Toward Better Tones in Natural Speech

At ACTFL 2008 one of the presentations on TCFL that I found most interesting was one called “An Alternative Way to Teach Mandarin Tones in Speaking” by Dr. Rongrong Liao of the Defense Language Institute.

The problem, as Dr. Liao presented it, is that many learners can reach a relatively high level of fluency in Mandarin Chinese, have excellent tonal accuracy for individual words, yet still make a large number of very unnatural tonal errors in natural speech. This is a common enough problem that educators really need to be looking for ways to address it.

The message of the presentation was, in essence:

  1. We’re giving students of Chinese the wrong picture of tones (third tone in particular)
  2. Tones are not of equal importance in natural speech
  3. Funny-sounding speech can be corrected most efficiently by focusing on certain key tones

Now I’ll break these different points down one by one.

We’re giving students the wrong picture of tones

The way students first learn tones is in isolation. You apply tones to individual syllables. The idealized tone contours of those tones in isolation look like the chart below.

Tone Contours in Mandarin Chinese

The thing is, in natural connected speech, tones don’t behave quite that way. Yes, there’s tone sandhi (tones in sequence affect each other in regular ways), but it’s more than just that. Third tone in particular has a habit of dipping but then not rising the way it should. (This phenomenon is known as the “half-third tone.”) So then is the not rising in natural speech the exception, or is the perfect rise in an isolated tone the real exception?

Dr. Liao suggests that it’s more useful to teach that the third tone is low rather than dipping. This could help with third tone problems in connected speech. The “model” third tone with a rising tail could then be treated as the exception to the rule.

The symmetry-loving perfectionist in me actually likes this a lot. This way you end up with two pairs of almost diametrically opposed tones (yes, we’re fudging a bit): high vs. low (1 vs. 3), and rising vs. falling (2 vs. 4). Dr. Liao also notes here that learners tend to confuse tone 1 and 4 with each other much more than with the other two, and tone 2 and 3 much more than with the other two. Very interesting.

This really struck a chord with me, as it matches nicely with my own observations. Taking all this into account and putting the actual tone contours aside for a moment, I put together my own experimental “idealized perceptual tone diagram”:

Perceptual Tone Contours in Mandarin Chinese

I have no idea if a representation like this could actually be useful to any students. Before you freak out by such a concept, though, let’s move on to the next point…

Tones are not of equal importance in connected speech

When Dr. Liao started talking about this, I had an immediate flashback to something my friend Alf said after studying Chinese in China for about half a year:

Tones are such bullshit. When Chinese people talk really fast, they don’t really use them. So I’m just going to ignore them and talk really fast like Chinese people, and I’ll be fine.

Ah, the “tones aren’t important” fallacy. Most students of Chinese have heard such sacrilege more than once in their long years of study, I’m sure. The thing is, like any good lie, there’s actually some truth to it.

Dr. Liao pointed out that in natural speech, some tones in a “frame” are “weakened” or “reduced” and lose many of their “idealized” properties. That is to say, if you look at their tone contours (remember how to do that with Praat?) in the sentence, they don’t all resemble the perfect angles in the classic chart we all know so well.

Here’s an example of what native speaker tone contours look like in speech [source]:

Tones in Connected Speech

You’ll notice that the tones of some words are clearly recognizable, while others are less so. What’s going on? Well, in natural Chinese sentences, certain words in each phrase are stressed. Stressed words will have a tone contour which most closely follows the idealized form, whereas the other tones are shortened, kind of run together, and generally goof off.

Funny-sounding speech can be corrected most efficiently by focusing on certain key tones

Here’s where Alf’s idea comes into play. Dr. Liao recommends that instead of correcting every mispronounced tone in a sentence (and there might be many), instructors should focus on the stressed words. When the tone(s) in a stressed word is mispronounced, the sentence will frequently sound quite bad to native ears, but when the stressed word is pronounced correctly, the other tones will often fall in line.

This is a cool idea, because if it works, it means (1) teachers can stop worrying about so many wrong tones, and (2) students can quit freaking about every tone.

Sounds good to me. It’s complex enough!


John Pasden

John is a Shanghai-based linguist and entrepreneur, founder of AllSet Learning.


  1. That sounds actually really interesting. Any chance for some pointers to more research about that? Could you perhaps upload the accompanying presentation or paper, if you got those? Thanks!

  2. DenisSuslov Says: December 11, 2008 at 1:12 am

    Very interesting, thanks!
    Any example of a sentence with stressed and non-stressed words?

