Cantonese tones

« previous post | next post »

If you ask Modern Standard Mandarin (MSM — Guóyǔ 國語 / Pǔtōnghuà 普通话) speakers how many tones there are in their language, most of them will tell you without much hesitation that there are four tones (1st, 2nd, 3rd, 4th) plus a neutral tone.

Chances are, however, if you ask a Cantonese speaker how many tones there are in their language, they will not give you a clear answer, or if they do, it will differ from what other Cantonese speakers claim.  That has always been my experience over the years, but I just did a little survey to reconfirm my earlier impressions.  The results are rather more amazing than I expected them to be:

I will list the replies from respondents roughly from those who are older to those who are younger.  Some just give a number, but some also give an explanation, in which case I include it as well.  In addition, I will note what I know about their occupation if I have that information:

[N.B.:  When the respondents speak of "entering tones" (rùshēng 入声), that refers to the -p, -t, -k endings, which are totally absent in MSM.]

All responses are from lifelong native speakers of Cantonese, except #4, an American who has been completely fluent in Cantonese for decades.


11 (?)    Don't know! I think (or I heard) that it's a lot! I'd say… 11. (librarian of the history of science)


9    [no comment] (professional)


9    Is that a trick question? …

Answer: 9 based on traditional counting that includes 3 Rusheng but 6 based on phonological system.

Is that what you are looking for, or are you referring to tonal mergers that may be reducing the total count? And that seems to be complicated. (professor — linguist specializing on Cantonese)


   I know the "right" answer is nine, but the entering tones are really pronunciation differences, so let's say six – except that the first tone can either be high level or falling. (professor — linguist specializing on Cantonese)


11    [no comment] (professor of Chinese language and literature)


7+    unless you count the rusheng (the p, t, k endings) as tonal.  Please do let me know the correct answer. (professor of Chinese literature)


   Depends on if you count the hard -k and -t endings that don’t exist in Mandarin?


ma = mother, sesame and same tone is so-so and also paternal grandmother, twins, horse.

mat = wipe, sock

mak = ink

So I count 7 there but there are some other tones that “ma” might not be used for…

(professional — for greatly expanded remarks by this respondent, see below at *)


11    I think, my answer is 11, if we include rùshēng 入聲 ("entering tones") and put the biàndiào 變調 ("modified tones") into it.

But I am not quite sure…… (professor of Chinese philology)


5 (?)    Haha I think I can recognize 5. I've heard the 6th… but it's so faint to me that I don't really distinguish it. On the other hand my parents have always said 9! (recent college graduate working in finance)


Zero idea    But somebody told me it's supposed to be 8. (graduate student in Chinese language and literature)


9 (?)    I think Cantonese has 9 tones? I know there are some disputes as to whether it's only 6 or a 6+3 combination, as some contest the last three tones are not the same as the first 6, linguistically speaking, but I generally go with 9 tones. (graduate student in Chinese history)


5 (?)    I actually don't know how many tones there are in Cantonese. I have attempted to count a couple years back, but some tones to me sound so similar I can't tell if they're different. Yet, I also think they don't sound the same. I don't remember how much I really counted up to, but I think there were at least 5. I'm pretty sure there are more than that though. (college student in the sciences)

It's uncanny that, in general, the older and more academically oriented to Chinese language studies the respondents are, the greater the number of tones that they tend to recognize / claim, and vice versa.

Now, if you're really curious about just how complicated and challenging Cantonese tones can be, I recommend that you take a peek here:

"The Phantom Tone" by I’m Learning Cantonese (download medium on the App Store or Play Store)

Cf. "Why learn Cantonese and one way to do it" (1/20/17)

[Thanks to Don Snow, Judy Weng, Norman Leung, Alan Chin, Timothy Wong, Nelson Ching, Marjorie Chan, Pui Ling Tang, Lily Lee, Howard Y. F. Choy, TinhVan Diep, Justin Wu, Carmen Lee, and Ashley Liu]


*Additional remarks by respondent #7.  He begins by expanding on the question of tones, but then moves on to the matter of literacy.  Although he was fluent in Cantonese from the time he was a child and right through adulthood, due to the fact that there was no suitable way for him to write Cantonese, he was greatly hampered in his progress toward literacy, being forced to learn characters in a Mandarin oriented context.  I list a few relevant Language Log and other posts after the conclusion of his additional remarks.


There is also confusion because Mandarin counts ma, man, mai as different , and in Cantonese depending on the word there's bleed also with mee, mu/mut, etc, the -E and -u sounds that all count as separate sounds entirely — the issue being that Mandarin doesn't have hard endings at all that I can think of.

