Tones for real

« previous post | next post »

For several years, John McWhorter has been studying Mandarin very seriously.  He and I have, from time to time, corresponded about the best, most effective, most efficient way to do that.  After years of assiduous learning, it seems that he has recently experienced a kind of satori about one of the most challenging aspects of acquiring fluency in spoken Mandarin:  the tones.

John writes:

"I'm taking intense Mandarin lessons now, twice a week, to really be able to do this.

But I'm having a tone problem.

On this, which is AMAZING — you must talk this up — I do it every day from "Elementary" which is so challenging I dread what the next levels are. It is the best language learning site I have ever known:

Popup Chinese

I hear real Chinese daily, and I can tell that quite simply, real Mandarin speakers do NOT use every tone. From the dialogues, it would seem that a good 60% of the time, the speakers just "talk," and you get the meaning from context. "Dajia hao," the greeter says, and frankly I do NOT hear the tones. "Mei banfa," someone says, and God, all I hear is stress on BAN and secondary stress on MEI.

But — teachers insist on the tones. They say that I have to start slow, and that even in rapid speech people are using the tones, just quickly. But I have learned from assorted posts of yours that this just could not be true.

So, I am thinking that Mandarin means first learning how to produce the tones but then learning how much you really need to use them in rapid speech.

So, my main teacher told me last week that I have two modes — what I could tell she meant as "citation / reading" mode where I use the tones, and then my "rapid" mode where she advised me I need to attend to the tones more — as in, when I try to SPEAK FOR REAL.

I feel like I'm imitating the people in Glossika and Popup Chinese. And I can tell there are places where I need to do the tones more to be comprehensible. But am I really supposed to use tone four for ZAI (location) when in rapid speech I know from walking New York streets that speakers just say ZAI the way anyone would, ignorant of tone at all?

So — my question is: how much must I attend to the damned tones in a sentence, as opposed to in citation, to really speak this language? I can tell the answer is somewhere between sounding like a pedant and sounding like an idiot. How did this this go for you? Is there a principle that I could internalize?"

 I certainly have my own opinions about all of John's questions, and I may later express some of them in the comments.  First, however, I'd like to hear what other Language Log readers have to say about the importance of enunciating the tones in everyday conversation.

Some related posts:

"When intonation overrides tone" (6/4/13)
"'Ni hao' for foreigners" (10/11/16)
"Dissimilation, stress, sandhi, and other tonal variations in Mandarin" (8/26/2014)
"Where did Chinese tones come from and where are they going? " (6/25/13)

There are many other posts on the origins of tones, their canonical forms, departures from the canonical forms, tones in the various topolects, and the evolution of tones.

A few posts that touch upon John's ideas concerning tones:

"McWhorter on the global linguascape of 2115" (1/26/15)
"John McWhorter responds" (1/29/15)
"Homophonophobia" (2/7/15)



38 Comments

  1. John Rohsenow said,

    February 5, 2018 @ 8:35 pm

    Again, way back in the Dark Ages, when I was in graduate school and beginning Chinese at U Michigan, one of the canards being bruited about in our Chinese linguistics class was that if Chinese really WERE dependent on tones, then in theory at least, it should be impossible to whisper it. So
    Prof. James Dew had the native speakers in the room play the child's game of "telephone", whispering a message sotto voce one to the next around the room. It emerged intact at the end, thereby laying at least that one myth to rest.

  2. A-gu said,

    February 5, 2018 @ 9:14 pm

    I learned the Zhuyinfuhao/Bopomofo keyboard layout early on, when it was designed to be unforgiving, and required tone input just to bring up the character list. That really helped ingrain tones into me.

    Between that and my classes, I was able to internalize tones enough to notice most spoken errors I make myself. Excluding words I'm just not really sure of, of course.

    My impression on the importance of tones: China is a huge landscape with a large varieties of native Mandarins, from Sichuan to Heilongjiang. And the way tones manifest differs widely. But they're undoubtedly important — I was often not understood or misunderstood with badly mangled tones.

