Hypercorrect Mandarin tones

« previous post | next post »

Here are two examples.  The first is the (in)famous one about the "Lion-Eating Poet in the Stone Den":

This is supposed to show that tones and characters are sine quibus non for intelligibility in Sinitic, except that nobody — but nobody — talks like this.  Without a character script, it's gibberish, and most Chinese throughout history have been illiterate.

The second example is one I've never heard of before, but could probably understand most of it upon first or second encounter if it were enunciated as clearly and correctly as here — even though it's a bizarre statement and is completely out of context:

The big difference between example 1 and example 2 is that the former is in Literary Sinitic /Classical Chinese (LS/CC) and the latter is in Modern Standard Mandarin (MSM) — grammar, syntax, morphology, vocabulary, the whole kit and kaboodle.  It is designed for oral / aural intelligibility.  Not so the one about the scholar eating ten stone lions.  It's book writing.  Not spoken language.

The young man's pronunciation is remarkably good, but he does make one very common mistake:  pronouncing "xue" as "shui".  I don't know how this happens, but I've heard many foreign speakers of Mandarin — some otherwise quite good — make this mistake.  What's funny is that, even when I point it out to fluent speakers, including native speakers, they don't really notice it.


Selected readings

[h.t. Laura Moreland]


  1. Philip Taylor said,

    October 15, 2023 @ 10:04 am

    “ The young man's pronunciation is remarkably good, but he does make one very common mistake: pronouncing "xue" as "shui" ” — as would I, presumably indicating that I failed to listen to my teachers, and/or the accompanying audio recordings (Kan Qian’s Colloquial Chinese), properly. Could you please describe how "xue" should be pronounced, and how that differs from the sound of "shui" ?

  2. John F said,

    October 15, 2023 @ 12:29 pm

    ‪Is xue pronounced shoo-eh?‬

  3. Jonathan Smith said,

    October 15, 2023 @ 1:17 pm

    first useful fact is that orthographical "-ue" syllables have high front medial [ɥ] i.e. ü glides, not high back [w]. second is that the orthographical "e" here is not a closing diphthong like [ei] as English speakers tend to produce but simply [e] ish. The rule that Mandarin syllables can't do U-turns like wVw / jWj is a useful generalization that applies here.

    orthographical "x" is also phonetic [ɕ] ish, conditioned by following [ɥ]/[j], which explains why orthographical "u" isn't an ambiguous representation here… but IMO this matters less to sounding OK then the first two facts above.

    Does this guy say a "xue" syllable in these videos?

  4. Duncan in London said,

    October 15, 2023 @ 2:40 pm

    In answer to Jonathan, indeed yes I think he doesn't say "xue" in either of the examples above – I think Victor is referring to this short by the same guy: https://youtube.com/shorts/cCf1k5WTyRM?si=b3Gs_EoaM6B17Jy6
    and yes he DOES pronounce xue as shui in this.
    Jonathan I think correctly puts it in IPA format – the simplest way to say it in "non-IPA" is that shui is pronounced SHWAY/SHOO-AY, but xue is more like SHOO-E with "E" as in "egg" (and in fact the OO is more like an umlaut EE-OO than a typical english "ooh" sound, and of course the SH is not really a SH but a slushy HS sound, but we are focussing on the difference between AY and E-as-in-egg).

  5. Victor Mair said,

    October 15, 2023 @ 4:12 pm

    @Duncan in London:

    Thank you for doing the leg/homework. A+!

  6. Philip Taylor said,

    October 15, 2023 @ 4:19 pm

    Thank you for that, Jonathan. Lacking any more authoritative source, I asked Google Translate to pronounce 血水 (xue shui) and while I can clearly hear your first difference ([ɥ] v. [w]), I cannot hear the second ("not a closing diphthong like [ei] as English speakers tend to produce but simply [e] ish").

  7. Vincent Chen said,

    October 15, 2023 @ 4:46 pm

    These two short videos are fun! Thanks for sharing.

