Learning Chinese characters – Revision

Visit Hacking Chinese instead: This post about studying Chinese is partly or completely obsolete. A revised version, along with much more related to language learning can be found at Hacking Chinese. This post is kept here for the sake of consistency.

Update: This article was written a long time ago, so even if most of what I say here is still relevant, I have developed my revision strategy a lot. Check this post for some ideas.

This article is the third part of three concerning the study of Chinese. I have composed these articles for various purposes, but mostly because I enjoy writing. I also feel that I have something important to share, and even though I am sure that others have written about the same concepts elsewhere, this is how I view it. Hopefully, these ideas will be helpful to other students. Another reason for writing these articles is that I want to show those of you that do not study the language a little bit how I feel and think about Chinese. These are the three parts:

Part 1 – Attitude
Part 2 – New characters
Part 3 – Revision

Learning Chinese characters – Revision
Having read the first two posts in the this small series of articles, one might draw the conclusion that learning what a character means and how to write it is the most difficult and time-consuming part of studying Chinese. This is wrong in every repgard. The really difficult bit of studying Chinese is to have the discipline and time to revise what one has already learnt; in other words, not to forget what the first two parts of this series focused on. Since this is a problem that increases with one’s knowledge of the language, it is also something I personally feel is very important to discuss. Remembering how to write a character is a real problem for native Chinese as well, especially for those who do not write by hand as a part of their daily lives. Writing using a computer merely requires one to know how a character is pronounced and what it look like, which is not the same as being able to write it by hand.

That being said, I have used trial and error to find a suitable revision method (admittedly, after studying languages for many years, I had a good idea of what I wanted before). My original intention was to publish this article in close succession to the previous two parts, but I began altering my revision strategy at the same time, so I decided to wait until that change was fully implemented. What I present here is by no means based on any serious research, but it is founded on my rudimentary knowledge of psychology and a lot of consistent blood, sweat and tears.

Committing characters to long-term memory
If the disposable time was unlimited, it would probably be best to keep track of sets of characters (such as from a chapter in a book) separately and make sure that any given set of characters was revised occasionally. However, time is painfully limited and for reasons I will explain in the next section, I have come to the conclusion that two separate systems are needed for revising characters. What I will say about the initial phase, committing characters to long-term memory, is based on my own revision schedule, which looks like this:

Click for higher resolution image.

It is difficult to analyse this kind of data without proper software or knowledge, but basing my writing on feeling, reflection and studying of the chart above will probably get me somewhere. The blue boxes are checks from pinyin to characters using ZDT. This is by no means anything fancy, it just means that I go through all the characters in a random order and make sure I know how to write them. If I do not remember how to write them or write them incorrectly, I count that as an error and continue with the rest. After one round, the incorrectly answered characters appear again, and normally I have no problem the second time. Later, I repeat this procedure for the same chapter, noting the number of correct answers (the figures in the blue boxes). The most important question here is of course how often to do this in order to optimise retention.

If working with simple repetition (which is a fairly poor sort of revision anyway), the human mind works in a rather predictable pattern. The most important thing is to find out what that pattern is like for a given learning situation. I have found out a couple of truths about how my brain works when it comes to revising characters, and I am pretty sure than most of them are applicable to other people in other situations as well. Here is what I have found out about revision intervals:

As I have already stated, I do the initial learning in the evening, and the first revision session is directly afterwards. Once I have studied each and every character carefully (as described in the second article), I simply read through the list of characters, refreshing my mnemonics and making sure I know what the various parts mean. This is often done very quickly.Then I go to sleep. Immediately upon waking (or very soon thereafter), I carefully go through the list again, using a pinyin to character test again. This is not done in a hurry, but I actually make sure that I can write all the characters. Usually, I am able to remember between 80 and 85 percent of the characters from the evening before.

Then I do not touch the characters for the remainder of the day, or if I do, I just quickly skim through the list on my mobile phone waiting for the bus and suchlike. The next revision session is the next day (the third day since, counting the initial learning as the first day). It does not seem to matter when during the day I do this, but it has to be done the third day. After than, I should revise again on the fifth and eighth day, which usually leaves me with a score seldom below 95 % and often close to 100 %. Even if i do not revise further, I can usually remember how to write more than 95 % of the characters several weeks after the initial learning (see the first article for a discussion about how good is good enough). If I realise that I have a problem with specific characters even after this time, I put them in a separate category so as to be able to practice them without having to go through the entire chapter that contains them. This is obviously saves a lot of time.

