Heisig for Chinese; are you sure?

[Hofmann]

Yes, let's talk about it. It might get ugly, but we have to talk about this in order to understand it.

James Heisig developed a method of learning the most common Kanji used in Japanese. Read about the method here. In 2009, a book Remembering the Hanzi was published that applies that method to Chinese. Although I find it a good method for Japanese, I think there are better options for Chinese.

First, why does it work for Japanese? Japanese usually has more than one way to read any Chinese character. Among those, the Kun'yomi have nothing to do with the character. Having no basis in the characters, many readings would be difficult to memorize. Therefore, Heisig ignores the pronunciation altogether and just focusses on the meaning. As about 90% of Chinese characters are phono-semantic compounds, there is often a component in a character that has nothing to do with the meaning. In order to incorporate that component into Heisig's mnemonic, he makes up a far-fetched story. In many cases, it has to be far-fetched in order to successfully incorporate the phonetic component. Furthermore, some components are given fake meanings just so that their usage in mnemonics is facilitated.

I remember a mnemonic for 切. We can explain this as a phono-semantic compound by saying it sounds like 七 and it has something to do with 刀. Before introducing 切, Heisig gave the fake meaning of "diced" to 七. Using that, his explanation of 切 is that "To the right we see the dagger [刀] and next to it the number seven whose primitive meaning we decided would be diced. It is hard to think of cutting anything with a knife without imagining one of those skillful Japanese chefs. Only let us say that he has had too much to drink at a party, grabs a dagger lying on the mantelpiece and starts dicing up everything in sight, starting with the hors d’oeuvres and going on to the furniture and the carpets…."

The idea is that eventually, the stories would be forgotten, and only the meaning of the character would remain. That's fine for Japanese, and I don't think I could make up a better method. However, for Chinese, I see better options. As Chinese languages have no other readings besides Chinese readings, phono-semantic compounds' pronunciations are always related to their phonetic components. If one used Heisig's method to learn Chinese, one needlessly ignores pronunciation and the phonetic information present in about 90% of all characters. For example, to ignore the way that 召 influences the pronunciations of 招, 昭, 照, 詔, etc. is a waste of good information. Furthermore, one needs to use the phonetic information as if it were semantic, requiring far-fetched stories, which is an unnecessary inconvenience.

Better would be to use the components as they are; using phonetic components to hint at pronunciation, and using semantic components to hint at meaning. That way, most characters can be easily and simply taken care of. Furthermore, the real etymology of the characters is learned, which may be more useful than one might think.

Therefore, I recommend the following method for learning Chinese characters instead of Heisig's method for Chinese learners:

1. Just learn it.

2. If that fails, actually pay attention to it and try and remember. Writing it once might not hurt either. Writing it once while paying attention is better than writing it 100 times and not paying attention. Don't write it more than 5 times.

3. If that fails, get more information about the character. More information associated with something makes it easier to remember.

4. If that fails, then try a mnemonic. If possible try to use the real etymology.

5. If that fails, then try something like Heisig's method.

I have to say, though, that if you're using Heisig's method for Chinese, and you like it, then whatever. However, if you're considering it, look at other options first.

[Groat]

I've been studying characters in multiple different ways and I agree with you to a certain extent. Picking up Heisig after I knew about 400 or so characters allowed me to fix the method a lot and include pronunciation (I learn Heisig with Anki with pronunciation in the answer side). I also do not use some of his fake meanings, like slices and whatnot, but use them for what they mean.

However, what the Heisig method excels in is the order of the characters. It's very structured and has helped me a lot. The fact that Heisig has a good number of characters using the same radicals grouped together helps me to remember them. It also helps me differentiate between them. I think it's quite logical.

One annoyance with Heisig is that I tend to lose characters when I read them "Heisig style" by radical. If I knew that character by heart before and could always recognize it in text I seem to forget it when I do my reviews in Anki with a Heisig mindset. I'm not sure what to make of this yet.

One thing that should be noted is, that I only study characters for passive recognition. Not for writing.

[Guest realmayo]

This was discussed here: http://www.chinese-forums.com/index.php?/topic/14001-remembering-simplified-hanzi-1-and-remembering-traditional-hanzi-1. I'm a huge fan of mnemonics, it's a proven way to greatly improve the basic human memory and has a very long tradition in Western culture. It seems a pity not to use this excellent tool.

But yes, Chinese characters include plenty of phonetic information, it seems a pity not to use that information when learning Chinese characters.

So, for me, the best approach lies simply in combining the two.

Better would be to use the components as they are; using phonetic components to hint at pronunciation, and using semantic components to hint at meaning

Well, in the example given of 招, this would result in: "hand that sounds like zhao".

What works better, for me, is to make up a story or an image involving hands and summoning or convening. (So, if a character's mnemonic involves "summoning/convening", I know it will have 召 and will probably sound like "zhao".)