    You're going to hear a lot of regional differences among Mandarin-variety natives, but most of them manifest regular correspondence that helps listeners recognize what's being said even when the tonal qualities differ from dialect to dialect.

  3. Jenny Chu said,

    February 5, 2018 @ 9:31 pm

    In Cantonese and Vietnamese, the two tonal languages I'm most familiar with, it seems to me that the distinction is not so stark/binary. It's not "use tones" vs "don't use tones" but, rather, the degree to which these tones can be affected/modified – in predictable ways. There is certainly a difference when, for example, you are shouting, or saying something sarcastically with exaggerated intonation, but it just as certainly does not overrule the original tone.

    Not sure if this is the same for Mandarin, however!

  4. dom said,

    February 5, 2018 @ 9:46 pm

    So basically he's saying, since *he* can't hear the tones, they must not really be there. Seems somewhat problematic….

  5. Neil Kubler said,

    February 5, 2018 @ 10:04 pm

    John is, of course, correct that "quite simply, real Mandarin speakers do not use every tone." In rapid, informal conversation, especially as spoken by older northern speakers, there are many (between a third and half) unstressed syllables. BTW, there is actually no such thing as a neutral "tone," it's just an unstressed syllable. However, I disagree when John says "a good 60% of the time, the speakers just talk and you get the meaning from context." No, on many of the syllables there really are tones, though often riding on a wave of intonation, so not so pronounced (pardon the pun). Note that John's teachers insist on tones when he is speaking SLOWLY. This is significant, because Chinese sounds weird if you pronounce many neutral tones but speak slowly. The fact that John is a non-native and in a classroom context is also relevant; the kind of classroom Chinese taught to foreigners has far fewer neutral tones than natural Chinese. But we really must know the underlying tones of syllables, so that when we speak slowly or want to stress syllables, we can do that. For decades I had believed that the pronoun for "we" is wo3men0, with a neutral "tone" on the men0. However, I am becoming convinced that even on this men there is an underlying Tone 2, so that when speakers pronounce this very slowly and deliberately on formal occasions (rare but it happens), they occasionally will say wo3men2. Also, it should be pointed out that the younger generation in all of China (Beijing included) is using far fewer neutral tones than the older generation. I think this is because of the influence of oral, group recitation of character texts in Putonghua in school, where every syllable is pronounced out loud in a very unnatural way. So what is poor John to do? He should start out slowly and learn the "proper" tones for words (including neutral tones where so indicated in his textbooks or dictionaries). Be patient, no hurry. Then spend several months or years in a Chinese-speaking society, gain fluency, and as he speeds up, his speech will gradually (if he keeps at it and has reasonable language talent) approximate that (though of course never attain) the level of a native speaker.

  6. Tim Leonard said,

    February 5, 2018 @ 10:15 pm

    John Rohsenow said, "one of the canards […] was that if Chinese really WERE dependent on tones, then in theory at least, it should be impossible to whisper it."

    Did no one test the theory to see if they could whisper tones with meaningless syllables? The difference between whispered 1st and 4th tones seems stark to me.

  7. Eric said,

    February 5, 2018 @ 11:42 pm

    In everyday conversation tones aren't everything, but they certainly aren't trivial. John Rohsenow raised the example of whispered Mandarin as evidence that, in theory, Mandarin can work without tones. True enough. But badly produced tones are not just *missing*, they may be actively *misleading*. Listeners have to not only fill in the missing tones, but also actively dismiss some of the ones they're hearing. That can be a potentially exhausting process, and native Chinese may need to 'practice' before they can really do it well. Of course, context can do wonders, but the more complicated and less predictable the stuff you want to say, the less helpful context will be. Even some simple things can be problematic. Walk into a store and say "I'm looking for [some word]" and butcher the tone on [some word]. Or hop in a cab and say "Take me to [some address]" and butcher the tones on [some address]. It's not likely this will go smoothly.

