Tang minus -ng, Tan minus -n

With a first-ever (experimental!) poll at the bottom of the post, and with apologies to Garfield minus Garfield

If Mandarin lost its -n and -ng suffixes overnight, would we all be eating phlegm and spitting candy tomorrow? It sure looks that way if you’re relying on a superficial view of Pinyin. I mean, once you take off the -ng and -n, what’s to stop you from doing tricks like this?

吃糖 = chī táng, eat candy
吐痰 = tǔ tán, spit phlegm / spit

morphs into

吃痰 = chī tán, eat phlegm / spit
吐糖 = tǔ táng, spit candy

But as any student of Mandarin phonetics knows, the /a/ in táng is not the same as the /a/ in tán. So as the -n or -ng fade out, which they are fond of doing in some people’s speech, the listener still has the vowel to help make the distinction.

A few of us here at Sinoglot started discussing this the other day after Victor Mair posted on Language Log about a Pinyin site whose sound samples he thought seemed to be reducing the -ng to -n in some cases. The Sinoglot consensus was that the enunciation of the syllables was well within the range of normal (listen for yourself by going there and pairing up, say, tàng and tàn). But Mair quotes former tester of Standard Mandarin Liwei Jiao as saying:

According to my training and experience, “bang4, hang4, lang4, qiang4, tang4, wang4, xiang4″ in your link could be judged as “wrong”, while “ang4, jiang4, kang4, nang4, pang4, rang4, zang4″ could be judged as “defective”. Basically your judgement is correct.

So whether there’s enough -ng on the -ng syllables appears open to debate, although as Sinoglot’s Randy Alexander notes in the comments:

Liwei Jiao’s perspective is that of a tester of the “national level” of the Standard Mandarin test. There are three levels for this test. Passing the highest level, what he calls the national level, is required of TV and radio announcers who broadcast nationally, and province-wide. They have to get 97% or higher to pass.

Quite right. Pǔtōnghuà is, by definition, a prescribed language, so it’s easy to say something is “wrong” if it doesn’t meet the definition of standard.

In the messy world of descriptive linguistics, though, it’s harder. Since the phonemic distinction between the -ang and -an endings is made up of phonetic differences in BOTH the vowel quality AND the nasal, an interesting question is: which has more weight? Could you make the distinction without the -n / -ng ending at all?

Well, give it a shot.

Recordings A and B, below, are from my daughter, an eight-year-old native speaker of Beijing dialect. I had her say tàn and tàng (or vice versa — that’s the mystery!) then clipped off the ending of each word so that only the initial consonant plus vowel remain. I don’t think you’ll have much trouble deciding which is which, though:

Recording A:

Recording B:

[poll id=”2″]

Now don’t cheat and listen to it before you take the test, but just for the record, here’s the original recording with -ng and -n intact:

Conclusion? If most native speakers can make the -ang / -an distinction without the -ng / -n at the end of the syllable, that seems to be a fair argument that if Mandarin eventually loses those nasal endings altogether, the language will make up for it with clearly distinguished vowels.


PS: A commenter on the Language Log post linked to a fascinating related article by E.G. Pulleyblank that I don’t have time to go into now, but I’ll leave a longish quote. If anyone knows where to find the Jie Zhang article that Pulleyblank references, that would be fantastic:

In a recent paper, Jie Zhang (2000) argues that in Mandarin Chinese the nasality induced on the preceding vowel by the /ŋ/ coda is perceptually more salient than the nasalization induced by the other nasal coda /n/ and that this is the explanation for the fact that when the retroflex diminutive suffix is added, resulting in loss of both preceding nasal consonants, the distinction between -Vn and -Vŋ is preserved by nasalization of the vowel in the case of the latter but not the former. Since degree of nasalization is not normally recognized as a possible distinctive feature, this requires the assumption that a surface phonetic contrast can emerge as distinctive in a morphological process. In the case in question, however, I shall argue that before adding the suffix there is already a featural contrast in the quality of the vowel preceding /ŋ/ as compared to /n/, namely pharyngealization, that is to say, secondary articulation of /ŋ/ by the pharyngeal glide [symbol unavailable]. This featural contrast serves as an enhancement to the distinction between the two nasal phonemes and is preserved when the suffix is added. The pharyngealization of /ŋ/ means that it has a uvular rather than a velar point of articulation, relatively close to the velum, which can also account for the spread of nasalization to the preceding vowel in the unsuffixed form, preserved along with the pharyngeal glide when the suffix is added.

unsuffixed form, preserved along with the
pharyngeal glide when the suffix is added.