  3. I like that a lot….But is there any rule generally to what words’ tones are stressed in a sentence??

  4. In my view, there is no symmetry between 2 and 4 except on paper. My grouping is as follows: there are three week tones: 1 is high; 3 is low; 2 is rising; and there is one strong tone (like syllabic emphasis in English): 4. Note also that the neutral tone is actually very important, and it’s one of the week tones. 4 is truly in a class by itself.

    The totally agree that the easiest tones to get confused are 1 and 4, and 2 and 3:

    1 and 4 start out about the same (high), but 4 is emphasized and 1 is not. By emphasizing 4, it falls. A weak, lackluster 4 is about the same as a weak, sloppy 1.

    2 and 3 start out about the same (low), but 2 rises and 3 mostly does not. 2 and 3 are the hardest to differentiate because one is not distinctly emphasized relative to the other.

    You can always tell when a native speaker is reciting or quoting poetry: suddenly their tones become very musical, which is contrary to normal speech. Normal Chinese speech relies on intonation (not so much pitch, but intonation), whereas poetry is more musical. IMHO.

  5. […] Toward Better Tones in Natural Speech | Sinosplice: Life A very interesting discussion of Mandarin tones by John Pasden (based on a talk by Rongrong Liao). Of course, you need to know which characters are stressed, which I don’t always know… […]

  6. That makes sense, when I pronounce something well it is usually because I have stitched together some long phrases that I have heard Chinese people speak and mimic the sound they made. My worse pronunciation is when I construct a new sentence entirely from individual words I know (I know the tones of the words but often have trouble running them together).

    The third tone doesn’t quite sit in the low flat low area for me though because the low flat tone in Cantonese sounds so very distinctive if you are used to hearing Mandarin.

    To make a comparison for an English speaker French is slurred in fast natural speech, yet German pronunciation is still clear even when spoken fast (generally agreed upon). To put things in perspective with Mandarin it should be possible to compare with other tonal languages. There may be a tonal language that tends to be a stickler for correct tone as opposed to one that really fudges it, where do Thai and Vietnamese fit in comparison for example?. Comparing tonal languages in this way gives the learner (and teachers) some perspective on how important the individual tones are.

  7. All of the three points make a lot of sense to me too.
    I’ve been learning Chinese in China for the last 8 months and the points made fit like jigsaw pieces.
    I find especially true the second point, living in Chongqing I frequently hear people pronouncing the first word overemphasised, and following it with a cascade of short and toneless bursts.

    I’d be ready to bet that teaching perceptual tones contours and the importance of stressed words would improve the learning process and reduce frustration on the whole.

    Thank you for this post, it’s 极品.

  8. Great post! I learned the tones are not exactly what they seem as a Chinese speaking missionary. One of my companions (who didn’t have the best Chinese) taught me that the length of tones change.

    One thing is true that some people just sound bad when they speak toneless Chinese.

  9. Interesting, I was having problems with 2nd and 3rd tone years ago, and after some intense listening and trial and error in class, I discovered that in the context of a sentence, native speakers would usually tell me i was correct if i pronounced the third tone as though it were essentially just a low first tone. I still tend to dip a bit so I guess there’s some difference, but it’s interesting to learn that someone who actually knows something about linguistics (a doctor!) has confirmed this random thing i made up a couple years ago just for my own use. Good to know I haven’t been wrong all these years.

    I’d be interested to know, too, what (if anything) governs what words are stressed in a sentence…in my experience, it seems to be mostly nouns (i.e. objects that need to be clearly understood to understand the overall meaning of the sentence) and sometimes also conjuctions…

  10. I think the Perceptual Tone Contours diagram is genius, and would be especially helpful for beginners. For me, the 2nd tone was the hardest to learn because I was learning from the chart that says the 2nd tone starts somewhere between the 1st and 3rd tones–I found this very confusing. However, the Perceptual Tone Contours diagram not only simplifies the 3rd tone, it also gives a clearer starting point for the 2nd tone. That is, start with the 3rd tone and work your way up. I’m not an expert linguist, but I have been learning Mandarin for the last four years and I’m willing to bet that this diagram would help beginners of Mandarin learn tones quicker and easier. At the very least, it would have saved me a few hundred hours of trying to figure out the 2nd tone.

  11. Max,

    I don’t have the paper; I’m relying on my own notes (so I hope I represented her points accurately!). I was really hoping to contact Dr. Liao herself, and perhaps link to the paper or get some comments from her personally, but I can’t seem to locate her e-mail address. Many of the presenters at ACTFL give out their e-mail addresses, but she did not, and I didn’t get it from her.