  8. Chris Button said,

    February 5, 2018 @ 11:44 pm

    "Mei banfa," someone says, and God, all I hear is stress on BAN and secondary stress on MEI.

    My suspicion is that the neutralization of the tone on "fǎ" is a result of the stress falling on "bàn" as /'ban.fa/. As such, John's comment here regarding stress is very apt.

    So, I am thinking that Mandarin means first learning how to produce the tones but then learning how much you really need to use them in rapid speech.

    I would say that "lexical tones" are affected and shaped by the environments in which they occur just like any other component of a syllable. However, they still need to be articulated within the spectrum of what renders them distinct from their counterparts even when they are distorted – this is perhaps most clear in their interplay with "intonational tones".

    @ John Rohsenow

    …if Chinese really WERE dependent on tones, then in theory at least, it should be impossible to whisper it

    My understanding is that information somewhat commensurate with tonal distinctions is carried in whispered speech too.

  9. Geoff said,

    February 6, 2018 @ 12:00 am

    You can adjust the pitch of the white noise you make whispering, to some extent, by shaping the mouth within the constraints set by the basic vowel.

  10. Geoff said,

    February 6, 2018 @ 12:10 am

    In respect to JM's problem: not a great Chinese scholar here, but my intuitive reaction is: as native speakers some of the phonemic contrasts we use are amazingly subtle, and we're amazingly highly tuned to notice them.
    Think "can" versus "can't" in varieties with the same vowel and an unreleased "t" . Or palatalised versus straight consonants in Russian.
    I would not be surprised if real use of tones is less than JM's classroom learning would suggest, but still more than he or I can hear.

  11. Jonathan Smith said,

    February 6, 2018 @ 12:12 am

    Pitch is the primary but not the only phonetic aspect of "tonal" distinctions in Mandarin, such that whispered speech is almost totally unproblematic.
    I agree with Neil Kubler's comments re: importance of stress loss in Northern Chinese and also Chris Button's re: environment. To this latter point, by comparison, it would be wrong to conclude based on the wide variety of specific phonetic realizations across contexts that the English distinctions p vs. b, etc., were in fact purely academic.

  12. Y said,

    February 6, 2018 @ 12:42 am

    Vowel reduction in English fast speech is also rarely taught, and is up to the learner to pick up by themselves. I imagine that tone reduction in Mandarin is comparable in principle.

  13. Fluxor said,

    February 6, 2018 @ 1:11 am

    While being able to hear subtle usages of tone (not to mention tone sandhi) is a challenge for learners, the issue of elision in regular speech further complicates the issue. Was that just a missing tone? Or a missing syllable? Or both? Or a merged syllable? If so, how did the tones merge?

    A classic example is 不要 bu2yao4 (don't want). What comes of the mouths of many speakers during regular speech sounds like b-yao4 (or biao4). It's not that the speaker didn't use the second tone from bu2; it's simply that the entire vowel and tone was thrown away altogether.

    Tones are there in regular speech. It's just not the "text book" type of tones. But it's there, it's fairly consistent within a certain accent region, and it's definitely important.

    Having said all that, Mr. McWhorter's problem is at much more basic level given his admission that he is unable to hear the tones in "dajia hao". Perhaps he can record his teacher saying "dajia hao" properly, then have her repeat using different tones on each syllable. He can then listen and re-listen until he figures it out. It's not something that comes by naturally for non-tonal speakers. I once repeated ma1,ma2,ma3,ma4 10 times to someone who asked what Mandarin tones sounded like. He still couldn't hear the difference after the tenth time.

  14. Bob Ladd said,

    February 6, 2018 @ 5:09 am

    This discussion is permeated by the presumption that tones are Weird and Mysterious. About whispering: voice is distinctive in most European languages (English bid / bit, fine / vine, rich / ridge etc. etc.), but we aren't mystified by the fact that you can pass a whispered message in English around the room. About connected speech: Mark Liberman has posted dozens of audio clips here over the years to show that lots of consonants – even whole syllables – just aren't there in the signal, at least not "there" enough to represent with a segmental IPA symbol. But that doesn't mean they're not really part of English.

