maximally complex Mandarin syllable finals, we find more restrictions in the latter: for example, you only get one "i" and one "u" per syllable, and in a syllable with a medial "ü", neither "i" nor "u" may appear in the coda.

    Does that make sense to you? I uploaded some materials to https://github.com/loveencounterflow/phonologie-auslaute-chinesisch which i plan to make more accessible and update over time.

  51. David said,

    March 6, 2015 @ 2:22 am

    @Zev Handel Actually both diǎ 嗲 (as in 嗲声嗲气) and niá (3rd person pronoun) are words in Mandarin. I think there is a character (俗字) for niá but I cannot recall it right now. I think 吕叔湘 might have mentioned this word and its origin as 人家.

    Also more on the word duāng, it is not a new word and it seems to me to be a pretty common onomatopoeic word. It originally imitates the sound of a blunt force, for example: 他duang地一脚把门踹开了. But Jackie Chan's pronunciation is apparently influenced by the English "boing", and he used it to mean something that's slightly different from its original meaning.

    Besides pia (sometimes written as 啪), there are many more words that cannot be written down with a Chinese character, e.g. bia (to paste), tsei (to break), piu (sound of a bullet),and many more.

    I think in the ancient rime tables there are sometimes possible combinations with no Chinese characters. These were sometimes noted in the tables or maybe in notes written by later authors. Those might be the equivalents to Modern Chinese words such as "duang" and etc.

  52. Lew Perin said,

    March 10, 2015 @ 5:09 pm

    Regarding the provenance of duang, and what might be called its place in Chinese culture, an acquaintance of mine has quite an earful (in Chinese):

    “Duang” 之我解

    "Duang" 的意思:形容对人,物的“冲击”,“撞击”。

    原本来自北京人(或北方人)对声音的模仿。例如:“duang"的/地一声就撞车了;”duang" 的/地一下就撞上去了(记得侯宝林的相声用过这个duang 音)。“duang”用来形象地描述两个或多个物体撞击时的“瞬间状态”。这种用声音词来强化和色彩化一个动作结果的用法其实很多,只不过有些声音有相对应的汉字, 例如: “扑哧”一声笑了, “垮差”一声倒了,“呼啦”一下儿围了上去,“哐”的/地一下就撞上了; 有些声音则没有对应的汉字,例如: “duang”, 还有“biang-biang” 面。记得小时候我们称北京的“有轨电车”为“diang-diang” 车 —- 取音于有轨电车的“敲钟”声音。

    成龙,一个香港武打演员,在为大陆的洗发液广告中用了这个音,夸大该洗发液对他头发的“强力”作用,用 “duang” 来形容洗发液的效力,以增强广告的效果。我想这个词应该是大陆编剧取用的。成龙本人在此前应该不知道这个“北方人的用语”。目前网络闲人就跟着这个广告 “延伸”运用这个“音字”来形容一个形容词的“超强烈”作用。其实这是“非规范”用语。由于中国有钱了,于是"Chinglish",和 “biang”, “duang” 这些 “伪汉语”也被中国人们“任性” 地使用起来了。我认为这反映了当今中国人在文化上的 “退化”现象。遗憾的是老外不明就里,也把“糟粕”当成了“精华”。

  53. Zev Handel said,

    March 11, 2015 @ 8:03 am

    Responses here to @Michael, @flow, and @David, in that order.

    @Michael, I don't know the answer to your question about "go" in various English dialects. I know that in singing, one learns to prolong the nuclear vowel rather than the off-glide in words like "high" [-aj] and "how" [-aw]. To my ear, prolonging either the nuclear vowel or the off-glide when singing "go" [-ow] sounds okay to me. I wouldn't necessarily assume that what works as a test in Mandarin works as a test in English. Maybe someone else on this list can answer your question better than I.

    @flow, thanks for posting this material. I'm afraid I don't have enough context to fully understand what is represented in the charts and what some of the abbreviations in the caption stand for. Are these groupings based on an analysis of rhyming texts in MSC? It's not clear to me why anything in Baxter 1992 is relevant to MSC rhyming. Maybe you can explain a bit more?

    You've clearly done a major research project on this topic. And I see from the beginning of your thesis you quote Chao's classic paper on the non-uniqueness of phonemic solutions. Do you feel that you've found something that previous scholars have missed which helps you to investigate the problem of the phonemicization of finals in a novel way?