    If anyone has her e-mail address I’d appreciate it if you could e-mail it to me. Thanks!

  12. Denis and Colin,

    Stress is really tricky, and it’s not my field of expertise. I’ll let you know if I find anything useful about it.

  13. Interesting enough for academic purposes. However, how do you deal w/the learning locations and their influence: such as tones in Taipei vs tones in Beijing? At that point, whose to say what the correct tone is? I will say that although there are differences here in Taiwan vs China, my family and teacher readily point them out when speaking to me.

  14. I think this is something that a lot of people internalize after spending a fair amount of time in an immersion environment, but it’s nice to see somebody talking about incorporating it into a classroom program too.
    I’ve been telling people for years that while they should worry about the tones to some extent, they shouldn’t worry too much since they’ll just have to teach themselves the tones all over again in a few years anyway. Would be nice to see a program that cut out the intermediary step.

  15. Lin,

    I’d say only beginners with a particular destination in mind need to worry about that, and that it will be addressed naturally as one’s studies progress. I believe that learners should definitely be exposed to a wide variety of accents.

  16. Brendan,

    Exactly! So the question is what people can actually internalize, what needs to be taught explicitly, and at what stage for more efficient acquisition.

    Some learners catch on (internalize) much better than others, but right now I think the ones that don’t are mainly just getting left in the dust.

  17. There is actually an interesting other consequence of this:
    It makes pronounciation less deterministic – and thereby less machine processable.

    I still remember testing out that Chinese speech-recognition / pronounciation training software. When my Chinese co-worker read the given texts naturally he was supposedly making tons of mistakes.. Only after he robotically pronounced syllable by syllable the program was satisfied.

    Giving the above findings, realizing a pronoundication training software gets much more complex. And it it needs additional informaiton on emphasis that it will never be able to deduct by itsself as software does not grasp semantics and context. That also means you cannot feed it with an arbitrary, unprepared text.

  18. […] Click here to read the article: Toward Better Tones in Natural Speech […]

  19. The “idealized perceptual tone diagram” looks good. I like the color-coding. Is it new, though? Something like it has been around in my head for years, since very early in my Chinese study. Unfortunately I don’t remember where it came from. Either it was taught me at some point, or your diagram is an eerily accurate version of my perception of tones. (It would be even more so if you stretched the oval shape for 1 and scrunched the oval shape for 3, to represent their respective bell-like and creaky natures.)

    As for when Dr. Liao’s points 2 and 3 should be introduced, perhaps at the intermediate level? From my experience as a student, rank beginners ought to be strictly forced to produce tones for every syllable, since it’s way too tempting for them to pronounce all tones weakly. (Tones at first seem so embarrassing and overwrought….) And beginners can’t say that many words in a row anyway. And they do need to learn the right tone contours. But I remember crossing a threshhold, around my second semester of college Chinese, to where tones actually started to make sense; that is, when I stopped having to pretend I was asking a question in order to say 2, pretend I was being emphatic when I said 4, etc. When students have reached that point, it would be safe to switch to the approach of focusing on stressed tones.

    In teaching English pronunciation, with our frightening array of vowels, you not only have to get students to clearly pronounce the stressed vowels in a sentence, you also have to get them to weaken unstressed vowels. It makes sense that a similar approach could work with Chinese.

    Finally, one random note: my teachers definitely taught the shortened third tone right from the beginning. Is that not then standard practice?

  20. Glenderful Says: December 11, 2008 at 9:37 pm

    Here’s my issue: why can I never, ever get the 4th tone, right? No matter how much I prepare for it, it just doesn’t work for me! Does anyone else have this situation? There’s times when I feel like I’m yelling and trying to sound really angry to get the 4th tone, but I know it’s not right. Any suggestions?!?!

  21. Yeah I tend to agree with Alf, but with an add-on. Since back in my Switzerland days I have always found that trying to emulate the pronunciation of the language being learned as exactly as possible includes not just pronunciation but stress, rhythm and tone, inflection and emotion. Talking fast has always been my problem – I speak too fast in my own mother tongue for other native speakers to understand at times. So by doing this in Swiss-German I would speak so fast that even my mistakes would sound as if I was speaking with a few different regional dialectic influences – the reaction would be one of amazement that I was such a well traveled Swiss person (I am an Aussie). So the same thig works for me in Chinese – I speak so fast with fairly accurate pronunciation that when I do make pronunciation errors they are less easy to pick up on, also speaking with the odd word in Ānhuī (Língbì) tǔhuá and some Suzhounese and even the occasional Taiwanese HaoHao Ci Eh! thrown in they just think I’m the bomb (until they realize after a few minutes of conversation that my actual articulation abilities are in need of some improvement).