    Think of a very reduced pronunciation of What'd he do? in AmEng, with only one (voiced) alveolar closure in the middle of the what'd he sequence. The closure crucially has to be just a little bit longer than a single /d/, otherwise it's going to sound like broken English What he do?. I think that's the analogue of what John McWhorter is doing in Chinese – he's inadvertently producing utterances like What he do? because he hasn't mastered the fine details of how the tones get reduced, and his teacher is reacting in the same way a native speaker of AmEng might react to a 20 ms. difference in the duration of that /d/ closure. It won't do to say that "the second /d/ isn't there", because in some way it is, and it's affecting the fine phonetic detail of the fast speech signal.

    Of course tones get massively reduced, but so does everything else. And like all other connected speech reductions, they can be remarkably systematic, which is why it's not simply a question of "they just talk". That's why John McWhorter's teacher says that he needs to attend to the tones more when he's in "rapid" mode, "speaking for real", even if, as a non-native speaker, he has trouble hearing the difference (and even if, as a native speaker of a non-tonal language, he has trouble believing that the tones really are just as much a part of the word as the consonants and vowels).

  15. boiko said,

    February 6, 2018 @ 6:28 am

    I agree with Bob Ladd and others that this is just normal language stuff. Language is squish. It has a lot of error-correction built-in, it has a lot of receiver-generated interpretation, and it can tolerate a lot of signal mangling. Yuu culd uvun dulutu vuwul dustunctuuns und stull bu undurstuud.

    I can't speak about Chinese tones but here's a tidbit about Japanese ones. We are a small team of non-native researchers doing fieldwork on the Japanese tonal system, otherwise known as "pitch accent", and how it varies between dialects. We were a bit distressed at how often our subjective judgment of the tone patterns wouldn't match. Looking at research on tone perception, we found data by Sugiyama et al. that, in a large-scale experiment, native speakers listening to their own (Tokyo) dialect failed to distinguish tonal minimum pairs about 30% of the time (much to the contrary of his own assumptions). Upon a closer look, the samples where most people made a mistake were the ones where the speaker didn't change the pitch very much between the "high" and the "low" surface tones. So it appears that the Japanese high tone/accent sometimes isn't very well marked in speech, and this poses no communicative difficulties due to context (as proved by the existence of non-tonal dialects in the east).

  16. David Marjanović said,

    February 6, 2018 @ 6:57 am

    It is true that completely unstressed syllables lose their tones. This holds not only for the particles, where it is conventionally transcribed in Pīnyīn and where the vowel qualities change too ( > de, liǎo > le), but also for cases like zìxíngchē "bicycle", where the tone on xíng "walk" is not mid-to-high rising at all, but low, simply staying where the top-to-bottom falling tone of "self" left off. I bet the same thing happens in dàjiā hǎo.

    On the other hand, a lot of what sounds like stress is actually tone. It's not hard to insert Běijīng with unreduced tones into an English sentence and have it come across as merely end-stressed.

    And then, as has been mentioned, there's intonation. When I first came across the "dry cups" expression associated with booze, I heard it loudly and clearly as "gāmbèi". So I figured it must be gān bèi. Nope, it's gān bēi with two high level tones; the falling I heard was real, but it's part of intonation. I haven't had enough exposure to Mandarin (let alone any other tone language) to figure out what, if anything, the surface difference between an underlying falling tone and a "hooray!"-intoned high level tone is – though I suspect it's length, the level tone being longer if all else is equal.

    if Chinese really WERE dependent on tones, then in theory at least, it should be impossible to whisper it

    Of course tones can be whispered, and so can intonation. Don't tell me you've never done that!

  17. boiko said,

    February 6, 2018 @ 7:29 am

    It makes sense that tones can be whispered, because tones are not just pitch.