    By the way, I don’t think these two shorts are so much about tones themselves, though. I think the guy was reappropriating the trope of "Lion-eating Poet" to follow a popular language meme, like this one in French: https://www.youtube.com/shorts/4jZE6y-uGJQ

    But indeed, Just like such a meme was originally meant to show how the said language is "easy," the tones being a difficult feature align perfectly with what these memes are trying to convey.

  8. Jonathan Smith said,

    October 15, 2023 @ 6:25 pm

    I see Wikipedia is writing [ɥe] vs. [wei̯]… the former is fine though in general I agree with Duncan that the value is lower than canonical [e]… whereas the latter is weird bordering on wrong; surely [wəi] is a better if imperfect representation…

    yes [ei] vs. [e] is not going to be easy to hear given English brain… it's also part of what's wrong with e.g. our Spanish :D

    unfortunately 'blood' is not going to be a good choice for google given the various ways this morpheme is pronounced… I tried some things and it is mostly producing xie3 including in contexts where it should be e.g. xue4… try 'snow' xue3 雪 vs. 'water' shui3 水

  9. Philip Taylor said,

    October 16, 2023 @ 2:42 am

    Yes, clear in that pair, Jonathan, and also clear (as regards [e]), in 大学 / 大学生. the two words/phrases with which I am most familiar that embed the "xue" sound. So my sincere thanks to you for clarifying Victor’s point.

  10. Chris Button said,

    October 16, 2023 @ 8:25 am

    @ Philip Taylor

    If it helps, it’s worth remembering that we speak in syllables rather than abstract phonemes. Alphabetic writing is very useful (and I am certainly not knocking it at all) but it is entirely artificial and not intuitive.

    So with “xue”, if you nail the front/palatal pronunciation of “x” the front/palatal pronunciation of “u” should follow more naturally without as much readjustment of articulators than would be required to produce a more English-style “u”.

  11. Chris Button said,

    October 16, 2023 @ 12:06 pm

    Athough I should probably add that historically i think it is the front/palatal nature of the "u" that gave the "x" its palatal articulation rather than the other way round. But it makes no difference to the larger point, and it is probably easier to think of it sequentially as in the order of spelling. You could of course nail the "u" and let the preceding onset follow accordingly because it will naturally be shaped by what is coming.

    I suppose it once again all goes back to what Ladefoged called the "phonemic conspiracy".

  12. Philip Taylor said,

    October 16, 2023 @ 1:20 pm

    A question for Chris, if I may ? When you speak of "nail[ing] the front/palatal pronunciation of “x”, may I ask if you are saying that it should be pronounced (very approximately) as/sj/ rather than as /ʃ/ ? If so, this goes against what we were taught by my first Chinese teacher, who told us that the correct pronunciation of "xièxiè" (thank you) was (once again, very approximately) /shɛ·shɛ/, and that the /sjɛ·sjɛ/ pronunciation typically affected by modern teenage girls was incorrect …

  13. Jonathan Smith said,

    October 16, 2023 @ 2:28 pm

    Chris Button is right in the sense that employing individual vowel "phonemes" like /e/ for Chinese and similar languages is not fit to purpose — as with all languages, we should build from the salient contrasts of which native intuition is (in part) composed. So "rimes" here can to some degree be treated as units. This is kinda the way the tradition has always had it. Cf. bopomofo, etc.

    But at least wrt synchronic analysis, pushing further here — say, deciding that /li/ does not involve /l/ + /i/ but rather /l/ + "medial" /j/, as some analyses do in order simply to reduce the number of "vowel" symbols to a minimum — becomes game-playing that flies in the face of observed contrasts and no longer bears any relation to a native speaker's intuitions about their language.

  14. Jonathan Smith said,

    October 16, 2023 @ 2:38 pm

    FWIW I think Chris's 2nd approach is better for learners — begin from rimes -i -jV -ɥV and apply the "high front" mouth/tongue position to the consonant. The result is not /ʃ/ but close to /ɕ/. These syllables, as is the case especially in Mandarin among the Chinese languages, involve a single or maybe rather a "minimally varying" articulatory gesture.

