Rule for sequences of 3rd tone syllables?

[mirgcire]

I have been explaining Mandarin to beginners for a few years now and recently I discovered a rule for tone change that applies to sequences of 3rd-tone syllables.  The purpose of the rule is to make it easier for mandarin learners to produce sentences "correctly".  The rule does not predict how every Mandarin speaker might handle tone change for sequences of 3rd tones, because many variations are allowable.  This rule simply defines the easiest method of applying proper tone change to a sequence of 3rd tone syllables.

 

For any sequence of third tone syllables, all but the last will change to 2nd-tone, except when the sequence includes a double-third tone word (such as 可以).  The second syllable of a double-third will be pronounced third tone no matter where it appears in the sequence.

 

For example, 我可以买你的手机吗？can be pronounced 223235315.

And,  我也很好 can be pronounced 2223.

I use the word can in italics because these are not the only ways to pronounce these sequences.

 

I am interested in hearing what others think of this "rule".  Is it correct?  If not, do you have a counter example?  Is it useful?  Is un-helpful?  BTW, I wouldn't be surprised if this has already been proposed by others.

 

Thanks!

 

 

 

 

[stapler]

To be honest I think this "rule" is a little misleading. It is misleading because third tones don't actually change into second tones when they are chained together. Rather the third tone's curve becomes more like a second tone's curve when linked up (but it's still a third tone). Native speakers have no trouble telling you when you that a two syllable word made up of two third tones sounds different from a two syllable word with a second tone followed by a third. You can confirm this by running words through something like Praat.

 

To take one of your examples:  ---  我也很好 can be pronounced 2223.

 

I think it would be a bit strange to suggest that this is to be read as "2223". It should still be read as "3333" but with the caveat that the third tone changes a bit when linked with other third tones.

 

I am no teacher, but if I had to teach someone this sentence in Chinese I would tell them everything is a third tone and then as they get use to hearing more authentic Chinese, they should slowly and automatically start making the adjustments to the third tone. To tell them there are a bunch of seconds tones, I can only see a beginner slowly saying "wo2 ye2 hen2 hao3" one syllable at a time. It'd sound quite weird. The chain of third tones should 'automatically' change once the student starts learning to speak a bit faster.

 

/end 2 cents

[Demonic_Duck]

I would pronounce “我也很好” as something like ³ 3 ³ 3, that seems most natural to me (^{3 3 3} 3 sounds slightly weird). I don't think saying it slowly makes much difference. It's interesting to me that you (@stapler) say the third tone with tone sandhi applied isn't exactly the same as a second tone - this is something I've begun to suspect for a little while now, but the accepted wisdom (at least in Chinese teaching and learning circles) still seems to be that they're identical.

 

In OP's example (“我可以买你的手机吗？”), if I say the sentence slowly, I think it's something like 3 ³³ 3 ³ 5 31 5. Seems like I'm applying tone sandhi across the neutral “的” to “你”, due to the presence of “手”. I'm not a native speaker, though.

 

Edit: to be clear, I'm still on the fence about whether or not ³ is identical to 2. Does anyone have any links to research on this?

[gaogaozhan]

 

我也很好 can be pronounced 2223.

 

No, it should be 2 3 2 3.

[Shelley]

It was explained to me as not where you go, down low and then up as in a third tone, but where you start, slightly higher up and without a pronounced down then up.

 

The third tone ends just above where the 2nd tone starts, so you go down a bit, but not so far down as you would with a 3rd tone. and then up to where the 2nd tone ends which is slightly above where 3rd tone ends.

 

This not easy to describe only being able to type so I have attached a picture of the tone chart I use and I hope you can see what I mean.

 

If you have a sentence that is 3333, then I would think it should be 3232, because once the second 3rd changes to 2nd the third 3rd doesn't follow a 3rd it follows a 2nd but the last 3rd follows a 3rd and should be 2nd.

 

I think the basic idea is to avoid an up and down sing songy result if you used 3rd tones all the way through.

 

I think its all a case of relativity,

 

I think this is one of the most difficult part about tones to explain, learn and understand.

 

I may have got this all wrong, but this is how it was explained to me.

[Demonic_Duck]

@Shelley: it sounds like you're confused on at least one aspect - when you have two third tones in a row, the first of the two should change pitch contour, not the second. Hence, 我也很好 becomes [³ 3] [³ 3], but definitely not [3 ³] [3 ³].

 

Also, the tone chart you attached is the classic one used in teaching, but it is misleading in that it fails to show that the vast majority of third tones (with no tone sandhi) are actually realised as low tones which do not go up at the end. The tone contour as shown on the chart is only used for third tones in isolation, and sometimes at the end of utterances.