  18. Chris said,

    February 6, 2018 @ 9:47 am

    I don't about Chinese, but…

    (1) teacher claims speakers are using all tones
    (2) student thinks speakers are not and "imitates" speakers
    (3) teacher can hear he is not using all tones

    If the student is correct, the teacher should not be able to hear whether he is using all the tones or not at speed. Maybe it's just me.

  19. Chris Button said,

    February 6, 2018 @ 11:59 am

    @ boiko

    . Upon a closer look, the samples where most people made a mistake were the ones where the speaker didn't change the pitch very much between the "high" and the "low" surface tones. So it appears that the Japanese high tone/accent sometimes isn't very well marked in speech, and this poses no communicative difficulties due to context (as proved by the existence of non-tonal dialects in the east).

    I think intonational focus in Japanese can confuse this a little since it can for example raise a lower pitched syllable to the degree that it can start to verge on a higher pitched one, but the distinction between the two is nonetheless maintained albeit in a reduced manner.

    @ Bob Ladd

    voice is distinctive in most European languages (English bid / bit, fine / vine, rich / ridge etc. etc.)

    That's an interesting example when a language like English is compared with say Spanish since the supposed "voicing" in English is very often not distinguished such that "fortis" and "lenis" are used in preference to "voiceless" and "voiced". As a result, the most salient distinction between words like beat and bead is very often the fortis clipping in the former that causes the vowel to reduce in length relative to the latter rather than any voicing distinction in the coda.

  20. Bob Ladd said,

    February 6, 2018 @ 12:57 pm

    @Chris Button: I know, but similar "redundant cues" are found with Mandarin tones as well.

  21. Victor Mair said,

    February 6, 2018 @ 1:32 pm

    "Just because he can't hear them means they aren't there???" Good lord, why are so many people so mean???

    It seems that several of the commenters are assuming that John has the typical problem of not being able to hear the tones. I can assure you that John has a very good ear and can both distinguish and produce the tones very well. What he does not hear are the tones used consistently in running speech — although it's evident from the o.p. that he is open to being told he's missing something there. So, yǎnjīng versus yǎnjìng? No problem. But his question is whether people observe that kind of difference in running speech.

  22. Y said,

    February 6, 2018 @ 2:41 pm

    This paper (J. Berry, Tone space reduction in Mandarin Chinese.
    J. Acoustical Soc. America 125, 2571, 2009) finds that Mandarin speakers may produce tones distinctly on the first token of a salient word within a discourse, but partly neutralize them in subsequent tokens. This is interesting but not surprising.

  23. Y said,

    February 6, 2018 @ 3:03 pm

    Marjorie Shostak writes in Nisa: The Life and Words of a !Kung Woman:

    There were four different click sounds and two throat sounds, all fairly easy to master with practice. But many words that sounded the same to our Western ears were actually distinguished by the use of four or more different tones—a dimension of seech that was difficult to hear, let alone master.

    By the time three months had passed, the !Kung sounds had become a little less strange, the clicks, glottal stops and fricatives a little more manageable, and I could finally hear the tones (although it took close to twenty months before I started using them correctly).

    (Whichever variety of Ju this is, it has far more than four click phonemes by modern reckoning, but evidently Shostak learned them well enough without proper phonemic analysis. Juǀ'hoan has four tones, but also some additional phonation contrasts on vowels.)

    If it's any consolation, tones may be harder to learn than clicks.

  24. Bob Ladd said,

    February 6, 2018 @ 5:04 pm

    Mandarin speakers may produce tones distinctly on the first token of a salient word within a discourse, but partly neutralize them in subsequent tokens.
    The same kind of reductions occur with "second mentions" of referring expressions in English (Bard et al., J. Memory & Language, 2000), but obviously reduction in English affects consonants and vowels, since there are no tones. Again, this provides no reason to think that tone is any different from other phonetic aspects of language.