So what is a tone and what is a separate sound? If I look up jyutping it all makes sense, and I use jyutping all the time now, but I've never studied jyutping formally or any other Cantonese transliteration system, and my (limited) Chinese schooling in Cantonese did not use any romanization whatsoever, it was all memorization of characters and the sounds that go with them.

So like a lot of Cantonese legacy learners my Chinese literacy is very low, I know a few hundred characters and can only now, returning to it as an adult and learning pinyin and some limited mandarin and discovering jyutping, that i can barely read newspaper headlines and menus. From the time I stopped going to Chinese school in 5th grade (and i went every after school, not weekends) until my mid-30s I never wrote a coherent sentence in written Chinese. Now with the help of online dictionaries I can crank out basic notes like "come at 4 pm tomorrow, we will have food and drink" and instructions on how to use a door buzzer and things like that. If i had to I guess I could write a very bad business letter, the reader would know instantly that this guy is an ABC. But she would understand it.

And verbally only in the past couple years have I started to give presentations to Chinese senior citizens in Cantonese and Toishanese . I spoke Cantonese and Toishanese with my family everyday and in public interactions in Chinatown, but never before in an "official" capacity. The experience has been great.

And that as somebody who didn't speak English at all until kindergarten and first grade — I remember not being able to speak English at all in school.

It's very humbling , it's like realizing that your mother language is about as recognized in current American discourse as Catalan or Liberian English, despite our 100 million speakers and preponderance still in the US.


"Is Cantonese a language, or a personification of the devil?" (2/9/14)

"Spoken Hong Kong Cantonese and written Cantonese" (8/29/13)

"English is a Dialect of Germanic; or, The Traitors to Our Common Heritage " (9/4/13)

"Cantonese and Mandarin are two different languages" (9/25/15)

"Token Cantonese" (5/16/15)

"Speak Cantonese" (6/10/16)

"Hong Kong Multilingualism and Polyscriptalism" (7/26/10)

"Identifying written Cantonese" (8/30/08)

"Cantonese novels" (8/20/13)

And there are many others.  Note that important references and comments are often raised in the discussions following the original posts.

See also:

"Written Cantonese" (Wikipedia article)

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

Snow, Donald B. (2004).  Cantonese as Written Language: The Growth of a Written Chinese Vernacular. Hong Kong University Press.

Cheung, Kwan-hin; Bauer, Robert S. (2002). The representation of Cantonese with Chinese characters. Journal of Chinese linguistics. Monograph series, no. 18.


  1. David Moser said,

    February 25, 2017 @ 10:15 am

    Fascinating post, Victor! Parallels my experience exactly. Could the lack of awareness be due to (1) fewer tones are easier to identify and categorize and (2) Cantonese speakers almost never receive formal instruction in the language, whereas Mandarin speakers in PRC and Taiwan etc. usually get some explicit tone instruction when they are taught Pinyin or Zhuyin fuhao? Cantonese speakers have also told me that Guangzhou Cantonese has one more tone than that of Hong Kong Cantonese.

  2. John Barnett said,

    February 25, 2017 @ 10:56 am

    I imagine you'd get a similar range of responses if you asked English speakers how many vowels there are.

  3. Alyssa said,

    February 25, 2017 @ 11:33 am

    I think this is mostly a symptom of the fact that without formal instruction in linguistics, native speakers are rarely introspective about their own language.

    Compare it to vowels in other languages: If you ask English speakers how many vowels there are in English, the vast majority will answer "five", even though that's only true of the written language. And I've been speaking French fairly fluently for years, but I couldn't even guess how many vowels there are in French. But vowels aren't particularly complicated or complex, they're just not something people think about consciously. I'd imagine tones are similar.

  4. Victor Mair said,

    February 25, 2017 @ 11:41 am

    I learned in first grade that English has five vowels: a, e, i, o, u. Don't laugh or sneer.

    Chinese students who go through formal schooling learn that Mandarin has four tones, plus a neutral tone. However, as we've learned in many posts on Language Log, there are numerous variants of the canonical 4 + 1 tones, including "half-third tone".

    Cantonese students don't learn anything about their mother tongue, including how many canonical tones it has, nor do they learn how to write it, whether in Chinese characters or in Romanization.

    This, for a language with 100,000,000 speakers!

  5. January First-of-May said,

    February 25, 2017 @ 1:10 pm

    If you ask English speakers how many vowels there are in English, the vast majority will answer "five", even though that's only true of the written language.