In summary
Day 1, initial learning in the evening
Day 2, revision in the morning
Day 3, revision
Day 5, revision
Day 8, revision

Future revision and retention of characters
Even if the aforementioned procedure leaves me with a satisfactory 95 % knowledge of the characters in any give chapter, even weeks after the initial learning, this does not mean that the understanding is permanent. If characters were left unused and unrevised, I have no idea what the curve showing loss of writing ability to time would look like, but I am pretty sure it would be an unpleasant affair, so let us instead make sure that they are not left unrevised.

As I mentioned in the first article, it should also be noted that the goal is very important. Is it necessary to have 95 % retention or is 90 % or even 85 % satisfactory? The relationship between percentage of correct answers and time spent on revising is exponential, so think carefully about this one.

Before I go on to explain why I think a computer is an essential and powerful tool to study Chinese, I will discuss the problem that faces all those who study foreign languages, namely that of effective revision (so far I have only dealt with intervals, not how one goes about the actual revision). Simply using a text book to revise is abysmally ineffective, because it means that too much time is spent on revising vocabulary that does not need revision. Imagine, for instance, a chapter where I know 95 % of the characters and need to revise 5 %. Studying the entire chapter is a waste of time, time that could be spent on revising other characters or learning new ones.

This is where the computer comes in handy. As I have already said, I use ZDT for this, but I do not doubt that there is a plethora of programs out there on the web capable of doing the same thing. I will use ZDT as an example simply to explain what I mean. The key feature is what is called an interval filter. Running any of the various flashcard checks in the program, this filter adds X points to a flashcard each time the user gives a correct answer, and deducts Y points every time an incorrect answer is given. X and Y are of course parameters set by the user. Furthermore, there are parameters for every given value of the sum X and Y, in other words, the current score for a certain flashcard (a high value meaning that correct answers are frequent). These parameters govern how often a flashcard is shown if the interval filter is enabled.

Why is this so marvellous? Because, provided that the interval filter is set up correctly, it means that it is possible to only focus on the characters that really need revision. If a character seems to be easy (many correct answers), it will appear less and less frequently, until at last almost never. If the character is difficult, it will keep appearing very often until the number of correct answers increases. Likewise, if an old character is forgotten, its score will tumble and it will appear more frequently until it is mastered once again. This is impossible, or at least extremely impractical, to do without a computer. It will save a lot of time, even including the fact that it takes time to key in the characters in the first place (please note that wordlists from many text books are provided by users, including myself, on the ZDT page).

That being said, simple repetition is a poor revision method. It will work for many characters, but if I feel that some character is refusing to lodge itself in my long-term memory, I will change method for that character. Here are some ways of learning difficult characters:

– Create a sentence of your own
– Find a better mnemonic
– Draw a picture
– Count the number of strokes
– Check the various parts again
– Check etymology
– Look at other words including the character
– Read the source text

The only thing that is required apart from a system like this is diligence over time, which of course is extremely difficult, but which is something I will not even try to approach in this article (even though I have written about related topics here and here).

Revision in context
Perhaps this ought to be the topic for an entirely new article, but I feel that my writing about learning Chinese is beginning to spin out of control, so I will try to keep it short. In essence, the best (and perhaps only) way to learn a language is to follow up studying by revision in context. By this, I mean that it is necessary to encounter the characters or words in context a couple of times before being able to use them properly. Simply studying a text book and grammar will never instill the feeling for the language which is necessary for a high proficiency. Obtaining this feel is only achievable through reading and listening a lot. And when I say a lot, I mean a lot. I am sorry, but there are no shortcuts, no smart strategies or other ways to help; spending perhaps thousands of hours immersed in the language is the only way.

By way of conclusion, I would like to invite everybody to some sort of discussion. Have you thought about the topics for these three articles and have you related them to your own situation? Have you some experience you feel is relevant for what I have said or do you just want to voice your opinion? Perhaps you even have pertinent scientific material or references to share? Whatever it might be, please, go ahead and post a reply.

Tags: , , , ,

  1. thark’s avatar

    That points-based flashcard thing sounds absolutely ingenous. I would have killed for something like that back in the day. :-)


  2. Svante’s avatar

    Det enda jag kan säga är att det hela gör mig lite smått sugen på att lära mig språk igen. Jag borde bättra på tyskan, och så vore det ju trevligt att lära sig franska. Någon dag …


  3. little_mike’s avatar

    Great!!! Thank you very much!!! Excellent contribution, definitely I will be able to learn faster.
    Just to let you know, there is another program, which is very nice too, it’s called Anki. I don’t know why, but there are the book 1 and 2 published there, but not the 3rd and 4rd.



    1. Olle Linge’s avatar

      Hi! This post is very old and even though I think most of what I say still holds true, what you point out is an extremely valuable shortcut. I’m in fact already using Anki and I’ve written about it here in greater detail: http://www.snigel.nu/?p=6283