  25. Christopher Coulouris said,

    February 6, 2018 @ 7:26 pm

    The great Jesuit missionary Matteo Ricci, the first European to master Chinese, also struggled with tones. He was deaf to the tones of Chinese. Later, when Cattaneo, another Italian Jesuit missionary arrived in China he and Ricci worked on a dictionary together. Unfortunately, that dictionary is lost. Cattaneo was an accomplished musician and had an ear for the tones of Chinese. Cattaneo developed the first system for marking tones. Soon afterwards Ricci began making progress with spoken Chinese. The Jesuits also developed a romanization system. On the cover of my edition of Jonathan Spence's The Memory Palace of Matteo Ricci is an example of Ricci's calligraphy and next to each Chinese character is its romanization with tone mark.

  26. Filter Fodder said,

    February 6, 2018 @ 7:53 pm

    I think one needs to realize, that in a sense, tones are phonemes with different allophones, just like, say, /t/s. Just because /t/s (in English) are sometimes aspirated, sometimes not, sometimes flapped, sometimes turned into glottal stops etc doesn't mean they're not there or not important. In a similar way, tones morph in all sorts of ways depending on context, emphasis and so on. Being able to perceive/produce pitch is just the first step.

  27. pt said,

    February 6, 2018 @ 8:42 pm

    @Victor Mair: The very fact that John thinks he's reproducing "real Mandarin as it's actually spoken", but his teacher can tell that he's not producing the tones correctly means that he is clearly missing something that's there in the speech of native speakers. It may not be the textbook four tones that's taught and used when spoken slowly, but clearly there is *something* that native speakers are adding to mark tones when spoken quickly that he is not hearing.

    I think it's very likely that some tones are abbreviated, some entire syllables may be elided, and there may be some complicated tone sandhi happening, but, regardless of how good you think his ear is, it's obviously not picking up everything in native speech.

  28. Chris Button said,

    February 7, 2018 @ 12:00 am

    @ Victor Mair

    But his question is whether people observe that kind of difference in running speech

    Leaving aside the undeniable influence of intonational tone, this really seems to be hitting on the question of stress more than any other environmental factors. An interesting comparison may be made with European "tone languages" like Swedish where only stressed syllables can carry lexical tone. Such an analysis may be in a sense flipped in an analysis of Mandarin whereby unstressed syllables may lose their ability to carry lexical tone. The question then is the degree of the loss – i.e. at what point is the lexical tone diminished to the degree that it is no longer discernible?

  29. RichardW said,

    February 7, 2018 @ 3:33 am

    In Yuen Ren Chao's "Mandarin Primer" (1947), all neutral tones are indicated in the first eight of the 24 lessons. (After Lesson 8, most of the neutral tones are unmarked.) For example, in Lesson 7, there is this sentence:

    Sweibiann nii gen .ta shuo sherm shyh, bugoan nii wenn .ta sherm huah, ta tzoong yeou huah gen .nii shuoh .de.

    No matter what you talk about with him, no matter what you ask him, he always has something to say to you.

    (Note that Gwoyeu Romatzyh (GR) is used for romanization in Chao's textbook. Chao uses periods (.) in front of syllables to indicate neutral tone.)

    What's interesting about this sentence is that it illustrates how carefully Chao paid attention to every single syllable in writing the transcript of the dialogs, because the same word "nii" (meaning "you") appears in this sentence both with its tone (nii : 3rd tone) and without (.nii), and the same can be said of "ta" (meaning "he").

    I have listened to the audio that goes with "Mandarin Primer" (mostly spoken by Chao himself, but with his daughters appearing occasionally). I have even listened to some of it slowed down to half speed, marvelling at how the tones are accurately produced even in rapid speech, and also at how true Chao's notation is when he indicates neutral tone for words like "nii" and "ta" that are normally written with 3rd and 1st tones respectively.

    Now here's another sentence. In this one, most of the syllables (10 out of 16) have neutral tone!