    And thus talking fast + the occasional phrase in dialect will also help cover up those tone errors.

    NOTE: At U.Q we were actually taught to make 3rd tone marks as an underscore underneath the vowels in the 3rd tone word.

  22. I believe my best friend from school and former roommate was taught by Dr. Liao in Monterrey. What a small world!

    I’ve always been of two minds about tones. On one hand, I want to break them down to the point at which I could use mathematical functions to describe them, regardless of how complex those functions would have to be. On the other hand, when it comes to actually improving my speech, mimicry has always served me best.

    By mimicry, I mean actually trying to deliver the lines I hear from a character on TV as if I were a method actor. Doing impressions of good friends seems to have helped a lot, too. I suppose this particular method isn’t open to every personality type, but it’s a lot more straight-forward than actually trying to memorize all the rules.

  23. geoffrey cook Says: December 12, 2008 at 4:37 am
  24. John, this is a very useful post. I just forwarded it on to one of my old Chinese professor from college for her to use as a teaching aid. I know characters can be difficult to learn, but I think one of the hardest things for a foreigner to accomplish is perfect Chinese tones. I have found that most of the time poor pronunciation is a result of bad habits developed early on in their Chinese studies. If more Chinese instructors were aware of what you mentioned in this post, then it could help them prevent their students from developing these bad habits in the first place. Thanks for this.

  25. After reading the comments it would seem that speaking a foreign language well is so complicated that it is best reduced to a biofeedback type approach (in which you learn to hear what sounds right and then keep adjusting until you sound the same, mostly ignoring technical analysis or complicated explanations).

    I taught myself Tuvan harmonic throat singing that way (you adjust the muscles in your mouth, throat and tongue so that the cavity formed is exactly correct to pick out various harmonics of a constant hummed note). The actual procedure is impossible to explain fully, the best way to learn is to listen, start with a few basic pointers and then keep going at it . You never know the exact combinations that work, but your body remembers for you as you increasingly get it right.

    I know that biofeedback is usually used in medical contexts, many years ago I learned to drop my heart-rate from about 80 to 40 in the space of a few beats, learning to do that though is similar to learning Tuvan throat singing and many aspects of language learning.

  26. I’m interested in the color coding used. I had heard it suggested that color-coding characters based on tone was a good way to help learn them (you’d put 1st-tone characters on yellow flashcards, 2nd-tone on blue, etc.). I’m thinking of trying out something similar in Skritter, where after you practiced a tone, the character would perhaps flash with the tone’s color.

    Are there standard colors that are typically used, or could I pick them arbitrarily?

    (Incidentally, I’m also interested in color associations with traditional vs. simplified, but I doubt there’s much there.)

    I only caught the end of Dr. Liao’s talk; did she suggest that the unstressed tones could goof off however they wanted, or that they would still have some restrictions, but ones that made sense, given the context, to a native speaker? I’m curious as to whether a student could make stuff up for unstressed syllables and not have it sound as valid as more naturally made up stuff.

  27. This totally makes sense to me, and I’ve been considering this method as I converse with the citizens of the middle kingdom and find that they do seem to understand me a bit better. What is more I feel like I’m more sense to myself, as far as tones are concerned.

    As to how tones ACTUALLY sound. I’ve found this; when two 4th are together. Make the first one sound like a first tone, and the second a strong 4th tone and the Chinese will completely get it!

  28. oh, this is one of the best posts in a long time, John. As every Chinese learner i have been so frustrated by the tones and this feeling doesn’t seem to go away completely, even after three years in the country. For a long time i have been trying to find a way to make myself use and recognize the tones correctly. I thought it was me, but this article finally agrees with what i have been thinking for a long time. The tones are no way absolute. They depend on the word itself, the place in the sentence, emphasize, combination with other tones, etc..

    The thing with tones is, that chinese people are best placed to explain but that like for any native speaker, it is more difficult to explain something obvious about your own language, than about an other. Not chinese learners on the other hand have many theories and methods but no way to know if their perception of the tones is accurate. Unless you have a built in ‘praat’ in your mind. For me anyway, I really don’t hear a lot of ‘tone’ differences (in a do re mi kind of way), I suck at this in music as well. But I do hear a difference in the way words with different tones are pronounced. I just can’t define it. So many times I have cursed my friends when they correct a word I said, while I was convinced I said the correct tone. After they correct me , it still often takes 4 or 5 tries before I get it right. This shouldn’t happen because after three years I can pronounce single words with whatever tone, or so i’d like to think.

