More Advice for Learning Chinese: Possibly Controversial

[Shelley]

I said "do not neglect the hard bits........" meaning anything you might consider hard when learning Chinese.

 

You didn't say it, I did.

 

For what its worth I can't understand the lyrics of songs sung in English a lot of the time, I think that is spurious argument for tones not being important.

 

There is even a word for the mishearing of lyrics - Mondegreen.

 

As I said before - the pronunciation of a Chinese character is the pinyin and tone together there is no separating them - learn it together.

 

People don't speak English in a monotone way, there are inflections for many things, questions, disbelief etc. We didn't learn them separately, they were part of learning English and were absorbed along with the words.

 

When native speakers appear to understand you it is because they are making allowances for you being a learner and will try really hard from context to understand. But having said that there are many examples of native speakers stubbornly ignoring your attempts.

 

IMHO its simple, Chinese has tones - if you want to learn Chinese you need to learn tones - end of story.

[imron]

At what point can you stop focusing on tones?

When you have internalised them to the point where you don't see tones as being separate from the pronunciation, and where different tones are as distinct as different consonants or vowels.

I never remember the tones - I remember the sound (with tone built in), and just like I would remember the difference and keep words like 'car' and 'scar' or 'scam' and 'scan' separate, likewise the same will happen with sounds containing different tones. If I need to know the tone I'll work backwards from the sound.

Once you get to that point you will have no need to focus on tones specifically.

[gato]

Now when I talk to someone on the phone, they think I'm a normal putonghua speaker, if not a 北京人。 But I can't tell you what tones are for 60-70% of what I say.

You don't mean you "can't tell what the tones are" for words you can pronounce correctly. You are pronouncing the correct tones, but you are advanced enough that you are no longer thinking about the tones consciously. Obviously, if you had to write the tones out, you can just by thinking about it a little more. You just don't think about the tones when you are speaking because you have internalized it.

This is just like native speakers don't think consciously about pinyin when you are talking or don't think about grammar when putting words together when speaking. It's subconscious. That is different than not being able to "tell" the pinyin or grammar when asked to turn the subconscious into conscious.

Another analogy is that many native speakers can't explain grammar rules because they haven't studied them in depth, but that doesn't mean grammar isn't important.

[anonymoose]

Yet, the existence of Chinese song lyrics doesn't fit with most of the arguments being made. Native speakers and advanced non-native speakers can understand the topic and 60% to 100% of the lyrics on first listening, without a lyric sheet.

 

Without tones.

 

How do you explain it?

 

But now when I listen to unfamiliar songs, or old songs that I never once looked at lyrics for, I can understand 60% to 90%. Sometimes I do think I understand something, and it makes sense in context, and then I look at the lyrics and realize I misunderstood. That happens maybe 10% of the time.

 

Those are things I'm trying to square with the relative importance of tones.

 

I don't think song lyrics constitute very good support for your argument as there are many factors at play other than pure language recognition. But taking your argument at face value, I'd say it pretty much demonstrates the importance of tones. After all, 60% to 90% comprehension is a dismally low range for efficient communication. In normal speech between native or advanced speakers, one would hope for 99% to 100% comprehension.

[陳德聰]

Man, this is lively!

OP, I get what you're getting at. There is definitely a way of empirically measuring how much processing is required for tones vs. other aspects, and I am pretty certain it has been studied before. Just gotta find the right journal, which I will do when I get home tonight!

[Ruben von Zwack]

I was just listening to a few humorous podcasts on my phone, and for lack of comparison I don't know whether they were typical or not, but the tones in the guy's speaking were very prominent.

Much more than in daily life speech, or in good speech you hear on a talkshow. I found that interesting in the light of this topic.

Spoken word comediens are basically brilliant rhetors.

So my personal conclusion from that is: if you intend to use words precisely, be well-understood, even surprising, then do not under estimate tones.

[li3wei1]

More anecdotal evidence: I attended a workshop for teachers of Chinese as a foreign language. Almost all the participants were native speakers. The speaker wrote a sentence on the board and asked us all to write it in pinyin. Hardly anyone got it all right, and many of the mistakes were tones. Not sure that proves anything, but I thought I'd throw it in in the spirit of open-minded enquiry.