    @David, the words you cited (diǎ 嗲 and niá (3rd person pronoun)) are great examples! But I would argue that they are not words in Pǔtōnghuà, they are words in colloquial varieties of Mandarin. They in fact demonstrate the same point I was trying to make about liǎ. There is no regular Middle Chinese source for Standard Mandarin syllables of the shape dia, tia, nia, or lia. From the perspective of strict historical development, these syllables shapes should no more exist than duang. But the fact that Chinese was written in characters did not prevent liǎ 'two people' from entering the language, gaining a written form, and becoming ultimately a part of the Pǔtōnghuà standard. (I suppose liǎ got created as a contraction of liǎngge, but I'm just guessing.) The fact that no extant character had the pronunciation "lia" did not prevent people from coming up with a character to represent the new syllable liǎ. My point is merely that pinyin is not necessary for a non-canonical syllable like "duang" to enter Mandarin and acquire a written form. It happened all the time in the pre-pinyin era.

    So back to diǎ 嗲 and niá. If people need to write these words down, they find a way to do so — a character can be invented that is conventionally associated with the word. (diǎ 嗲 is a case in point.) If the need to write these words is felt widely, then the invented character will spread. If the word is so colloquial or dialectal that hardly anyone feels a need to write it, then it will not acquire a standard written form (niá is a case in point)–it may have no written form, or multiple competing written forms. As you note, pia 'slap' is sometimes written 啪. If any of these words migrate from colloquial/dialectal forms into Pǔtōnghuà, and thus into written texts and dictionaries, you can bet they will soon have standard characters.

    In short: There is nothing inherent in the Chinese-character writing system that prevents ANY Chinese word or morpheme or syllable from being written. All it takes is for someone to create or borrow a character and get others to use it. With all the changes in pronunciation and vocabulary over 3000 years of written Chinese history, the writing system has always proven flexible and adaptable enough to accommodate everything the language has come up with.

    I have nothing against pinyin — it's a great tool. And of course it could be a functioning orthography for Standard Mandarin. But there's really nothing wrong or defective about Chinese characters, and in fact they have many advantages over alphabetic writing. I wouldn't personally advocate for one system over the other even if I felt I had standing to do so.

  54. Victor Mair said,

    March 11, 2015 @ 10:11 am

    @Zev Handel

    I don't think anyone on Language Log is advocating for Pinyin over Chinese characters. What we try to do here is describe what is actually happening in China linguistically, and what we are seeing is an emerging digraphia.

  55. Zev Handel said,

    March 11, 2015 @ 10:12 am

    @Victor, I absolutely agree.

  56. Victor Mair said,

    March 12, 2015 @ 8:21 pm

    @Zev, I'm glad that we've reached agreement.

    For whatever reason, "pia" has never developed into a "legal" syllable in Modern Standard Mandarin, and yet many people use it in their speech.


    Even though it is alleged that 啪 is sometimes called into service to write "pia" (how would one know?), I have never seen "pia" listed anywhere as an acceptable pronunciation of 啪, which I only see given as "pā".

    For the widespread acceptance of the alphabet in Chinese writing, see Mark Hansell, "The Sino-Alphabet: The Assimilation of Roman Letters into the Chinese Writing System," Sino-Platonic Papers, 45 (May, 1994), 1-28

    See also LIU Yongquan (Institute of Linguistics, Chinese Academy of Social Sciences), "On the Problem of Chinese Lettered Words, Sino-Platonic Papers, 116 (May, 2002).

    For a summary of an important, unpublished paper by Liu Yongquan, see "Creeping Romanization in Chinese" (8/30/12), from which I here quote the concluding portion:



    Liu Yongquan, a noted applied linguist of more than half a century's standing, is a specialist on zìmǔ cí 字母词 ("letter words", i.e., words used in Chinese that consist partially or wholly of Roman letters). In an unpublished review, Liu states the obvious, namely, that the sharp increase of zìmǔ cí 字母词 ("letter words") in the 6th edition of the Xiàndài hànyǔ cídiǎn 现代汉语词典 (Dictionary of Modern Chinese) reflects their increasing frequency in daily usage.