    Yes certain stylistic registers and some regional Mandarins use normatively frowned-upon /s/ here…

  15. Chris Button said,

    October 16, 2023 @ 7:44 pm

    @ Philip Taylor

    You should end up coarticulating sj as sʲ.

  16. Philip Taylor said,

    October 17, 2023 @ 1:11 pm

    “ You should end up coarticulating sj as sʲ ” — as in Cantonese Siu mai, Chris ?

  17. Chris Button said,

    October 17, 2023 @ 2:48 pm

    @ Philip Taylor

    I was suggesting you articulate [s] and [j] at the same time, which combines the voiceless alveolar fricative features of “s” with the palatal feature of “j”.

  18. Philip Taylor said,

    October 18, 2023 @ 12:15 pm

    Yes, I understood that Chris, but I was trying to map your suggestion to a real-world example with which I am familiar. Cantonese siu mai was the best example that I could come up with, where the [s] and [j] are effectively indivisible, whereas in English "assume" (for example), they are separate elements …

    It also seems to me that there may be an analogy with Czech "ř", which native Czech speakers insist should be pronounced with the [r] and [ʒ] pronounced simultaneously, but which many Britons (including myself) can approximate only by pronouncing them sequentially.

  19. Vampyricon said,

    October 18, 2023 @ 6:08 pm

    @Philip Taylor

    >Cantonese siu mai was the best example that I could come up with, where the [s] and [j] are effectively indivisible, whereas in English "assume" (for example), they are separate elements

    I would consider English "assume" an infinitely better example than Cantonese siu mai, because there simply is no palatal coärticulation / palatalization anywhere to be found in siu mai. Whereas in English, one can either say [əˈsjʊwm] or [əˈɕʊwm], in Cantonese, it is always [siːw⁵⁵maːj³⁵] with no palatalization whatsoever.

  20. Ethan Elasky said,

    October 18, 2023 @ 9:43 pm

    In the first video, the shì meaning room written wrong — he has written the simplified version of 實, which is actually shí and means real, instead of 室.

  21. Philip Taylor said,

    October 19, 2023 @ 3:28 am

    Thank you, Vampyricon, I stand corrected. Although Cantonese is spoken regularly in my extended family, it is not a language that I have ever studied or learned, and my pronunciation of siu mai is simply an attempt to produce the sounds that I hear; clearly I either do not hear the sounds correctly, or I fail abysmally in my attempts to replicate them.

  22. Martin Dorey said,

    October 27, 2023 @ 5:53 pm

    Finally I can clumsily enunciate 学生 and be understood by Duolingo (or is it Apple's speech-to-text engine? Any sufficiently obfuscated interface renders the parts of the technology indistinguishable from one another; I do notice that the interface and behavior is quite different with no internet). Thank you to the inimitable, indefatigable Philip Taylor and his (here) patient interlocutors. Something of what Chris Button wrote went in, even though much of the terminology and notation goes over my head, while the Wikipedia pages on linguistics send me spiraling down rabbit holes of increasing incomprehension. The bite-sized comments, often as mind-expanding as the original posts, keep me coming back here year after year. Perhaps shutting them off was part of the always entertaining Geoff Pullum's performance but, I fear through Prof Liberman's thankless efforts at moderation, it's often a wiser crowd here than most anywhere on the internet. I suppose having his critical notices attached to the performance made them harder to ignore.

  23. David Marjanović said,

    November 1, 2023 @ 11:45 am

    In Standard Mandarin, x is a dorso-palatal sibilant. There is no way to render this in IPA.

    I've only heard [ɕ] – warning: very small sample size! – from southerners who also don't retroflex (i.e. merge sh ch zh into s c z).

    However, there is some variation in how the dorso-palatal sibilant actually comes out. Teenage girls stereotypically go for a cutesy pronunciation that closely approaches [sʲ]; [sʲ] is Russian сь if that helps (…though not the reflexive verb ending сь, which is actually [s]…).

RSS feed for comments on this post