Now, maybe 招 was not a good example, because 召 is surely not just phonetic here, it also indicates the meaning. But even for characters where the phonetic part is purely phonetic -- ie having checked to see that there is no accepted or credible etymological explanation for a character -- I still use this method.

I'd also point out that, once one knows about Chinese radicals, it is of course usually very easy to work out what is the phonetic part, and use this as a hint towards the sound.

With a tool like Wenlin, which gives you plenty of information about the radical, the phonetic part, other characters which share the phonetic component, and etymologies or etymological guesses, I've found it fairly simple to learn characters. I used another book, which gave "stories" for the first 800 characters, and with Wenlin, have done the rest myself.

Perhaps my concern with Heising is that if you follow him, you don't have to do any of this work yourself. I understand his first book covers much more than 800 characters, and that there's a sequel coming? And I believe he doesn't deal at all with the pronunciation of characters: part of the mnemonic I use for each character includes the sound and tone. But again, this is covered in the thread linked above.

[renzhe]

However, what the Heisig method excels in is the order of the characters.

I tend to feel that this is the most important part too.

I learned characters the "conventional" way, generally starting with common characters first and working towards the less common ones (this is the way most textbooks will work too). It worked for me, but the first 2000 characters were quite painful, with very complex characters consisting of many parts introduced early, before you knew either the meaning or the pronunciation of the parts.

If I were starting again, the main thing I would change is the order in which I learn characters. That's the one thing Heisig got right. It doesn't matter that 警 is far more common than 苟, you should first learn 言, 夊 and 苟 (after you've learned 句).

[Guest realmayo]

I agree renzhe. For quite a while now, whenever I have a new character to learn, I will make sure I know all its components first. Then, if its phonetic component is used by other characters, I'll learn the meaning, if there is one, of the phonetic component. If there isn't a meaning, I'll make one up. Then I'll learn at least one or two of the more common characters using this new phonetic component. I'll also learn any other characters which contain the new character I want to learn. Finally, I'll learn at least one 词, one multi-character "word", for each new character I'm learning.

So a while ago, learning one new character could easily mean learning 10 or more items.

[Guest Pipas]

Better would be to use the components as they are

As they are. The thing is they are complicated creatures..

Who wants to know there is NO crescent in the 有. And in the 歺 either!, though its components look similar to those in 外.

And who cares there's a cowry(贝) inside "得"?

And you are saying people ignore the phonetic information. They used to ignore from the beginning.

Don't write it more than 5 times.

why not? Isn't it a pleasure after all? 學而時習之， 不亦說乎?)))

[Hofmann]

Who wants to know there is NO crescent in the 有. And in the 歺 either!, though its components look similar to those in 外.

And who cares there's a cowry(贝) inside "得"?

I don't know who wants to know that there is no 月 in 有 and no 夕 in 歺, but if that was a rhetorical question that implied that nobody wants to know, I disagree.

得 has 彳 and 㝵. 㝵 has 見 and 寸. I don't know who cares, but if you imply that nobody cares, I disagree.

People have their own reasons for learning about different things. In my first post I suggested getting more information about characters to see if it helps one's memory. I didn't suggest that one learns things like this if it doesn't help.

And you are saying people ignore the phonetic information. They used to ignore from the beginning.

Let's express ourselves clearly here. Some people who have no reason learn and use phonetic information may understandably ignore phonetic information. Others may benefit from learning it, and heed it. If you mean that nobody heeds it, I disagree.

why not? Isn't it a pleasure after all? 學而時習之， 不亦說乎?)))

You asked me why not write a character more than five times in an effort to remember a character if just learning it upon initial exposure fails. I assume the objective is to remember how to write a character, its meaning, and its pronunciation as efficiently as possible. What writing a character repeatedly accomplishes is refinement of writing technique, but only if done correctly. While it should help one remember a character's writing, it does nothing to strengthen one's memory of a character's meaning and pronunciation. Furthermore, there are other ways (e.g. learning more about the character) that aid memory of writing which are more efficient than writing something repeatedly. I have never written most of the characters I know. Writing it once (while paying attention) is usually enough for me. Twice while paying attention and it's practically permanent. I allowed a bit more headroom, so I said 5 times should be more than enough.

You asked me 學而時習之不亦說乎. I don't know, but if you implied that I instructed our readers not to learn, you have misinterpreted me. For example, although writing something repeatedly is a way of learning, there are more efficient ways to accomplish the given objectives. While it may be a pleasure to learn, it should be a greater pleasure to learn something in less time and with less effort.

Edited November 7, 2009 at 08:52 AM by Hofmann

[Kai13]

I finished "Remembering the Kanji", and believe me, it's really easy to learn chinese characters now (even the simplified, which I'm learning). I had been struggling with kanji, I finished the book precisely when college started. And my characters are beautiful and I have no problems in writing them. I must say that it is much more easier to memorize characters for chinese rather than japanese. Even though you have to use more. Just don't ask why, I just feel this.

[chrix]

I would like to bump this thread.