[alanmd]

Nathan, none of your anecdotes contradict what I said. It is of course possible to understand toneless Chinese, as it is possible to understand English with no vowels, but it is more work for the listener/reader, and it requires more context.

If you ignore the tones completely when you learn Chinese, you may still be able to make yourself understood, but you will sound about as bad to a native speaker as semwin wha spokes Anglush weth ull thi wrang viwols (someone who speaks English with all the wrong vowels).

I agree with the points made by @imron and @shelley in their last posts. Tones are as much a part of correct Chinese pronunciation as the intonation of words in English. It's a bOttle, not a bottLe, stressing the second syllable is still understandable, but just sounds wrong.

[querido]

Alanmd, did you say vowels?

"The most interesting result is that identifying the tone of a syllable is at least as important as identifying the vowels in the syllable."

The Functional Load of Tone in Mandarin is as High as that of Vowels (pdf)

[li3wei1]

At last, some science!

 

May I request that we refrain from any more posts to the effect that ignoring tones is a bad idea? I don't think anybody disagrees with that. Let's move on.

[alanmd]

I made up that analogy to highlight how important I thought tones were, I didn't think that it would be backed by research! Very interesting, thanks. Your link didn't work for me, this one does: http://www.isca-speech.org/archive_open/sp2004/sp04_099.pdf

[Nathan Mao]

I'm surprised they found that tones were as important as vowels!

From the article:

 

One could argue that it is more important to effectively teach or accurately model a contrast that carries more information.

This was my point from the very beginning.  Not to completely ignore tones, but to spend less time on them relative to word order, grammar, and stress.

 

I do think the research was flawed in that it assumed that stress is equivalent to tones, and that Mandarin has no stress.  However, I don't think that would have changed the computations enough to result in a different conclusion. 

 

Moreover, I want to point out that this scholarly article used a proposed method of calculating functional load of various elements of speech.  Meaning, this research:

1) is based on an unproven theory of Functional Load. If you reject the premise, the conclusion is immaterial, and

2) did not actually test comprehension; the actual impact of tones versus vowels may be much different than these calculations

 

That being said, this research is directly applicable to the topic at hand, and even more importantly, is the best science we have available to the discussion.

 

As such, I don't reject the premise, and withdraw my assertion that one can study tones slightly less than one studies the other elements of Chinese Mandarin and still succeed.

 

More scholarly articles on this topic would be welcome, if anyone knows of any.

[Ruben von Zwack]

I'm surprised to hear functional laod is an unproven theory. I did study linguistics quite extensively (my major was art history but for a German Master we were required to have a full second field) but it's a few years ago, so I guess I'm outdated.

[Nathan Mao]

Functional load isn't an unproven theory.

The computational framework for measuring the functional load is an unproven theory.

 

From the paper: "Using a recently proposed framework..."  (emphasis mine).  'Recently proposed' is absolutely not 'vetted and verified', and as such, the framework itself is unproven theory.  The calculations work, but do the conclusions apply in real-life comprehension?  That's beyond the scope of the research.

[陳德聰]

I've just finished reading a paper linked here.

Something that supports OP's inference that tones are less important than other parts of words is quoted below.

 

 

In order to determine how much information each subpart contributes, we can compute the size of the equivalence class that results from specifying only one subpart (Altman, 1990). For example, the equivalence class size (ECS) for the onset /p/ is defined as the number of syllable + tone combinations that begin with /p/. Of the 1251 syllable + tone combinations in the database, 56 (4.5%) begin with /p/, meaning that the ECS for /p/ onsets is 56. Simply by knowing that a syllable + tone combination begins with /p/, a listener has a 1/56 (4.5%) chance of identifying that particular combination correctly as compared to the base chance probability (1/1251). Thus, ECS provides an index of the relative informational value of a given subpart. The larger the ECS value, the less informative it is. By computing the ECS for each onset (rime, tone) separately, we were able to derive an average ECS across all onsets (rimes, tones). The average ECS (SD enclosed in parentheses) for onsets, rimes, and tones were, in order, 52.1 (12.7), 36.8 (23.0), and 310 (37.6). A one-way ANOVA of ECS with three levels (onset, rime, tone) revealed a significant main effect, F(2,59) = 314.05, p<.001. Post-hoc Tukey HSD pairwise comparisons (onset vs. tone; rime vs. onset; rime vs. tone) were all significant (p<.03). Since informational value is inversely proportional to ECS, we may infer that rimes are significantly more informative than consonants, and consonants, in turn, are significantly more informative than tones in Mandarin Chinese.