    Liu divides zìmǔ cí 字母词 ("letter words") into several categories:

    1. those that come directly from English or other foreign languages, e.g., WHO, CBD ("Central Business District")

    2. those that combine a foreign usage with a Chinese translation, e.g., B chāo B超("type-B ultrasonic"), X xiàn X线 ("X-ray")

    3. acronyms and abbreviations of Chinese terms, e.g., RMB (from rénmínbì 人民币 ["people's currency"]), HSK (from Hànyǔ shuǐpíng kǎoshì 汉语水平考试 ["Chinese Proficiency Test"])

    4. English expressions created by Chinese, e.g., CCTV ("China Central Television"), CEPA ("Closer Economic Partnership Arrangement"), ECFA ("Economic Cooperation Framework Agreement", a policy for managing cross-strait relations with Taiwan)

    It must be born in mind that all of these terms (and many others like them) are used freely and naturally by Chinese individuals while speaking Mandarin.

    Liu goes on to propose that China should create more acronyms directly from Mandarin pronunciation, e.g., BWDX Běidǒu wèixīng dǎoháng xìtǒng 北斗卫星导航系统 ("Beidou [Big Dipper] Satellite Navigation System"), instead of BSNS (following the English translation, which is the usual custom now).

    Liu's review concentrates on many other aspects of pinyin usage in the latest edition of Xiàndài hànyǔ cídiǎn 现代汉语词典 (Dictionary of Modern Chinese), such as proper orthography, details of pronunciation, grammatical niceties, and so on.

    When the most famous literary character of modern Chinese literature is named Ah Q (Ā Q 阿Q), and when the Chinese stock market could scarcely function without A, B, ST, and G, the alphabetphobic scholars who are waging a campaign against the inclusion of Roman letter expressions in the Xiàndài hànyǔ cídiǎn 现代汉语词典 (Dictionary of Modern Chinese) would seem have lost the battle before they began it.

    Essentially, what we are seeing is an emerging digraphia, with Chinese characters being used in parallel with Hanyu Pinyin for those purposes that are suitable to each of them. This is a natural process, one that will not be substantially slowed down by the naysayers nor measurably speeded up by the reform enthusiasts. It is happening because users of the language find it convenient and suitable to proceed this way.


    Finally, for the rise of digraphia in Chinese, see the many posts called up here:

  57. Victor Mair said,

    March 12, 2015 @ 8:22 pm

    Because it was written in Chinese and appeared nine days after the original post, few people noticed the remarks of Lew Perin's acquaintance. They mostly just repeat, somewhat awkwardly at points, what others had already said in earlier comments. But I wonder how Language Log readers would react to his / her closing sentiments:


    Qíshí zhè shì “fēi guīfàn” yòngyǔ. Yóuyú Zhōngguó yǒuqián le, yúshì "Chinglish", hé “biang”, “duang” zhèxiē “wěi Hànyǔ” yě bèi Zhōngguó rénmen “rènxìng” de shǐyòng qǐláile. Wǒ rènwéi zhè fǎnyìng le dàngjīn Zhōngguó rén zài wénhuà shàng de “tuìhuà” xiànxiàng. Yíhàn de shì lǎowài bùmíngjiùlǐ, yě bǎ “zāopò” dàngchéng le “jīnghuá”.

    其实这是“非规范”用语。由于中国有钱了,于是"Chinglish",和 “biang”, “duang” 这些 “伪汉语”也被中国人们“任性” 地使用起来了。我认为这反映了当今中国人在文化上的 “退化”现象。遗憾的是老外不明就里,也把“糟粕”当成了“精华”。

    In fact, this is a "non-standard" usage. Because China has money now, "Chinglish" and "pseudo-Chinese" words like "biang" and "duang" have been capriciously adopted by Chinese people. I think this reflects the phenomenon of cultural degradation of contemporary Chinese people. . Regrettably, uninformed foreigners also take the "dregs" to be the "essence."


    "Laowai: the old furriner" (4/9/14)


  58. Victor Mair said,

    March 14, 2015 @ 5:56 pm

    From Bob Ramsey:

    BBC World had a segment on the emergence of that word on their program this morning, but I was a little disappointed that all they did was marvel that the Chinese had just created a new character (advertised as visible on their website). It would have been nice if they’d taken a look at your blog beforehand!

    [VHM: This comment was originally sent to me on March 3, 2015, but I mistakenly posted it elsewhere.]

RSS feed for comments on this post