memorising & RETAINING Chinese characters

[huaqiao06]

hoping people can share their techniques for someone with a western educational background to learn new characters.

I have a high attrition rate for writing recently learned hanzi from memory, and I'm tired of being had for breakfast by Korean classmates, who seem to do it naturally...

[adrianlondon]

Yeah, but their pronunciation sucks, so you can get revenge during shuo hua classes.

[imron]

Can we also share our techniques about searching for previous threads on the same topic?

(don't forget to try with American spelling also).

General consensus seems to be that that Spaced Repetition Systems like Anki, Supermemo or Mnemosyne, are the go.

[haton]

Hello huaqiao,

I had (have) exactly the same problem. That's why I wrote a small memory game in order to train Chinese vocabulary. It is called Pingrid. Used regularly, it is a great help.

You can get it at:

http://ehaton.blogspot.com

[kdavid]

I use ZDT. Input new vocab into a new category and learn them in this pattern:

Say I have 100 new words:

1 - 5

5 - 1 (backwards)

1 - 10

10 - 1 (do this until I know them all)

1 - 20

20 - 1 (repeat until I know them all)

20 - 30

30 - 20

20 - 40 (repeat)

and do this until I've worked all the way up to 100.

I can usually do this whole process in a little over an hour. After that, review a couple times a day until I can go through the whole 100 in a few minutes. Then shelve them and review once or twice a week.

[self-taught-mba]

SRS programs are great once the knowledge is partially there but you need "a knocker" to get it in there (knock it in) in the first place.

A character isn't like another fact so it needs to be repeated more often than many srs programs do at the initial stage.

so I think kdavids suggestion is great for initial retention. folllow up with SRS.

--typed

[newyorkeric]

Do you use any kind of mnemonic device? I find that having a story that summarizes each character and pronunciation helps a lot. I memorize the story and then use Anki for review (the question is the word in English and the answer is the character and pronunciation). The method with stories for 800 characters is explained in Learning Chinese Characters by Mathews and Mathews. It's a very nice book if you like this approach.

[OneEye]

Two parts to the question. Memorizing and retaining. Both are easy but they require different answers.

1. Memorizing is done most effectively by creating mnemonics. These should be as simple as possible, but not so simple that they won't work. No need to write a complex story when a few words will do. Make stories about the different parts (radicals, 部首, components, primitives, whatever you want to call them).

The best books to guide you on this are Heisig's Remembering the Kanji (it's for Japanese, but can be adapted for Chinese easily until his Hanzi book is released), the Matthews and Matthews book (although at only 800 characters until part 2 is released, I wouldn't use this one, though the book is quite good for what it has), Harbaugh's Chinese Characters: A Genealogy and Dictionary, and McNaughton's Reading and writing Chinese. Take your pick. I'm using McNaughton's book because it is presented in a pedagogical order like the Heisig book, but is for Chinese and includes readings.

2. Now for retaining. An SRS is the way to go, hands down. Once you've learned the character, doing your daily SRS reps will ensure that you don't forget it. You can "forget about forgetting," as Supermemo's page says (do I have to pay royalties for that?). I'm adding about 20 characters per day to my SRS (I use the flashcard system in Pleco because it has a space to write the character). I set it to prompt me with the pinyin and definition, and I have to write the character from memory. Any wrong stroke, and I count it wrong (even a gou where it shouldn't be). So far (several hundred characters into it) my retention rate is above 95%. Not bad, considering that many of the characters I hadn't even recognized before.

Check the method at for more info (after reading the overview, read through all the links under each Phase at the right). I also write about this kind of approach quite a bit at my own blog (from which I receive no profit, so hopefully it's ok for me to plug it here), located at the end of that link in my signature.

[newyorkeric]

The problem with Harbaugh and McNaughton is that they don't include any mnemonics (as far as I know). IMHO, it's better to have someone create the mnemonics for you in the beginning until you understand how they work. Then, you can apply the technique on your own.

[OneEye]

That's easy enough to solve if you're willing to put in a tiny bit of extra effort. And it's worth the effort. The Matthews' book is good, but for the purpose of becoming functionally literate, the others are better, just by virtue of how much more thorough they are.

Download the sample of Heisig's book. Read the intro and the first few chapters, or all of the sample. You'll have a pretty good understanding of the method after this.