    Sentences on the other hand are quite a different matter… Just a week ago my teacher was explaining about the intonation of some sentence he just said. The student was supposed to say this or that part with a higher pitch and the last part starting high and ending low. Now both parts were mostly first and neutral tones, so how can you continue a high to low sound while producing first tones, following the tone chart? Even if you can control your voice enough to say the first first tone higher than the second one and thus having an average drop in pitch, you don’t follow the tone chart. I haven’t even considered this with second tones. I would like to get further into this because if i don’t stop intonating and emphasizing my sentences like I would in my own language i will never sound even close to natural. It is very hard to find information on this though and on how it affects the tones. It could be helpfull to know about this more because it could explain why we have such a hard time recognizing some tone, if maybe the reason is intonation or stress. Like Michael Max said, if you have two fourth tones following each other, the second is usually stressed and since many foreigners seem to associate the fourth tone with stress, like some of the commenters here, the first fourth tone doesn’t appear like a real fourth tone to our ears. As is a fourth tone that is pronounced by one of my friends who always seems to pronounce it very slow and long, making it actually harder for me to recognize it.

    About the perceptual tone contour diagram. In my subjective perception the fourth tone would be drawn shorter with most of it’s length on a very high level. Or at least I would draw the beginning of the arrow thicker and the end very thin. For me the only tone that really changes a lot in pitch is the second one. And the third one in certain circumstances, but usually not in combination with other tones in one word.

    Although this article is very interesting, I doubt if we will ever find an optimum way to explain students about tones and to teach ourselves. Diagrams and whatever ways to visualize the tones are great and all, but shouldn’t we regard ā á ǎ and à as different vowels instead of starting from the ‘a’ sound of or own language, making it sound chinese and adding a tone to it? This doesn’t solve the problem of making the actual sound but would take away a lot of the confusion about chinese pronounciation in my opinion, especially for beginners. “It’s the same word, but with a different tone.” is a statement that is just false. If it has a different tone it is not the same word. ‘Car’ is also not the same word as ‘bar’, with differently pronounced consonant. ‘Hit’ is not the same as ‘heat’ (although it is a variation of the same sound), and so on…

    Well, whatever we say about it and what is true or at least helpful and what is not, at least now i can relax some with the thought that tones are relative and not of the utmost importance for every syllable.

    And now i am going to experiment with Praat. 🙂


  29. Nick,

    I have no idea if there are any color-coding conventions. In the absence of any better reasons, I chose warm colors for tones that start high (higher pitch, higher temperature), and cool colors for the ones that start low (lower pitch, lower temperature). To me, tone 1 and 4 also feel higher energy, which goes well with warm colors as well.

    I don’t think that Dr. Liao suggested that the unstressed tones could goof off however they wanted, but the exact principles by which they were “weakened” or “reduced” were out of the scope of her talk.

  30. check out beijing yuyan daxue, hanyu yuyin jiaocheng (北京语言大学 汉语语音教程), published in 2002. also Chen, Tone Sandhi Patterns Across Chinese Dialects.

    i get so annoyed when students have only learned stressed tone 3.

  31. […] Sinosplice on a better way to teach tones in Mandarin. […]

  32. Great post. Not sure I get the third tone as just a low tone, as the first is a high steady tone. There has got to be a dip (according to my humble ear), albeit a truncated dip much of the time. Try saying 美国 mimicking the 3rd tone as a low steady tone… it won’t sound right. I see the tones more as a framework, not as the gospel, and the rest is simply listening to native speakers and doing as they do. This, of course, will vary from region to region and person to person, yet if you sound like ANYONE from China or Taiwan you are already a great success. And, Chinese, as I’m sure you are aware, isn’t the only language difficult to speak accurately just because of her tones. Try explaining the French “r” or Spanish “rr” to learners. Let’s take the Spanish “rr”, infamously difficult for Americans. “Just roll your r’s”, Spanish teachers will explain. Here is what is not explained: sometimes your roll is longer for emphasis, other times it is truncated. So what happens is non-native speakers sound kind of funny rolling too long, stopping too short. The only thing to do is perk up your ears and go with it. There were a few posts to that effect. French perhaps is more comparable to Chinese in our capacity to absolutely slaughter the sounds. I hear Americans reach very advanced levels in both languages and be far off the mark with the pronunciation/tones, intonation. On the other hand, I also see the rare beginner come closer to nailing it. My recommendation for anyone learning any language with challenging sounds is to get addicted to podcasts (,, etc.). And do not listen to glossy, dramatized Chinese such as the audio for Integrated Chinese; it should be natural sounding language. Another thing, which I did at one point, is get a private tutor and read from simple texts at first (beginner textbooks), have the teacher stop you every time you sound a bit funny, highlight those words/passages and keep trying to copy the teacher until you get it right.