    Ranq .woo kann .a, ranq .woo kann hair fei.de.shiah.chiuh .fei.bu.shiah.chiuh.

    Let me see, see if I can still fly down.

  30. APOLLO WU said,

    February 7, 2018 @ 4:06 am

    Ancient Chinese used tones to avoid using polysyllables, probably in the name of brevity. Thus mai3 means buy and mai4 means sell. However, the trend of using more polysyllables cannot be stopped. As in a noisy environment, such as on the stock trading floor, such tonal difference could lead to confusion between vital difference between the buy or sell instructions. Thus, it is now the practice to use polysyllables of 'mairu' 买入 for buy and 'maichu' 卖出 or 'guchu' 沽出 for sell. Cantonese has nine tones (including those with ptk endings, while the relatively new Mandarin only has four. With polysyllables, the need for tonal accuracy has been reduced. Hopefully, Chinese will de-emphasize tonal accuracy in their use of more polysyllables.

  31. Phil H said,

    February 7, 2018 @ 5:35 am

    Bob Ladd said everything I was thinking, so I won't repeat that. But this is interesting as a pedagogical problem. I do some teaching of Chinese to colleagues at work, and there is a real problem when someone gets hung up on "how do Chinese people actually say it" as opposed to the rather simplified, canonical versions of language that we teach. (This can be pronunciation, grammar, even vocabulary.) I'm not sure whether there is something inherent about language that means it's better taught through simple archetypes; or whether it's just that most people learn language that way, starting as a child, so native speaker discourse tends to be built as extensions and developments of simple forms… but children are often not taught archetypal pronunciations, so in the case of phonology at least there must be some intrinsic feature…
    Blog hosts, I hope you don't mind this aside, but: personal to Chris Button, you must be a guy I went to university with. Hello! Drop me a line on Philip underscore hand at hotmail

  32. Chris Button said,

    February 7, 2018 @ 6:35 am

    @ Richard W

    I have even listened to some of it slowed down to half speed, marvelling at how the tones are accurately produced even in rapid speech, and also at how true Chao's notation is when he indicates neutral tone for words like "nii" and "ta" that are normally written with 3rd and 1st tones respectively

    That's interesting in terms of clearly destressing the pronouns as might be expected.

    @ Fluxor

    A classic example is 不要 bu2yao4 (don't want). What comes of the mouths of many speakers during regular speech sounds like b-yao4 (or biao4). It's not that the speaker didn't use the second tone from bu2; it's simply that the entire vowel and tone was thrown away altogether

    Such compression effects in connected speech seem to be noting the difference between "cannonical tone" (i.e. on a word spoken isolation) and actual spoken tone. Disregarding sandhi effects, when a word is uttered in isolation it will usually have much larger pitch variations (e.g. a rising tone may start lower and rise to a higher point) than in running speech when it often simply has less time to move as much. In English this is associated with unstressed syllable reductions (e.g. "interesting" or "ingredient" having 4 canonical syllables but usually 3 in speech) and such an analysis may be applicable to Chinese where it is the less stressed syllables that compress the most and whose tonal contours may become less distinct to the point of then sometimes disappearing entirely.

    @ Apollo Wu

    With polysyllables, the need for tonal accuracy has been reduced.

    Kuki-Chin languages that retain more prefixal elements than their counterparts tend not to make so many tonal distinctions either since the prefix can make clear a distinction that is only made clear by tone in other languages.

    @ Phil H

    I think you might be right – Greetings!

    By the way, in terms of teaching Mandarin, the common high level onset on the first syllable of an intonational phrase combined with the falling, rising and falling-rising nuclear tones (e.g \mine as a statement, /mine as a question and \/mine correcting someone who mistakenly thinks its theirs) in English makes a nice crude comparison with the four lexical tones of Chinese and helps people to get their heads around them.