So I'd like to ask those of you who have successfully worked through Heisig: Would it have helped if information about the pronunciation would have been included? I know that's not part of the Heisig method, but how do you then acquire accurate knowledge of how characters are pronounced? Are there book on that, like the "Heisig Phonetic Complements Companion" ?

I just think an approach that teaches about character components should probably also include the phonetic components in a systematic way.

[M-Segments]

Chris, in my opinion, the Heisig method would not be improved by including the character's pronunciation. I've blogged about this quite a bit before - but I think people often miss the point with Heisig.

Ultimately, Heisig is just a tool. It should be one of many that you use when studying Chinese. And by cramming more into the method - pronunciation in this case - it feels like people are trying to expand it into the one-tool-that-teaches-you-everything.

If I want to improve my listening skills, I use podcasts. If I want to practise speaking, I'll use a language partner. If I want to learn the meaning of written Chinese, I'll use Heisig.

In my experience, the mind's ability to create & remember images is much less limited that the brain's ability to memorise by rote, like with pronuncation. When I tried learning pronuncation alongside my Heisig work, my progress was much slower. By dumping the rote stuff, I really sped up.

A few months after finishing Heisig, I am going through the book again. But what I'm finding is that I already know a lot of the pronuncations because of the flaschard work that I have been doing to learn compound words. And because I now know the hanzi, I am finding it much easier to learn the pronuncation on second read-through.

So my recommendation is to use Heisig as it was intended - for the meanings of the characters. You can learn the pronunciation later - it'll be much easier then.

[Laughing Squid]

I used Heisig to learn the meaning of 2,042 characters in Japanese. I did not learn the readings when using Heisig, but there is no real systematic way of learning the KUN reading, it can only be done on a case-by-case basis. Still, knowing the meaning of a character is better than not knowing the meaning, and the time investment has more than paid for itself in my case. I developed a system for memorizing the ON reading of Japanese characters, and it has been quite successful at systematically learning just about any Japanese compound word.

Heisig for Chinese makes even more sense because each character has only one or two readings, usually only one that is frequently used. The challenge of learning the Chinese character readings is very small compared to Japanese.

Basically, learning the meaning of Chinese characters puts you on the same level as a Japanese person learning Chinese or visa-versa. You know what they mean, but you must learn how to pronounce them. Although it takes some time to learn the meanings of many characters, it greatly evens the language playing field. Now that I've learned the meaning of about 2,000 Chinese characters, learning new words is like shooting fish in a barrel. The only troublesome problem I'm encountering is "What words should I learn next?"

[James Stange]

M-Segments... I couldn't agree you with more and I found your blog very helpful when I worked through Heisig myself. So thanks.

Before Heisig when I would attempt to learn a character I would subject myself to information overload. It was all too much to try to learn to write the character, recognize the character, learn the most common definition(s) and most common pronunciation(s) and tone(s). Progress was slow and was often confused by similar looking characters. At first, I didn't like Heisig's approach either as it didn't include the pronunciation and I would have to spend time thinking up mnemonics. As my current learning method wasn't working I eventually opened my mind and gave it a try. The method was quite arduous but combined with SRS Software (PlecoDict!) it worked. Afterwards I found that I could better "process" new written Characters which freed me to concentrate on learning pronunciation and tones (which still need more work). FYI, I did include pronunciation on my SRS flashcards to get "exposure" to it but I didn't score a card as incorrect if I got the pronunciation or tone wrong nor did I spend time trying to remember the tone. Later after completing Heisig I cleared my flashcard scores and on the second pass I included pronunciation/tone in my testing. Also, occasionally I didn't find Heisig's "key work" vivid enough so I would make my own "key words" based on visual appearance or entomology. I found http://kanji.koohii.com/ and http://www.chineseetymology.org useful at these times.

Before I was always stuck learning around 500 characters I had thought I would never get to 1000 or more characters. Being a visual learner having access to the written language has really boosted my learning of Chinese vocabulary. I've since learned another 1000 characters and have read through several of my wife's old 小叮噹 books... maybe someday I'll be able to read some of her favorite 倪匡 books.

Heisig is an excellent resource for those learning Chinese Characters which due to their complexity, compared to alphabetical systems, really deserved isolated study. Of course later study will be required to learn the "word" behind the "character" but Heisig lays a foundation for this later study. It's not for everyone but those learning characters owe it to themself to have an open mind and try the free chapter available on Heisig's website.

[dreamon]

Suppose that the forum conducts a "census" of successfully memorized characters across a wide variety of Chinese learners. The result of the "census" are several thousand sets of characters, each set represents what one student learned so far. Some sets are very small, some medium, some large. Now, can we use a machine learning algorithm on these sets in order to identify the most optimal ordering of characters to study? For example, if character X is almost always memorized after a subset {A01, ..., A99}, then the optimal ordering places this subset prior to X. It may turn out that more than one optimal ordering exists (i.e. there are several local optima), reflecting several types of learners: visual, auditory, mnemonic etc.