    Meanwhile, Russian really has five vowels (or maybe six, by some analyses, or maybe a bit more depending on how you count the unstressed vowels), but the orthography makes it look like it has ten (or maybe nine, if you forgot the somewhat optional "yo"). I wonder what would Russian speakers actually answer about the number of vowels…

  6. WSM said,

    February 25, 2017 @ 2:39 pm

    kind of interesting description of the 入声 as referring to phonemes with specific endings that are missing in MSM. Pertinent to observe that such 入声 words are voiced very short and low, which explains why many classical lyricists (writer of 詞), for example, chose lyrical patterns that used 入声 rhymes, since doing so lent a choked-off quality to the scansion reminiscent of someone sobbing. This persisted as an aesthetic tradition even at a later date, such as early Qing, when the 入声 had most likely vanished from the actual spoken language (for example the notoriously melancholic early Qing poet Nara Singde favors 入声 patterns, when he isn't simply using his own patterns).

  7. mg said,

    February 25, 2017 @ 6:00 pm

    If you ask an English speaker how many vowel sounds there are, you will likely have them start counting on their fingers while running through them in their minds. Blaming English speakers who aren't linguists for assuming you mean the vernacular understanding of the question (which we were all taught in grade school as meaning "how many letters are vowels?") is unreasonable.

  8. Victor Mair said,

    February 25, 2017 @ 6:41 pm

    Who's blaming English speakers who aren't linguists?

  9. Guy said,

    February 25, 2017 @ 9:18 pm


    I doubt that most English speakers would be able to identify all the English vowels (in their dialect, with reasonable leeway for theoretical issues like whether /ju/ is a vowel or a consonant followed by a vowel) jut by running through them in their minds. How would they know they haven't missed any?

    I think the issue is very much like Cantonese tones. The relative clarity with vowel letters (but why don't we count y?) is very much like the clarity with Mandarin tones. People are consciously aware of facts (and sometimes misinformation) about their own language when they are taught it specifically and formally, but they are usually not at all consciously aware of facts about their language outside of that, even though they "know" these facts in some sense, since they learned to speak the language.

  10. Victor Mair said,

    February 25, 2017 @ 9:28 pm

    Guy understands.

    Key point that some people are missing: Mandarin speakers confidently assert that their language has 4 tones + 1 neutral tone. How do they know that?

  11. Victor Mair said,

    February 25, 2017 @ 9:35 pm

    From a Cantonese radio broadcaster (age-wise she would be in the middle of the 12 individuals listed in the o.p.):

    The only way for me to find out how many tones there are, or rather, are "possible", is to actually sound out some words as an experiment.

    What I found out:
    The most common is 4-5 tones
    Infrequently 6 tones
    Rarely (or never…but never say never…) 7 tones

    I could be wrong because it's possible that I've left out some tones in my samples.

    "ma" (5 tones — the "/" sign separates words that have the same tones), but I don't know the tone marks so they aren't arranged in any particular order. Not an exhaustive list of "ma", just to give you some examples.


    "si": 6 tones


    "mai": 3 tones

    謎/迷, 米/咪 , 瞇

    "yuet": 1 tone


    "suk": 2 tones

    熟/屬/淑/蜀, 叔/肅/宿/縮/粟

    "yim":6 tones

    嚴/鹽/儼/閻, 驗/艷/焰, 掩, 醃/奄/閹/淹/冉, 厭, 染

    VHM: I find this respondent's method for determining the number of tones in Cantonese — sans any standardized instruction from pedagogical authorities — to be both interesting and empirical. But why should Cantonese speakers have to do this for themselves, whereas Mandarin speakers do not have to resort to such a procedure every time somebody asks them how many tones their language has?

  12. Victor Mair said,

    February 25, 2017 @ 9:42 pm

    Additional note by the previous commenter:

    I forgot to add this bit of information. I know that it's possible to have *many* tones in Cantonese, but if you pronounce a word in some tones, they are just gibberish that don't carry any meanings, so I'm not sure if that actually counts as a "tone". So, my answer to your question is, it looks like 6 tones is the "max" and rarely exceeds 6.

  13. x said,

    February 26, 2017 @ 12:18 am

    Are Mandarin speakers not taught about tones when learning Pinyin/Zhuyin?

  14. Guy said,

    February 26, 2017 @ 12:21 am

    Now I'm curious what the maximal size for a minimal set for English vowels might be in some standard varieties of English. I know the context /bVt/ produces a large number of words, but not every vowel is represented (there are no words /bʊt/ or /boɪ̯t/, for example). Is there an even larger set that can be produced?

  15. Jonathan Smith said,

    February 26, 2017 @ 12:28 am

    It was instructive early in my Chinese learning experience to observe one native Mandarin speaker explain to a group of others the nature of "tone" and its status as a special feature of Mandarin relative to English. The "naive" speaker is indeed quite oblivious to it; awareness comes with explicit instruction. So the introspection of Shen Yue, etc., is really most impressive… and leads one to wonder exactly how long prior to the 5th c. tone had actually been phonemic in Chinese.