  33. More pitfalls of letting linguists teach language. Always explaining things in a way that makes sense to linguists, but is baffling or nonsensical to everyone else.

  34. I usually ready your blog on the way to work, so I missed out on this one and all the pictures. I can totally relate this entire concept the learning scales in music theory..

    To summarise: it all depends on where you start. Looking at the speech pattern diagram this is extremely obvious. If you are starting on a high 1st tone, then the next syllable is going to start up there and end in a shallower position, relative to the starting position. Whereas, if you start on a third tone then you are going to be starting the next tone, regardless of what it is, at a much lower relative pitch. Even in the case where you use another 1st tone, it’s definitely going to be higher than the highest point of the 3rd tone that went before it, although not that far up in pitch.

    As you say, if you individualise each syllable separate to the next, then you will end up with a set of perfect “example” words but not a flowing sentence. This is very similar to the way music flows from phrase to phrase.

  35. As others have expressed here in various ways, I also see this kind of discussion of second language acquisition as describing two different aspects of learning– the conscious and intentional type of learning that happens when we are consciously trying to learn something, and the subconscious (internalizing) kind of learning that happens over time with sufficient exposure and interaction in real-life (non-academic) situations using that language.

    Memorizing tone patterns in new vocabulary is tedious, but perhaps is still a necessity at a certain stages of learning Chinese.

    But very early on (in learning any language) I think it is a good practice (assuming one has a native-speaking teacher) to begin learning whole phrases, sentences and even dialogs by heart in large chunks, and without analyzing grammar too much. For example, it is enough for the student to know that this phrase means “Are you single?” or “Please bring us two more beers”, etc. In that way, you are swallowing and regurgitating whole ideas, and hopefully being coached to get the stress and ‘meoldy’ as close as possible, basically mimicking your teacher.

    And this is somewhat related to the “tones don’t matter so much in rapid speech” idea. A beginner can learn some vocabulary words and can try to string a sentence together, but if it doesn’t come out pretty close to the way a native expects to hear that idea expressed in normal speech, then it won’t be understood –no matter how accurate the learner may be with the tones on the individual words.

    But if the (Chinese) listener hears a whole sentence (in gramatically-colloquial usage), then it will be understood pretty much regardless of tone-mistakes. That’s because the sentence will be understood as a sinlge, intact IDEA. The individual syllables that make up that idea are understood IN CONTEXT, as parts of an larger pattern, rather than having to to be guessed at one by one….with inaccurate tones to confuse the listener.

  36. […] On the Jump to Comments Hard-core students of Mandarin Chinese will be interested in this article on the subject of teaching and learning the pronunciation of the four standard tones of Mandarin, posted on the excellent Chinese learning site Sinosplice. […]

  37. I’ve often noticed that if there are a heap of 3rd tones together it really ruins the flow of the sentence.. I can’t think of the last sentence I saw like this, other than it had 4 3rd tone syllables all together at the start of the sentence. I found this hard to pronounce in a fluid way.

    I find that when a chunk or whole sentence flows easily in a fluid way, it makes it exceedingly easier to remember. Take a phrase such as “what is your name?” “ni jiao shenme mingzi?”, this sentence flows remarkably well and makes it not only easy to remember but also allows the beginner to get a taste of how whole sentences should sound when spoken without hesitation.

    Getting back to the similarities of music theory and language learning, this can all be said to be like learning scales. Many beginners in music hate going over and over their scales. However by doing so, it means that when they use them in their music later on, they will do so effortlessly.

    When you learn the pronunciation of tones, initials and finals in mandarin, it provides the same foundation that scales do in music theory. Grammar inadvertantly arises early on as well but one without the other is useless.

    In English I would know thousands of words by now but ask me to name them all and I cant . And even the ones I don’t know can be easily worked out because of a solid foundation.

  38. […] Toward Better Tones in Natural Speech – Dr. Liao (of the Defense Language Institute) suggests re-conceptualizing tones in Mandarin to take into account natural speech patterns and tones in sentences. […]

  39. I have been using this system for a long time now, because I noticed the important words or words that easily can be misunderstood were being stressed and a lot of the other speech was fast and not distinctive when a native speaker spoke. Like the word Shui(3) I would usually heard stressed with the third tone. Very interesting post John.