  33. Chris Button said,

    February 7, 2018 @ 12:22 pm

    Incidentally the negative character 不 reconstructs in Old Chinese as *pə̀ɣ. Pulleyblank suggests that the ɣ (or ɰ) coda might not have been present underlyingly in such grammatical particles, but rather added epenthetically to make a fully-formed syllable when pronounced in isolation. A coda-less clitic such as *pə would therefore have been unstressed back in Old Chinese as well.

  34. David Marjanović said,

    February 7, 2018 @ 6:15 pm

    the negative character 不 reconstructs in Old Chinese as *pə̀ɣ

    Which reconstruction is that? Baxter & Sagart (2014) reconstructed 不 as simply *, did not reconstruct any [ɣ] or [ɰ] anywhere in the sound system, and did not use any accents – are you using the grave accent to indicate a Type B syllable?

  35. Chris Button said,

    February 7, 2018 @ 9:28 pm

    @ David Marjanović

    It's Pulleyblank's of course!

    The /ɣ/ or /ɰ/ corresponds to Li Fang-Kuei's /g/. I posted about this on a separate LL thread. While there were clearly no voiced stop codas in Old Chinese, its complete removal in reconstructions like Baxter&Sagart's is really a case of throwing the baby out with the bath water since some kind of velar fricative/approximant is still needed. Take for example 來 *rə́ɣ (whose /j/ coda in Middle Chinese would incidentally then go unexplained) and its undeniable association with 麥 *mrə̀k and 賚 *rə́ks.

  36. Philip Taylor said,

    February 8, 2018 @ 7:41 am

    Re John McWhorter's "[This] is the best language learning site I have ever known: Popup Chinese", I took a look (and listen) today for the first time. Whilst I can appreciate the site's potential value, I find that the PDF transcripts are of less use than they might be : although they are interlinear in nature, the two scripts are Latin (for the English) and Hanzi (for the Chinese) — pinyin is relegated to the "Vocabulary" section which follows the transcript. For myself (and perhaps for others), a tri-scriptal interlinearisation would significantly enhance the potential value of the transcript as learning resource.

  37. Victor Mair said,

    February 8, 2018 @ 4:51 pm

    I am in complete and enthusiastic agreement with Philip Taylor that it would be of enormous help for learners to have Pinyin versions for all sentences.

  38. Chris Button said,

    February 9, 2018 @ 6:36 am

    @ David Marjanović

    Just so as not to make Pulleyblank seem like an outlier and to give fair credit to Li Fang-Kuei, Li (1945 – OC loanwords in Tai) is approvingly cited by Pulleyblank (1995 – Role of Glottal Stop) regarding a shift of -əg > -əɣ > -əɯ (note Li's /ɯ/ here is the vowel rather than Pulleyblank's glide /ɰ/).

    Also regarding the grave accent, yes that is Type-B with an acute used for Type-A. It follows Pulleyblank's valid (and largely ignored) suggestion that the A/B distinction resulted from a prosodic accent giving syllabic weight on the first mora (Type B) or the second mora (Type A). The syllabic weight correlate is found in Kuki-Chin languages where its surfacing as a quantity (i.e. length) distinction has been confused by almost all researchers as underlying phonemic vowel length (Type-B) which convenient ignores the secondary lengthening that also occurs with sonorant codas (Type-A) in Kuki-Chin.

    By the way, the superficially redundant contrast in Kuki-Chin between surface long vowels with short codas and short vowels with long codas (the vowels also have a tense/lax quality distinction which is also ignored by almost all researchers) finds a nice typological comparison in Swedish although with a very different origin (in the Swedish case it results from a requirement for a basic length in stressed syllables versus their unstressed counterparts and hence you can have long obstruent codas which are not possible in Kuki-Chin where it is restricted to sonorants)

    It should also be noted that the surface vowel length distinctions were often created by analogy as something the language could exploit (this has a historical precedent ultimately connected to the ə/a ablaut) such that any attempt to directly correlate the surface vowel length distinctions in individual Kuki-Chin syllables with Old Chinese Type A/B syllables is doomed to be inconclusive.

RSS feed for comments on this post