  16. Jonathan Smith said,

    February 26, 2017 @ 12:35 am

    It occurs to me that learning about tone in the PRC is presumably just part of learning pinyin — not sure why else it would be thought useful.

  17. flow said,

    February 26, 2017 @ 4:23 am

    To those who mention the question of numbering the vowels of English—I was just about to make the exact same comment, "five" (as in, 'aeiou'), only to demonstrate that often what we think we know about language turns out to be a mess once we really sit down and 'do the math'.

    This, of course, is not to disparage speakers of English or Cantonese (or, for that matter, Mandarin); we can't all be linguists. What's more, the answer to deceptively simple questions like "how many vowels/consonants/tones does language x have" is surprisingly difficult.

    The answer "four" or "four+one" that comes so conveniently to anyone with some sort of education in Mandarin *may* be right, but where is the experimental evidence that this is true for sizable portions of native speakers? (There might be a lot or none; I don't know.) And how many vowels does Mandarin have? That may be considerably harder to answer in a meaningful way, and will also decisively depend on how 'vigorously' you apply phonological principles (e.g. do 'gou' and 'guo' have just one vowel each? or are they sequences of the same two vowels in reverse order? are the nuclear vowels in 'lei' and 'lou' the same one (the same phoneme) or two distinct ones?)

    To anyone who is interested in phonetics, do visit Geoff Lindsey's, especially the blog at He does a very good job at picking apart all those things you might have heard about or think to know about vowels in general and vowels of English. For starters, he has a look at the so-called vowel space at, where he says "I’m not sure that languages ever contrast ɨ and ɯ". I feel vindicated by this remark.

    More to the point, there's a nice discussion of the British English vowels at (those charts in your school textbook got it all wrong).

    As a closing remark, how do we know / think we know how many tones are there in Cantonese? And even discounting the entering tones that are so obviously different—are there experimental studies on how many tones can reliably be produced and perceived by native speakers? Do really all (for some sense of 'all') speakers maintain a six-way distinction? Do we know how well those tones persist, in a distinguishable manner, in everyday language (as apart from more declamatory styles such as stage speech, radio announcements and citation forms)?

  18. liuyao said,

    February 26, 2017 @ 8:07 am

    Had YR Chao's Gwoyeu Romatzyh prevailed, Mandarin speakers would also have a hard time counting their tones (assuming they weren't instructed in first grade), and might think third tones as double vowels instead. Since entering tones in Cantonese are already spelled differently, many of the respondents weren't sure whether to include them as tones.

    The four tones (聲) of Shen Yue almost surely included many more tones by today's standard.

  19. John Swindle said,

    February 27, 2017 @ 3:38 am

    Since someone mentioned English … in elementary school in mid-century mid-America we learned that the vowels were a, e, i, o, u, "and sometimes y." The other letters (and sometimes y) were the consonants. We were confusing our language with the elements of its customary writing system, something that has reportedly been done with other languages as well.

  20. Rodger C said,

    February 27, 2017 @ 7:45 am

    I'm reminded of the late Roger Price's skit about his English professor from Mississippi: "Today we'll luhn abaht the vaals: Aey aow yeew!" (I looked for this on Youtube, but "Roger Price" is now some fundamentalist lecturer.)

  21. Andreas Johansson said,

    February 28, 2017 @ 5:26 am

    Offhand, I can think of reasonable arguments for (my variety of) Swedish having between 9 and 23 vowels, depending on whether you think long and short vowels are different phonemes and which if any diphthongs you recognize as phonetic units. The lesson, I guess, is that "how many X's does language Y have" is frequently dependent not so much on the language itself as on your analysis thereof.

  22. FM said,

    February 28, 2017 @ 2:15 pm

    @Guy: biggest I can think of is 16 (assuming GenAm with no cot/caught merger, though I'm not confident about assigning those):
    kid Ked cad cod could cud keyed [ar]cade cowed cawed code cooed cued curd card cord
    Almost a complete set except for *[kaɪd] and some r-ful stuff that only goes in open syllables. Of course some may suggest that "cowered" is also one syllable, as well as the hypothetical *[kaɪɚd].

    Here are some close ones:
    pit pet pat pot put putt Pete pate pout [com]pute pert part port (13)
    kit Ket cat cot cut Kate kite caught coat coot cute curt cart court (13)
    bid bed bad bod bud bead bade bide bowed bawd bode booed bird bard bored (15)

  23. Nicki said,

    March 10, 2017 @ 5:14 pm

    Cured? And possibly Skied if you work really hard… Blue skied horizons?

RSS feed for comments on this post