20 responses to “Tang minus -ng, Tan minus -n”

  1. Karan Misra says:

    That’s really interesting. After listening to the recordings of your charming daughter, when I went over to the site you linked to (http://tool.httpcn.com/Zi/YinJie_nan/), and listened to their recordings, it sounded as if àn and àng just varied in the final nasalization. However, clearly your daughter (and many speakers) use a different vowel for tàn than for tàng (the former sounding closer to the vowel in “and” and the latter sounding closer to the vowel in “aunt”, when they are both pronounced in American English.) Are pronunciations given on that website you linked to are considered, perhaps, the “proper” ones? Those are the ones that were taught to me when I was learning Chinese.

  2. Considering the unbelievable variation in Chinese accents and dialects, I’m pretty sure removing the -ng from all -ang sounds would just turn putonghua into another random bendihua.

  3. ETH says:

    I’m not convinced that the vowel distinction holds up in all cases, whether the case involves native speakers or not.
    I suspect it works pretty well with -an and -ang, but not so well for pairs like -en and -eng.

    While I was in Taiwan recently, I spent some time trying to pry apart various sounds that tend to get lumped together (such as the expected [pinyin] pairs like si – shi, c – ch). I wanted to find out if my ear was off or if there really was a blending of pronunciation for some terms. So, I asked a number of friends to tell me the phoneticization (in their case, zhuyin) they would use for pairs like these:

    針 爭

    枕 整

    All the people I asked told me that the members within these pairs had identical pronunciation and they wrote out identical bo-po-mo-fo. (These were all adults who hadn’t had to use zhuyin much at all once they were out of school). Sometimes, people would correct themselves (usually after noticing how surprised I was) and add the -ng sound, but their first reaction was invariably that these four had different tones (first pair: first tone; second pair: third tone) but the same “spelling” in terms of romanization/phoneticization (pinyin: zhen).

    If the vowel were different in these cases, even if they drop the ‘g’, it seems like they would have reported different pronunciations and phoneticizations.

    That’s just my bit of anecdata–I wasn’t particularly systematic. And I only tried it with pairs that stood out to me as being conflated, and -an/-ang wasn’t one of those.

    My own pronunciation of -en and -eng does use a different vowel sound, but I couldn’t hear one in native speakers of Mandarin in Taiwan and apparently those I spoke with didn’t consciously recognize one in their own speech.

  4. Syz says:

    @ETH: Good point about how ng/n might be more important with some vowels. Maybe I’ll do the same sort of test with the others.

    I’m very interested in the “spelling” test you created. The problem with both Pinyin and zhuyin in this situation is that they can’t account for a phonemic difference in vowel quality. That’s what I was trying to say with the “eat phlegm” example where you leave off the ng/n: Pinyin only offers one “a” even though there are clearly two, at least in Beijing speech.

    Going back to Taiwan, it’s possible (I’m not saying it’s true — that’s what we need to test), that even your Taiwanese speakers who don’t spell the syllables differently actually can tell the difference. The difference would not be in the ng or not but in the vowel. If you feel really inspired it would be cool if you could record a local speaker like I did with my daughter. Just have them say your two words. If no one can differentiate them, well, I guess that’s the end. If people can, I’d guess it’s because of the vowel.

  5. Syz says:

    @Karan: good question but I don’t really know which ones are “proper” anymore, esp. since, as you saw in the quote, the former putonghua instructor thought they were wrong. Is there any absolutely authoritative putonghua model that’s publicly available? If you come across one, let me know.

    Re the linked pronunciations: I think the vowel might be closer than in my daughter’s speech, but still different.

  6. ETH says:

    Agree that my spelling test just hints that there may not be a difference, but doesn’t really show it.

    I may try the recording and cutting off method you used next time I speak with one of those friends I originally tested.(Unfortunately, this might be a while.)

    It might also be interesting to see if there’s a difference when the “test” word is spoken by itself (i.e., when the speaker knows that enunciation is being looked at) and when it’s used in a sentence and the speaker is being recorded but doesn’t know what the test has to do with.

    I wonder about -in/-ing. When it’s yin/ying, it sometimes sounds (again more in Taiwanese Mandarin speakers) like the sounds are the same or nearly so, and the vowel sounds the same. But when it’s something like bin/bing, the vowel and the -ng are both part of the distinction.