  40. Is it possible to get a list of easily misunderstood words that need to be stressed? It would be interesting to see this “stressed words” methodology incorporated directly into a learning course. Not just discussed in a sidebar or introductory paragraph but really incorporated into the overall learning structure.

  41. This gives me a bit more confidence in speaking if i think of tones as such.

  42. Glad to see this being noticed in the real world. We did some phonetic experiments a few years ago that showed this kind of stuff. It seems that (on average) alternating syllables are stronger and weaker. The first syllable is generally strong and you do the tone correctly, then the second is weak and instead of doing the tone, you use that syllable to get ready for the tone on the third syllable.

    If you want the technical details (with math), the paper is at

  43. […] into the nuances of Chinese life. Its author John Pasden, of Chinesepod fame, recently wrote this great entry concerning the learning and use of tones in Chinese. Perhaps it helps that he has a degree in […]

  44. Interesting. The small danger I see is that people like to swing to the extremes: “tones are everything”, or “tones are not important at all.” Both are ludicrous, the latter probably more so.

    It’s true that tonal production in Mandarin varies widely native Chinese speakers, particularly those with strong regional accents. Why, then, does a native speaker recognize this as ‘native’ production, whereas the foreigner stands out so much?

    I agree with Dr. Liao to an extent, but I think an even bigger part of it is that these people use a syntax and structure, and a rhythm to their speaking that a native speaker recognizes as genuine, even if rustic or regional. This is what a true non-native speaker lacks, and it’s the dead giveaway even for the foreigner with quite standard tonal production, because it’s the most difficult aspect to master.

    One last point, I’d be interested if Dr. Liao has been able to put her theories in practice at DLI. I’ve actually met several people who learned Chinese there, and I would have to say their speaking abilities generally are rather poor. DLI has always stressed listening comprehension, since mainly what the students do once in their military posts do is analyze voice transmissions, so they can understand quite a bit.

  45. […] while ago, John Pasden posted on his Sinosplice blog an article about an alternative representation of tones for chinese in connected speech. I’ve been […]

  46. […] my own perceptions of those tones. I had even (briefly) considered color when I sketched my “Perceptual Tone Contours” […]

  47. […] something similar to a second tone plus a third tone, but I think most people get that. Please read this post for a suggested different way of picturing the third tone. Although perhaps not more accurate, it […]

  48. This article really helped me a lot actually. I am a foreign American student in China and I’ve found that when I pronounce Chinese with the correct tones some Chinese don’t even understand me. But if I blur words together, and don’t pronounce things thoroughly, like with the 3rd tone I noticed that people speak low mostly, I am understood. It’s really baffling sometimes. As for stress words, it’s a bit difficult understanding that concept since I myself can understand more than speak.

  49. I have also noticed that Chinese people don’t use the tones all the time. As light487 suggested, it would be very helpful to have a list of easily misunderstood words that need to be stressed.

  50. I would say that half of the problem is that most orthodox Putonghua training is teaching Beijing pronunciation which in my opinion is sort of like teaching English with a Scottish accent. Only some people talk like that, it sounds horrible, and the mainstream putonghua is much softer and easier to pronounce.

  51. Thanks John.

    Hope you do not mind that I am going to cite your tones graph in my phd application proposal on perception of Mandarin Chinese tones in Chinese adoptees in the Netherlands.

    But I will mention the source 😉

    Great job


    • Selina,

      I would love to hear more about your research! I’m happy to her that Sinosplice is useful to you, and please let me know what happens.

    • Selina –
      Your research topic interested me and I was able to find your paper online to read more. I am not a linguist but was drawn to your idea since our daughter whom we adopted at 15 months from China is now 8 years old and has been learning Chinese for about 5 years. We assumed she had next to no language remaining from her early time in the SWI but your research indicates otherwise. She is reasonably fluent now and Chinese friends (both in the US and China) tell us her pronunciation, including tones, is like a native. I wish I could develop such facility with tones!
      Thank you and best wishes on your further work.

  52. Matthew Patenaude Says: February 25, 2010 at 9:51 pm

    A note on the fourth tone. I think foreign Chinese students, if they are native English (or perhaps German as well?) speakers tend to put a forceful stress on to the fourth tone syllables. Then, if they think there should be stress on other tones, then these become pretty much identical to their fourth tone stressed syllables as well.

    Chinese speakers do not as a rule put that kind of stress on any syllables as a function of tone. Tone is more pitch than anything else. The only time you will hear the kind of stress that English speakers use – a forceful volume of air coming out, as in the stress on English syllables – is in emphasis, particularly emotional.

