Efficient Learning Strategy of Chinese Characters Based on Network Approach

[Yadang]

This is a quite recent (Aug. 21, 2013) study done on how to efficiently learn Chinese characters. The full research article can be found here: 

 

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0069745#pone.0069745-Yang1

 

Basically, as I understand it, the authors thought that learning characters just by frequency ignores their inter-relatedness that can be used to help learn them more efficiently. However, learning them just from a bottom-up approach based on their just components, although very efficient, leads to learning a great number of characters that are very infrequent before getting to the frequent ones.

 

So what they did here is try to combine these two methods where both frequency and components were taken into account.

 

 

 

A very cool idea, I think. Also, even cooler, is that they provide their findings (aka, the list of characters in the most efficient learning order based on their algorithm) for us to use!!!

 

Traditional: http://www.learnm.org/data/DNWorderT.txt

Simplified: http://www.learnm.org/data/DNWorder.txt

 

I think it would be totally cool if someone could incorporate this into a series of textbooks... Or even use something like this but instead of just using two node weights (character frequency and character composition), maybe perhaps three node weights could be used - character frequency, character composition and word frequency (although I'm not sure if this would actually work).

 

Anyways, what do you guys think - useful? Or just another character frequency list that ignores actual words and context?

 

 

 

[imron]

Or just another character frequency list that ignores actual words and context?

In my opinion, for someone who doesn't already speak the language, random lists of characters will never be as efficient as learning them from context (including text books).

 

These sorts of character-only lists might be useful for a heritage learner who spoke the language already and just wanted to figure out how to prioritise which characters to learn.

[renzhe]

I find it very interesting. I've often thought about the best way to go about learning characters from scratch, because it is such a fascinating topic. Learning based on frequency is clearly not the easiest path.

However, I don't think that this will be the silver bullet so many are hoping for. Perhaps it will cut down the needed time by 10-15%. However you do it, learning characters takes time.

[lechuan]

If you're going to learn from a character list, I see a lot of value in the approach. Tuttle's "Learning Chinese Characters" book already does that, only goes up to 800 characters though.

 

I'd love to see a combination of this type of "list" with corresponding reading exercises that make use of the characters learned thus far in each section.

[li3wei1]

The problem is that most students learning characters are also learning grammar and sentence patterns, and want to be able to say simple things about everyday objects in the first couple of weeks. So you need to learn 我，有，是，没 etc. whether they're on the list or not. Unless a course was designed to ignore characters for the first X months, and then attack the list when the students have already learned the first 100 or so orally. There's another thread on that subject somewhere.

[character]

If you're going to learn from a character list, I see a lot of value in the approach. Tuttle's "Learning Chinese Characters" book already does that, only goes up to 800 characters though.

There is also "Reading and Writing Chinese: Third Edition" which systematically builds characters based on strokes. From the Amazon description, the 3rd edition is using the HSK as the basis for choosing characters to present. I've used earlier editions; this latest one sounds good too.

 

I'd love to see a combination of this type of "list" with corresponding reading exercises that make use of the characters learned thus far in each section.

 

Sometimes the old ways are best...

http://www.chinese-forums.com/index.php?/topic/31477-why-dont-more-people-use-john-defrancis-chinese-reader-series/

[li3wei1]

Source 32 of the above paper describes how, at the University of Vermont, they teach speaking and reading separately. So you learn the most common words in the speaking class, without worrying how to read them, and the simplest characters in the writing class. At some point, presumably, they bring the two together. I'd be interested to hear from any graduates of this program.

[renzhe]

That approach is not new. John DeFrancis reported on similar experiments with Chinese kids a long time ago. He says that the results were that the kids who started reading/writing in pinyin only for the first year, and only studied characters later, outperformed the students following the "classical" approach.

I can dig out the reference later, if anyone is interested.

[Guest realmayo]

Interesting -- I've certainly said before on the forums that separating characters from listening/speaking seems the best approach for beginners, because it means that you can teach the key components of characters first, as well as some of the principles behind them. Something like this http://www.amazon.com/Fundamentals-Chinese-Characters-John-Jing-hua/dp/0300109458 book would be ideal.

 

I agree with li3wei3 it'd be interesting to hear what happened when they 'merged'.

[Guest realmayo]

Ha, actually just checking the University of Vermont and the guy who wrote that book I mentioned is in fact the chair of their faculty! John Jing-hua Yin. And yes they do indeed use his book.

[Yadang]

Renzhe, I would be interested, if it's not too much trouble...

Yeah, I'm not sure how useful this list is on its own, but I think it could definitely be cool if someone incorporated it into a series of textbooks... I think if they designed a course where they completely separated characters from speech for a while, and then slowly brought them together, and used something like this as the basis of the character learning part, and added in a lot of reading practice, plus all of the great tips here on chinese-forums.com, they'd have an awesome textbook.

[renzhe]

It doesn't seem to be in DeFrancis, unfortunately. I remember reading about it somewhere, maybe something from here, but can't seem to find it now.