And both books have simple mnemonics for some characters. You'll have to create some of them yourself, but that's GOOD. You'll remember it better that way. And you have to do that with Heisig's book too. I think that if anything, the Matthews' book is too helpful, spoonfeeding you everything. If you put a little mental effort into it, you'll be more likely to remember it.

But that brings me to the question of how many characters each book teaches. The Matthews' book has 800. The McNaughton book has around 2300 and the Harbaugh book has 4280. Sure, no book will teach you EVERY character. You'll have to learn some on your own one way or the other, but 800 just doesn't seem like enough to be worth the money. And I don't even know if they have plans to release vol. 2 anytime soon (at least not that I've been able to find).

[huaqiao06]

thanks all, should have posted this question years ago... you'd think that memorisation technique would be the first thing taught to foreigners, but it's as if my teachers here in Beijing (all masters candidates writing theses on teaching Chinese as a second language) never covered it at all

[OneEye]

Well honestly, Chinese learners have been a little slow on the uptake with this mnemonic thing, especially Heisig-style character learning. Japanese learners have a serious leg up on us in that department, so we'd do well in taking a page out of their book.

A common argument against the system is that if you use Heisig's approach, you lose the helpful phonetic information in the characters. There are two possible ways to deal with this. One is to say "screw the phonetics," and just stick to learning to write the character out given the English keyword. The other (which is equally valid) is to do what I do and include the pinyin in the question side of the flashcard. It sounds like cheating but it isn't. It has actually helped me to remember the pronunciation of the characters I'm learning, even if they are completely new to me.

For example, I know that any character in my flashcard deck pronounced 'zhu' (any tone) will most likely have 主 or 者 in it. And when I see those characters in context, I remember them pretty accurately, even if it is the first time I've seen it outside of the flashcard deck. So, pinyin and definition in the question, character, pinyin, and definition in the answer.

And it really isn't surprising that you're not finding more effective techniques being taught in your class. Classes usually are not geared toward learners. They are geared toward grades. The teacher has to have some measurable way to show his superiors that you are learning Chinese (or whatever subject), so the whole class is based around tests and teaching to those tests rather than basing it on what works best for the learner.

Schools have deals with publishing companies, so they have to use those textbooks, which kill the language, dissect it, and stick it under a microscope for you to study. Instead they should present the language as a living thing and allow you to experience it as it is in "the wild" (real life), and teach you the most effective ways of being able to handle the language in real life situations.

Unfortunately, unless the educational system changes drastically, this won't happen, and so the learner has to do most of the learning on his own time, his own way, rather than the school's way. Really, will you get better at Chinese by using NPCR or David and Helen than you will by watching tons of Chinese dramas, reading tons of Chinese books, listening to tons of Chinese music, and hanging out with Chinese friends? Is it more efficient to learn new characters when the book presents them to you over the course of a few years, or all at once, up front, in the space of a few months, so that you rarely encounter a character you don't know and can work instead on building real literacy and vocabulary using real sources (books, magazines, newspapers, movies, TV shows, and websites intended for native speakers)? Think about it.

[imron]

Chinese learners have been a little slow on the uptake with this mnemonic thing

Personally, I don't like the whole mnemonic thing as it just adds another layer of things you need to remember, and it isn't the way that native speakers think about their language. I can understand how some people find it useful, but to me it seems like training wheels on a bike, i.e it might useful at the very beginning, but you need to get rid of them eventually if you ever want to learn to ride properly.

I should add however that I do think it's incredibly useful to break down a character into its component parts, and remember the character like that. e.g. 安 = 宝盖头,下面一个女字. It's the whole "create a story" thing I don't go for.

[roddy]

For example, I know that any character in my flashcard deck pronounced 'zhu' (any tone) will most likely have 主 or 者 in it.

Listing characters pronounced 'zhu', any tone, in Wenlin, there are 24 within the set of characters for which Wenlin has frequency info (which goes up to about 3,000 I think, so a decent chunk). Less than half of those have 主 or 者 in them. There are also very common characters which don't follow this rule - 往, 都.

