The miracle of reading and writing Chinese characters

« previous post | next post »

We have the testimony of a colleague whose ability to write Chinese characters has been adversely affected by her not being able to visualize them in her mind's eye.  See:

"Aphantasia — absence of the mind's eye" (3/24/17)

This prompts me to ponder:  just how do people who are literate in Chinese characters recall them?

Of course, now that computers and cell phones can write characters for you, there's no longer a need to remember all of their strokes nor even their overall shapes:  just enter the sounds of the characters and your electronic device will convert them into characters.  You still have to be able to recognize that you got the right characters through this process, but at least you don't have to produce all of the strokes in the proper order and configuration from scratch.

But there is still a need from time to time to write characters by hand.  Furthermore, usually after an initial period of reliance on phonetic annotation (Romanization, kana, bopomofo, etc.), whether in China or elsewhere, if you want to become literate in characters, you will have to go through a period of practice writing them (and that includes the dreaded dictée exercises and tests).  Somehow or other, you have to work the intricate shapes of the characters into your memory and store them there sufficiently securely so that you can write, or at least recognize, them when necessary.

Two nights ago at a dinner party, I did a simple experiment in an effort to determine whether individuals who are literate in Chinese characters actually visualize them in their mind's eye.  There were about ten people present, half of whom were native speakers of Mandarin and half fluent speakers of Mandarin who had learned it as a second language.  All of those who were present are highly literate in Chinese.

I asked everyone to close their eyes and attempt to visualize the character 德, which I identified through the usual method:  "dàodé de dé 道德的德" ("the dé ['virtue'] of dàodé ['morality; ethics']").  I then asked them if they had actually "seen" the character 德 in their mind's eye.  Their replies were "not really", "sort of", "not clearly", "no", and so forth.

The question then becomes, if highly literate individuals are not clearly visualizing the character 德 in their mind's eye, how can they reproduce it when writing or recognize it when reading?  If we weren't at a dinner party and I had a lot more time to pursue the experiment with those who were present, I would ask them to attempt to visualize many other characters and explain to me how they recall the characters when reading and writing.  Since that wasn't possible last night, I simply interrogated myself, attempting to be brutally honest about how I write a character like 德.  Reading 德 is much easier than writing it, since in reading it the full form of the character is right before your physical eyes, whereas when you write it, you're starting with a blank space on a flat surface.  Consequently, in what follows, I will concentrate on describing how to write 德.

Here's what happens with me when I try to visualize 德 in my mind's eye.  Whether my physical eyes are open or closed, I can't really see the character 德 as though it were typed or written out.  If I try very hard, I can force myself vaguely and hazily to retrieve from my memory groups of strokes (⼻ ⼗ ⽬ [on its side] ⼀ ⼼) blurrily, but I cannot force my mind's eye to see them all together clearly as one neatly formed character.

Since I cannot visualize the character 德 clearly in my mind's eye, how then can I write it?

Here's what I do:

1. I want to write the character that is pronounced dé in Mandarin and means "virtue".

2. Since I've written it thousands of times, I know that it begins with ⼻.

3. I start the 德-writing motor.

4. Then, in a rush, yī kǒuqì 一口氣 ("in one breath"), I just let the four other components FLOW forth sequentially, until I reach the last stroke, the dot at the bottom right, after which the 德-writing motor shuts off.

Mutatis mutandis, I do the same thing with all the other characters that I am capable of writing:  jiǎng 講 ("talk"), jiāng 疆 ("border; frontier"), tīng 聽 ("hear; listen"), tái 臺 ("platform"), shùn 順 ("go along in succession"), and so on.  Shùn 順 ("go along in succession" — see here for about 40 other translations) is an apt character to focus on in this context, since it conveys well the sense I have when writing a character that it has a certain CONFIGURATION or PROPENSITY that you have to follow along with to successfully complete it.

This reminds me of a book by a French philosopher named François Jullien titled La Propension des choses: Pour une histoire de l’efficacité en Chine (Seuil, 1992); translated as The Propensity of Things: Toward a History of Efficacy in China (Zone Books, 1995).  The book is a disquisition on shì 勢 ("configuration; potential; tendency; momentum; inertia", etc.), which is a key concept in The Art of War:  Sun Zi's Military Methods (Columbia University Press, 2007).