  7. Syz says:

    @ETH: I’ve never heard the yin/ying vowels as being the same, but I’m pretty sure there’s huge regional variation so it might well be true over there. I will get around to posting a sample of it from Beijing: the dipthong effect in ying is enormous.

  8. hsknotes says:

    “The problem with both Pinyin and zhuyin in this situation is that they can’t account for a phonemic difference in vowel quality. That’s what I was trying to say with the “eat phlegm” example where you leave off the ng/n: Pinyin only offers one “a” even though there are clearly two, at least in Beijing speech.”

    Zhuyin does account for a phonemic difference in vowel quality.

    pinyin an and ang both have an ‘a’, but zhuyin’s ‘ㄢ’ and ‘ㄤ’ are two different and unrelated symbols.

    Any pronunciation issue in this situation that you’d like to pin somehow on the pinyin system certainly can’t be that easily pinned on the zhuyin system.

  9. I haven’t had time to poll my Jiangnan 江南 flatmates and get recordings. But I’d say at least as I learned it down here, the -ang/-an ending does have different vowels. It’s yin/ying that gets all mushed together and has no vowel distinction.

    Maybe I’ll try to push the microphone in their collective face this afternoon.

  10. Syz says:

    @Hsknotes: I was a little careless, and I’m not blaming either system. I shouldn’t have used the word “problem” at all. I’m just saying that neither Zhuyin nor Pinyin shows clearly that the difference between ang and an (or between ㄢ and ㄤ) is both a difference in nasal and a difference in vowel quality. It might be that ETH’s Taiwanese informants really have no difference, which is why they would spell them the same. Or it might be that they maintain a different vowel but merge the n/ng. I just don’t think spelling alone answers that question.

  11. I’ve always seen the distinction between ㄢ and ㄤ as strictly consonantal and not in any way reflecting vowel differences. This is probably because in any case where I’ve seen Bopomofo explained, even as far back as a century ago, the transcribed vowel was always simple “a”.

    This is especially true in early uses of Zhuyin for transcribing Wu, in which case a third glyph (丄) is used for “-ang” when the quality of the vowel is different from that of ㄤ and ㄢ.

  12. hsknotes says:


    I think this touches on a much deeper question of what alphabets/syllabaries/etc actually “show clearly.”

    In an alphabet like the Latin one used in the english language, the letters are not fixed to single values.

    I think the problem you get into when say something like “neither Zhuyin nor Pinyin shows clearly that the difference betweeen…” is that you get an issue where you can say the latin alphabet in english doesn’t show clearly the difference between ‘t’ and ‘d’ is one of both voice and aspiration as opposed to merely aspiration.

    Your interest in this case is the ‘phonemic difference in the vowel quality’, but it could easily be any part of the language.

    With zhuyin, and pinyin as well I would say, the symbols, or the letters, are representations of a pronunciation, all aspects of the pronunciation, at some point in time. How people choose to pronounce those symbols, or letters, or words that they, or others would claim are composed of those building blocks, is another story. Even in places where the syllabaries and alphabets are essentially fixed to certain sounds, in theory, like many Indian languages, and I would argue pinyin and zhuyin, as opposed to english, you have pronunciations that wildly differ based on linguistic change, accent, etc.

    So. In english, does the alphabet make clear all the distinctions that exist between ‘t’ and ‘d’? If it doesn’t, is that really the right question to be asking? Doesn’t it in actual use, because of people’s knowledge of the sounds of the language and how they create a correspondence with the alphabet, “make clear” the two different sounds just as clearly as ㄢ and ㄤ do for someone who speaks a language that distinguishes these two sounds? (Note. Just like many southern places do not speak a language say that completely, or ever, distinguishes many sounds in Mandarin, or local languages, there are going to be places that speak a variety of mandarin where ㄢ and ㄤ really do not live separate existences.)

    There was more, but I forgot.

  13. A good example might be Cherokee where symbols are very representative of Latin letters but with distinctly different sounds, e.g. R = /e/, T = /i/, M = /lu/ etc. But then, it’s not really T but just something that looks like T.

    I guess my very irrelevant question is, were we lacking the massive amount of foreign influence on English spelling and vocab, would our interpretation of a letter’s corresponding sound really be any less fixed than IPA?

    hsknotes: Not sure I follow on the Indian languages part. Do you mean to say that devanāgarī doesn’t allow variance in the pronunciation of a particular glyph, or do you just mean there’s more of a 1:1 relation between letter and sound than you find in English?

    (p.s. Flexibility was always the thing I hated most about IPA. What good is a fixed system if it permits variance, after all.)