Summary:

In terms of informative value, the relative ranking in Mandarin is finals > initials > tones. This is relatively intuitive, considering the fact that there are are only 4 possible tones.

The study itself actually just measured interference effects of consonants vs. vowels vs. tones, and found that vowels have the strongest ability to confound the other two dimensions. The reason I felt the above quote was relevant is that it is a reminder that OP is totally founded in asserting that tones are not as informative as the other two features (consonants, vowels).

Sometimes it seems that there is a bit of ideological resistance in this forum. Tones are clearly important and definitely phonemic, but they are by far the least salient of the three dimensions.

Edit: After having read the Tong et al. paper (quoted above), I'm a little bit sceptical about the Functional Load conclusion there. If FL measures importance of a feature for creating contrast, and vowels and tones are comparable in their ability to create this contrast, does that necessitate tones being as "important" as vowels? I can see how this would support the control results in the study by Tong et al., since reaction times for classifying the two should be comparable. But I am a bit confused about why, if having comparable FL results, the interference of vowels on tone classification ability would be so much stronger than vice versa.

[alanmd]

I think all of these academic results should be taken with a grain of salt, although they are very interesting.

Learn everything, all at the same time, and don't worry too much if you mess up hear or there occasionally because language has redundant information and context to save you...

[陳德聰]

I think it's a bit silly to dismiss research that lends itself to the very theoretical discussion going on here.

[Nathan Mao]

I concur that the research links are helpful. 

 

Even if the research/scholarly works don't definitively answer the question, they are still more structured than discussing anecdotes and experiences that may apply to only that one person's specific circumstances.

 

Probably the question won't be answered within several lifetimes, much less ours.

But the linked research has given me a better idea of how/if I can do my own research on this when/if I go for a PhD in Chinese Linguistics.

[lechuan]

The stronger you are in other areas of the language, the better your ability to understand by context. At that point, I'd call bad tones an accent.

 

If I listen to a fluent speaker (vs english), with an accent, speaking english, I probably won't have much problem understanding them. But, I assume that most learner's goal is to speak with a minimal accent.

 

I know a few learners, with pretty good Chinese, who have no idea what tones they're using, but they get the majority right because they learned through imitation. They may not know their tone symbols, but they still took care to imitate accurately.

 

But I think it's hurtful to the beginner to tell them not to worry too much about tones.  It's a lot harder to undo badly learned tones practiced over years, making it hard to get rid of the accent later on. Speaking from experience, I learned the tones wrong on a few words early on in my studies, and it took a long time to fix them when it was pointed out (by a native speaker) that they were wrong.

[gato]

In terms of informative value, the relative ranking in Mandarin is finals > initials > tones. This is relatively intuitive, considering the fact that there are are only 4 possible tones.

Since it's unlikely a learner is going to leave out finals and initials when speaking, the more relevant question is whether the information value of tones sufficient large that if one pronounces everything with a neutral tone, one can still be understood.

Comparing toneless vs initial-less Chinese isn't all that meaningful from the perspective of learning Chinese. Besides, the calculations in the paper are still too simplistic even given their goals. For example, it doesn't consider the similarities between certain initials, such as zh and z, ch and c. You are more likely to be understood if you are making a z/zh mistake than if you were substituting, say, d for z. You need to consider "similar soundingness" if you want to figure out what kind of mistakes are more tolerable than others.