    I agree that the perceptual tone model seems to fit the actual native Chinese speech patterns fairly well.

    • Matthew
  53. Dear John!
    I liked this article the first time I read it more than a year ago.
    At that time I was a Chinese newbie, and it really helped me with my tones.
    I am very grateful for that.
    Recently I have made a Russian translation.
    Hope you will not object.

  54. An important observation that should be standard teaching material. I had to figure out for myself that there are “full” and “weakened” tones in natural speech. Although I have essentially given up trying to achieve “conscious competence” with these more complex and contextual patters. Rather I think I’ll do better to let this develop as a sub-conscious learning process by listening to and emulating full sentences and phrases spoken by native speakers as often as possible. What sucks is, this internalization of proper speech patterns would be a much faster process in an immersed environment. I think those of us struggling to achieve some sort of fluent, nevermind “native like”, speech outside of China are in for a long slog. I intuitively suspect though, that listening to dialogs well above our level is a useful excersise. Maybe we could listen to recordings of a native reading Harry Potter (or whatever) in our sleep to help speed up the process.

  55. For some reason this reminded me of that one strange discovery of rdeiang wdros mxeid up. English speakers can read and understand it because we know the words by the context of what the words should be in the sentence and by being given the correct letters, only in the wrong places. For Chinese, from what it sounds like (from what I read), if you stress the correct tone in a sentence in a correct way, given the context of the entire sentence and the topic at hand, to a native speaker the rest of the tonal meaning conveyed (unless it’s completely butchered perhaps) will fall into place.

    Ah, tones.

  56. […] read this post for a suggested different way of picturing the third tone. Although perhaps not more accurate, it […]

  57. I found this article (and the underlying research) so interesting that I decided to write my bachelor thesis in Chinese about teaching the third tone. My conclusion is similar to Dr. Liao’s, i.e. that there is call for a different approach to the third tone, regarding it as an essentially low tone instead of a falling-rising tone. In case anyone is interested, the article and the thesis can be found here:

  58. […] the 3rd tone should just be called the “low tone”. Here’s a short article about John Pasden’s better tone diagram. Olle Linge recently came to the same conclusion in his thesis for Lund […]

  59. […] provided at the end. Before you check this out, though, you might be interested in having a look at an article over at Sinosplice which deals with the third tone. Now the abstract: The goal of this paper is to examine various […]

  60. […] do the same thing. It means that it’s even more important to get those key tones right. See this article on for more about […]

  61. […] and how it relates to the teaching of Chinese, but some places to start are the discussions on Sinosplice and Hacking […]

  62. I know this is an old post, but I had to comment on this. I’m having a 现代汉语 classa at the moment and actually in our textbook (written for Chinese students) there is a rule:

    3rd tone + not 3rd tone = half 3rd tone + not 3rd tone(上声+非上声=半上+非上声).

    When a 3rd tone is 214, the half 3rd tone is 211(调值).

    My Chinese teacher in Finland (from Tianjin)never thaught me this, but apparently this is commen knowledge in China among students that study how to teach foreigners Chinese. Actually 3rd tone is rarely pronounced as 3rd tone, because it often changes to 2nd tone or half 3rd tone.

  63. […] Toward better tones in speech – by Sinosplice This article is about how tones are really used by native speakers. Native speakers swallow the tones on many words, depending on whether the word is emphasised. Besides, the 3rd tone is not what it’s often made out to be: there are variations. […]

  64. […] often does not actually always sound like a clear down-up when it is spoken in normal conversation. This article suggests an excellent alternative way of thinking about it, where the 3rd tone is viewed as […]

  65. 哲明 - James Says: March 23, 2013 at 11:05 pm

    Very enlightening. Thanks!

  66. […] chart once you’re in the real world. (For a great discussion of this, read this great article Toward Better Tones on […]

  67. I recently created a free web resource for learning Chinese tones based on differentiating minimal pairs. The site has a “boot camp” section for new learners and this article is one of the resources I link back to from there (hope that’s ok!). Some people might think of this as an advanced topic but it is very confusing and discouraging when you are told things are supposed to sound one way and it isn’t what you hear, so I think the earlier learners can be made aware of this kind of thing the better.

    Anyone wanting to check out their tone recognition skills can find the website at

  68. […] well. So, if the resources you choose to use simply say it “falls then rises,” you should read this article on Sinosplice and this one on Hacking […]

  69. learn tones just like learn vocal music

Leave a Reply