  14. Sima says:

    I’m a little surprised you hear no difference between the vowels in -an and -ang on the linked website. They sound pretty much as clearly different as the recordings of PBS to me. Could you have a check back there and tell us if you still hear them as being the same?

    As far as I can tell, certain pairs of rhymes, like the pair you mention, are simply not distinguished in certain regions/dialects. There’s no need to look for a difference in vowel quality if the rhymes are not distinguished at all.

    In Chinese teaching (in Mainland China, at least), I’m not aware of the rhymes being broken. That is to say, -an and -ang are introduced to the learner as whole rhymes and simply learnt as separate items. Of course, speakers of English, learning Chinese with the use of Pinyin, may well look at the spelling and think of these two rhymes as being made up of some kind of standard /a/ vowel + nasal final.

    Though a Chinese phonology course might analyse these syllables as /-an/ and /-ɑŋ/, I wonder whether it may be a mistake – not because of the distinction between the vowels, but because of the final consonants themselves.

    What I find really intriguing about the clipped recordings of PBS is that both come out sounding as though they are on the first tone. Does anyone else hear them that way? Would it be fair to say that the fourth-tone fall has not yet begun at the point the recordings end? If so, what does that suggest?

  15. hsknotes says:

    hsknotes: Not sure I follow on the Indian languages part. Do you mean to say that devanāgarī doesn’t allow variance in the pronunciation of a particular glyph, or do you just mean there’s more of a 1:1 relation between letter and sound than you find in English?


    In Zhuyin, and pinyin (if things like an and ang are properly looked at symbols) the symbols are representations of sounds. In english we can’t look at a letter and say what it’s sound is. We talk about the “names” of our letters as opposed to their sounds: ‘double u’, “h”, etc. What sounds does a “c” make? What sound does it represent? A “see” or a “kay or what?

  16. Codfish says:

    My experience is that the people who don’t have much of a n/ng distinction also tend to eliminate the vowel distinction. One of my friends, a Chinese-American from the Huaiyang area who still spends a lot of time there, has what I consider all three of the “standard Southern manglings of Mandarin” – she pronounces n and ng the same in finals, confuses n and l in initials (even in English, which is kind of hilarious), and tends to lose the h in zh/sh/ch.

    Her last name is 郑, which is a problem because it makes two of the problems come out. She doesn’t say “Zheng”. She says “Zen” – obviously not even a standard pinyin syllable, but that’s what I’ve heard. And the vowel she uses is, to my ear at least, distinctly the one I would use for -en endings (shen, zhen, sen, what have you) rather than the one for -eng endings.

    I’m in Beijing, so I don’t have much experience with the phenomenon, but I think that two Taiwanese people I know with the same regiosyncracies (is that a word?) in Mandarin do the same thing. Have other people encountered this?

  17. So then it’s an issue with English spelling, not with the Latin alphabet, right?

  18. hsknotes says:

    “I’m in Beijing, so I don’t have much experience with the phenomenon, but I think that two Taiwanese people I know with the same regiosyncracies (is that a word?) in Mandarin do the same thing. Have other people encountered this?”

    It’s a de facto standard.

    So then it’s an issue with English spelling, not with the Latin alphabet, right?

    I think it’s either an issue with word formation in languages like English that use the Latin alphabet, or perhaps just an issue with languages that use alphabets in general, as opposed to syllabaries. However, I don’t think it’s the second one. I think with an alphabet and and stricter, or even 1:1 correspondene, and strict enough rules on word formation, the issue goes away.

    So, strictly speaking, I suppose the problem is now with the Latin alphabet, per se, as it’s just a series of symbols no different than say zhuyin or pinyin. I’m not sure how it has been applied throughout history though, I’m not expert at Latin.

    All this hurts my head, but I like the term regiosyncracies .

  19. Karan Misra says:

    @Sima I went again and listened to more samples. The first time, I had only listened to àn and àng, and honestly I still think he’s pronouncing them with the same vowel. However, then I listened to tàn and tàng, and pān and pāng, and yes, I think there is a difference, but it’s very, very slight, and I hope I’m not making up a difference simply because this time I was really listening for it.

  20. Ross says:

    Interesting article. I had a Shanghainese boyfriend who would replace most “n” finials in Mandarin with a very nasal “ng” and had a hard time distinguishing between these sounds both in speaking and listening. I’m sure this is because of the differences in Shanghainese/Wu dialect?

Leave a Reply