[OneEye]

Sorry, roddy. I should have clarified that better. I'm using McNaughton's book, and so far that has been the case. However, I'm not under the impression that all characters pronounced 'zhu' contain those components. But I'll learn other phonetics as I get further into the book. And actually, I think your post would be a pretty good argument in favor of the "screw the phonetics" camp, if one were to go the Heisig/Matthews/bulk memorization via mnemonics method.

imron,

The problem is, I'm not a native speaker of Chinese. And as a Westerner I see no need to think about the characters the way a Chinese person thinks about them, at least as far as learning them. I'm more concerned with learning them as quickly and efficiently as possible. And if that means adding an extra layer of info (the story), then I'm willing to do that (as counterintuitive as it may seem). And the stories do disappear pretty easily. Pretty soon it becomes "just another character" and I don't even have to think about how to write it. It's just there, as naturally as "你好."

And, most of my stories are nothing more than the names of components, and a few words to make sense of it. Nothing complicated. "Make things as simple as possible, but not simpler." - Einstein

[necroflux]

Here to second everything OneEye is saying, extremely sound advice. It's almost as though I'm reading my own experience! I started with Helsig, and now three years later still sometimes use stories for completely new characters but it's usually not necessary. And the stories do disappear after a while after the character just "is what it is".

My study method for new characters is also almost exactly the same as OneEye, a bit different in that I quiz myself by giving the english definition only, and I have to write the characters from that. So I'm testing myself on the meaning/sound/character all at once. I use a web program I made that keeps track of what I make mistakes on, if I get one wrong then I have to write it correctly twice before it's "cleared". If I get it wrong twice, then I have to get it right three times, etc. Each miss means I have to write it correctly one additional time.

95% of the time, a month later while I might not be able to write the character from memory, I will recognize it in reading.

[newyorkeric]

But that brings me to the question of how many characters each book teaches. The Matthews' book has 800. The McNaughton book has around 2300 and the Harbaugh book has 4280. Sure, no book will teach you EVERY character. You'll have to learn some on your own one way or the other, but 800 just doesn't seem like enough to be worth the money. And I don't even know if they have plans to release vol. 2 anytime soon (at least not that I've been able to find).

I contacted the publisher of the Mathews' book. He said volume 2 will be released in 2009 or later. Not such a problem for me because I plan to apply their technique on my own after I get through the 800 characters in volume 1.

[renzhe]

Personally, I don't like the whole mnemonic thing as it just adds another layer of things you need to remember, and it isn't the way that native speakers think about their language. I can understand how some people find it useful, but to me it seems like training wheels on a bike, i.e it might useful at the very beginning, but you need to get rid of them eventually if you ever want to learn to ride properly.

What I've ended up doing is trying to use components to remember meanings, and only create funny stories with characters that I absolutely cannot retain at all (we all have such characters). In these cases, it really helps me move the character into long-term memory, even if I have to replay the story often in the beginning.

But with most characters, I don't really need a big story. "hui4" with silk radical = to draw, "zong" + foot radical = footprint, etc. This sort of rough combination of sound, radical and complete character shape is usually enough for me to retain a character long enough. I do recognise over 2500, though, this may not work as well with a total beginner, where mnemonics (preferably built around the real component parts) are very useful.

But the most effective thing for me has been an insane amount of volume. If you want to learn a lot of characters relatively quickly, you need to review every day, morning, noon, evening, as often as you can, and read as much as you can. Nothing etches a character into your memory like volume, IMHO.

[necroflux]

Indeed, volume is key. Everywhere I go I look for characters in my environment to figure out their meaning, like a sickness (but a good kind ). Reading was the foundation of my native language, and so I figured it would be good to make it the foundation of my Chinese - so far it's been a great decision. But it won't pay off for years, so if you decide to go that route you really just have to stick with it through the initial months/years of utter frustration. i.e. it took me a year and a half before I started regularly encountering complete sentences I didn't need to look up words to understand.

And yes some characters are more conducive to "stories" than others. Try making a story for 喜 or 濤... I hate these characters with a passion as they don't readily break down into recognizable parts and can really only be memorized as a complete unit.

[Guest realmayo]

Certainly agree with renzhe: stories are great for those characters you can't remember by just breaking down the components ... but a waste of time if you can quickly learn the character without needing a story.

imron: your analogy is a good one: it is like learning to ride a bike with 'training wheels', on a character-by-character basis: bear in mind that for most characters, you can take these wheels off pretty soon, ie discover early on that you no longer need the story because you can write the character without it.