DeFrancis reports on Mao Dun saying that his granddaughter had mastered Pinyin by the end of first grade, but this lapsed when they switched to characters only in the second grade. He also reports on letters from first-graders in Harbin in 1962, who mixed Pinyin and characters and were thus functionally literate.

I remember reading about a school in which one class learned only pinyin in the first grade (and then characters), and the other one learned characters only, but I can't seem to find it now.

[character]

^ Thanks for looking.  Honestly, such a study didn't seem to have wide applicability, as it sounded like it was of Chinese children, presumably in an immersive environment.  There is good evidence such students already have the best success at learning Chinese.

 

Based on fellow classmates and my own experience, I think even passively learning characters helps adult learners keep similar-sounding words separate in their minds.

[renzhe]

I think the point was that kids can be literate using pinyin much faster than with characters, and that delaying learning of characters did not hurt their progress later. But yeah, these were native speaker children, so a different story.

I should stop posting about things I read years ago, as my memory is apparently not as good as it used to be...

[Yadang]

Thanks for looking! Looks like some cool stuff in the link you posted...

[roddy]

Renzhe, you're ringing a bell with me, but I'm not sure if I'm remembering the same thing. I seem to recall a comparison where first grade Chinese kids learning pinyin only outperformed first grade English kids, as pinyin was a more phonetic system. But as soon as the characters were introduced, the English kids sped ahead. 

[akdn]

 the results were that the kids who started reading/writing in pinyin only for the first year, and only studied characters later, outperformed the students following the "classical" approach

 

 

This article reviews the experiments that were carried out investigating and demonstrating a positive impact of pinyin on literacy among primary children in China in the 1980s (in Heilongjiang) and early 1990s:

 

Liu, Y.B. (2008),  'A Pedagogy for Digraphia: An Analysis of the Impact of Pinyin on Literacy Teaching in China and its Implications for Curricular and Pedagogical Innovations in a Wider Community', Language and Education 19(5) [Online] pp.400-414. Available: http://www.tandfonline.com/doi/abs/10.1080/09500780508668693#

 

The original research is reported here:

 

Ding, Y.C., Li, N. and Bao, Q.E. (1985) Zhuyin shizi, tiqian duxie shiyan baogao [The report on Pinyin annotated character recognition that promotes earlier reading and writing]. Yuwen xiandaihua 8, 134–48.

 

 

my memory is apparently not as good as it used to be

 

 

To sum up, your memory appears to be fine.

[renzhe]

akdn to the rescue! Thanks!

[hedwards]

I haven't seen that article in a while, but I think the thing to realize is that this is about the efficiency in learning to read all the characters you'll need. As in probably a few thousand of them. And from that point of view, it's a reasonable way of going about it. Because of that, it's probably more applicable to people that are being forced to learn to read rather than those who have it as an option. As in, Chinese people and people living in other places where the written language is Chinese.

 

The problem that I have with this sort of organization is that it means that you're spending a lot of time learning characters that aren't necessarily high value characters because they're faster to learn than ones that might be higher value. And for individuals that are particularly dedicated and driven, I think it's likely that it does save time. The problem though is that Chinese has such a high burnout rate for people that are learning to read, that whether or not it's more efficient to go about it like this winds up being pointless in most cases.

 

IMHO, it's been a lot easier to familiarize myself with most of the radicals, in terms of meaning and reading, and then just look at as many characters as I can find. It might not be as efficient in the long term, but following that basic path I've stuck to it. Whereas when I've been memorizing random high frequency words, I've had a really hard time with motivation. It also tends to lend itself better to mnemonics for the more complicated characters. 龍 winds up being a piece of meat stood up away from a village to lure the dragon a long way away. Going with this route, I might or might not get the pieces I need to do that before I get to the character.

[Yadang]

hedwards said:

 

The problem that I have with this sort of organization is that it means that you're spending a lot of time learning characters that aren't necessarily high value characters because they're faster to learn than ones that might be higher value. And for individuals that are particularly dedicated and driven, I think it's likely that it does save time. 

I think this occurs somewhat, but far less than with other methods that try to teach characters in a more "efficient" way (like studying them by common component, going Heisig/Matthews/Hoenig/T.K. Ann style or things like that). However, because one of the two criteria that went into making this list was frequency, I think on average, most of the characters that are learned are going to be of pretty high value (although it might be a problem that the list was made with character frequency and not word frequency... I'm not sure how much a problem that causes though...)

 

 

The problem though is that Chinese has such a high burnout rate for people that are learning to read, that whether or not it's more efficient to go about it like this winds up being pointless in most cases.

That's an interesting point.

 

 

It also tends to lend itself better to mnemonics for the more complicated characters. 龍 winds up being a piece of meat stood up away from a village to lure the dragon a long way away. Going with this route, I might or might not get the pieces I need to do that before I get to the character.

The other criterion for their algorithm was to know all of the components that went into a character before learning that character. While I think because they tried to strike a balance between frequency (the first criterion) and knowing all of the components to make up a character before learning said character (the second criterion), I think for the most part, one will know all the components that make up a character that they are learning. I'm sure there are times though, that because of the frequency criterion, one of the components is left out until you move further down the list (although, I suspect they would be pretty close on the list..)