[Shelley]

@Demonic_Duck thanks for clearing that up for me. As I have mentioned before I am starting again from lesson 1 as it were for this very reason either through bad information or my lack of understanding I have got some of these things wrong.

 

Started this week with a new course with edx, Learn basic Mandarin. hopefully it will clear up this sort of thing, I am sorry if i have confused the issue but at least one person (me) has learnt something.

[anonymoose]

Maybe not 100% accurate, but I think it would be much simpler to just regard the 3rd tone as a low pitch (flat) tone. That may not be exactly how a native speaker would pronounce it (but I suspect more or less indestinguishably close), but then native speakers don't all have exactly the same pronunciation anyway. Tone charts are all good and well, and of course neccessary for pedagogical reasons, but I think they mask the variations between real speakers, and the contour for the 3rd tone is just downright misleading.

[mirgcire]

Thanks for all the thoughtful comments. I definitely agree with Stapler, Demonic Duck and Ananymoose that the standard "tone" chart is somewhat misleading, and that the contour of 3rd is fluid at best.  But, I have never seen any book or web site that has stated any rules about the correctness of pronouncing strings of 3rds, so I am assuming that there is no correct way.

 

My goal is to identify pronunciation hints that are correct enough, and simple enough to get people speaking with confidence.  All rules have exceptions, but having rules helps us get started.  If we use the language, we eventually learn the exceptions without even knowing it.

 

Here are some sentences to consider.

你好吗？

我很好，你呢？

我也很好。

 

I think it is pronounced

Níhǎo ma?

Wó hěn háo, nǐ ne?

Wó yě hén hǎo.

 

... or would it be Wó hén hǎo, nǐ ne?

 

I think the middle line is speakers choice.

[陳德聰]

I don't think you should be explaining things you don't understand to beginners just yet.

[dwq]

You should consider the two halves of 我很好，你呢？ separately, as the comma stops the flow of pronunciation, so to speak, and tone change does not happen across a comma.

i.e. The first half would NOT be pronounced 232.

[dwq]

Here is what the book 普通话水平测试实施纲要 (Syllabus for the PSC exam) says about tone change.

It has rules for 3 character words, but unfortunately no rules for multiple words chained together.

 

[dwq]

One interesting bit from the above is that it asserts 「实验语音学从语图和听辨实验证明，前字上声、后字上声构成的组合与前字阳平、后字上声构成的组合在声调模式上是相同的。说明两个上声相连，前字上声的调值变得跟阳平的调值一样。」

 

i.e. a tone-changed 3-3 sounds IDENTICAL to 2-3 from experiments.

 

If you are wondering what the PSC exam is, it is a national exam recognized by the Chinese government mainly for natives (whose mother dialect might not be Putonghua).  Here is a thread about it.

[mirgcire]

@dwq - very useful input.  Thanks!

[Altair]

One interesting bit from the above is that it asserts 「实验语音学从语图和听辨实验证明，前字上声、后字上声构成的组合与前字阳平、后字上声构成的组合在声调模式上是相同的。说明两个上声相连，前字上声的调 值变得跟阳平的调值一样。」

i.e. a tone-changed 3-3 sounds IDENTICAL to 2-3 from experiments.

Most of the literature seems to agree with this point; however, there is some disagreement. There is also a possibility that there is variance according to region (e.g., Beijing vs Taiwan). If there is a difference, it is slight and probably results from some speakers (unconsciously modifying Tone 3 into its sandhi version, rather than merely replacing it with Tone 2).

[mirgcire]

@Altair, thanks for the reference to a study.  In English no less It has a lot of interesting observations.  I am going to give it a thorough read to see what I can glean.  I noticed it was published in 2003.  Gives me hope that other studies have been done.

[Altair]

Since no one else has come forward, I will give may best understanding, based on the papers I have read.

The citation tone of a particular syllable can be affected by syllabic accents in a multi syllabic word, by sentence intonation, and by the tone of surrounding syllables. Accenting a syllable tends to raise the upper limit of pitch contrasts. Removing the accent tends to depress any high pitches.

The requirements of sentence intonation also adds its effect to each tone, without obliterating the contrasts. For instance, pitches at the beginning of an utterance, before minor pauses, and at the end of an utterance may be altered somewhat to reflect the sentence intonation.

Speakers also tend to adapt change in pitches to minimize vocal effort while still maintaining sufficient contrasts to be understood. For at least some speakers, changing a tone-3 syllable to a so-called tone-2 syllable is not a substitution, but merely a vocal adaption that deletes the initial downward glide (perhaps subconsciously attributing it to the drop in pitch during or after the preceding syllable. The half tone-3 is similarly just a deletion of the terminal upward glide and attributing it to the rise to the start of the following syllable.

All of the above effects mean that any two given tone-3 syllables will not necessarily have either the same pitch or even the same contour as each other or as any of the canonical pitch contours. When someone says that a particular syllable should be pronounced as tone 3, they are not giving sufficient information to tell you what contour the syllable actually has. Even if they do give such information, some research shows that speakers are not good at isolating changes in pitch from other vocal effects, for instance, what someone describes as a rise in pitch may actually be something else, like a change in volume, length, or clarity.

In a sequence of two or more tone-3 syllables, the penultimate one will always undergo some sort of sandhi that emphasizes an upward glide at the end.

In sequences of more than two tone -3 syllables, a speaker always thinks at least one or two syllables ahead. In considering what to do on any but the last two syllables in the sequence, a speaker considers whether the following tone-3 syllables form an indivisible sense unit. The sense unit is not limited to a word, but can include a phrase (e.g., 很好). On the other hand, even words may be divisible into obvious parts (e.g., 纸老虎 into 纸 and 老虎). If such an individual sense unit exists, the speaker will use a descending or low half tone 3. If not, a speaker will use or add an upward glide like tone-2 or perhaps a full tone 3 followed by a slight pause.

The issue is that tone sandhi in Mandarin is determined by what follows and yet a speaker must first utter the words that precede without necessarily always knowing in advance what will follow. A speaker can only predict a certain distance into the future and so thinks of suitable places to pause for the sense and grammar to be clear. Such pauses are actually an integral part of spoken Mandarin irrespective of tone issues. Before such possible pauses is where a speaker will use the half tone 3 in a sequence of three or more tone 3 syllables or use a full tone 3 instead of a simpler tone 2.

It is possible for a speaker to put two half tone 3 syllables in a row or to use a "full" tone 3 instead of a half one, but this seems to be avoided by most speakers unless they perceive a place for an optional sentence pause. I think this happens when speakers make slight mistakes in the sentence parsing as they speak on the fly, since this seems to occur only nearer the beginning of an utterance. In any case, I think speakers rarely put two full tone-3 syllables in a row.

An example from the research I cited is the following sentence:

lao3 shou3 zhang3 mai3 jiu3

老首长买酒

the old senior officer buys wine

In this example, among examples taken from six informants analyzed by machine,the first and the last three syllables were consistently: tone 2, half third, tone 2, and tone 3, respectively. The second syllable was toneless in statements; either toneless or half third in unmarked questions; or half third, toneless, or tone 2 in particle 吗 particle questions.

Another example is:

laogudong jiang3 lao3 shouzhang mai3 jiu3

老古董讲老首长买酒

the old fogey/fuddy-duddy talks about the old senior officer buying wine

Here again, the last three syllables were consistently half third, tone 2, and tone 3, respectively. Out of nine occurrences of 讲, eight were half third, and one was a full tone 3 (in a particle question). The first and the second syllable were also consistently tone 2 and toneless, respectively. The sequence 老首 Was pretty consistently tone 2 and toneless, respectively, with slight variation. The syllable 董 showed the most variation, but even here, the full tone 3 was used in the majority of statements and a toneless version was used in most of the questions.

[Guest realmayo]

By "full tone 3" you mean people first dipped, assuming they are dippers, then pronounced low, before coming up at the end? I'd understood that people rarely bothered coming up at the end of the third tone. I wonder why every respondent in the survey above gave the rise for the two 酒s.

[Altair]

I don't have a good explication for the discrepancy, but here are my speculations. All or none of the following might be true.

The paper was analyzing the effect of intonation on tone and so treated tone sandhi from that perspective. The tone graphs it showed sometimes show a very slight upward slant at the end of sentence-final tone-3 syllables. I guess some people could interpret these as canonically dipping tone-3 syllables, but others might hear them as mostly flat.

The paper stresses that individuals' perception of tone is colored by several factors beside pitch, and so even when people perceive a lack of a rise, it may nevertheless exist and be identifiable by machine. An example might be that some people radically decrease the volume at the end of a third tone, and this might make it seem flat in pitch.

The lack of an apparent rise on a final tone-3 syllable may be more an effect of sentence intonation than a true variation of the tone itself. The interplay of these two effects might color the perceptions of both speakers and listeners as to the true nature of the tone.

Pronouncing a full tone 3 seems connected with sentence stress, which correlates with how long a syllable is pronounced. In the sentence I described, all the full tone 3s were pronounced longer than other syllables, and the final syllable was often pronounced twice as long as the others. That might mean that unstressed third tones might lose the rise; whereas stressed ones would not.