"Victor Mair on the Art of War" (8/7/08)

It's no wonder that Jullien engaged in a celebrated contretemps with the Swiss Sinologist, Jean François Billeter (Contre François Jullien [Allia, 2006]).  Although the debate engaged with larger philosophical issues, Billeter's emphasis on calligraphy (L'art chinois de l'écriture [Skira, 1989]) and Jullien's recurrent attention to shì 勢 ("l’efficacité") and other topics related to calligraphy ensured that it had inescapable implications for Chinese writing.  (See the Afterword below for additional details concerning the controversy between Billeter and Jullien.)

To return to the matter at hand:  to write a given character, I start at the top left (or top, if there is no top left), then work my way from top left or top to bottom right.  To write a character correctly, you have to know the total number of strokes and their proper order.  For most people, once you get started writing, you just have to keep going — almost as though you are in an automaton-like state.  You can't think about what you're doing, what the next stroke is.  If you start thinking about the next stroke, or if for some reason you miss / mess up a stroke, you usually fail to complete the character.  That is why you see so many scribbled out characters in handwriting (like this).

I have often reiterated that writing Chinese characters is a highly neuro-muscular act.  You have to etch the characters into your nerves and muscles so that you simply do not have to think about how to write them.  That is why Chinese children — hour after hour, day after day, night after night, week after week — write the characters hundreds and hundreds of times.  If you want to be literate in Chinese (to be able to handwrite Chinese characters), you eventually will have to do this too.  And keep doing it.  If you only want to be able to read Chinese, and not necessarily to write it by hand, then read as much of it as you can, with phonetically annotated texts when they are available. Now there are many wonderful electronic learning tools like WenlinABC, Pleco, and so on to ease and speed look-up of unfamiliar terms.

How to write Chinese characters? — practice, practice, practice!

How to read Chinese character texts? — read, read, read (while looking up as efficiently as possible unfamiliar terms).

There's no magic bullet for learning to read and write Chinese characters, and you don't have to be particularly smart to do it.  But you do have to be willing to invest huge amounts of time in practicing them.

"Good good study; day day up" (1/14/14)

A final observation

Some of the most brilliant people who start out to learn one of the Sinitic languages think it's not worth all the time and effort to master handwriting of Chinese characters.  For them, acquiring fluency in handwriting characters is neither fun nor a miracle; they view it as drudgery.  On the other hand, some artistically inclined individuals who know no spoken Chinese and can neither read nor write Chinese texts, take great pleasure in practicing Chinese calligraphy as an art form.

Different strokes (bǐhuà 筆畫 / 笔画) for different folks.




François Jullien's reply to the charge that he portrays China as "an alterity" appears in Chemin faisant, Connaître la Chine, relancer la philosophie. There he argues that the unreferenced quotations used by Jean-François Billeter are fabrications and that Billeter attempts to construct an imaginary version of François Jullien's work to argue against. The crux of the matter for Jullien is that exteriority and alterity are not to be conflated. China's exteriority, Jullien's point of departure, is, he argues, evident in its language as well as in its history, whereas alterity must be constructed and, as internal heterotopia, is to be found in both Europe and China. Rather than relegate China to a separate, isolated world, Jullien claims to weave a problematics between China and Europe, a net that can then fish out an unthought-of (un impensé) and help create the conditions for a new reflexivity (réflexivité) between the two cultures.

Jullien has dealt with the question of criticizing Chinese ideology several times in his work: La Propension des choses, chapter II; Le Détour et l'accès, chapters I to VI; Un sage est sans idée, final pages; etc. He thus separates himself from those who, out of fascination with strangeness or exoticism, have upheld the image of China as an "other." He separates himself also from those who, like Jean-François Billeter, permit themselves to dip into a "common fund" of thought and thus miss a chance to benefit from the diversity of human thought, which for Jullien is its true resource. He argues that we must reject both facile universalism (which springs from ethnocentrism) and lazy relativism (which leads to culturalism) in favor of a "dia-logue" of the two cultures: the "dia" of the écart, which reveals the fecundity of multiple lines of thought, and the "logos," which allows these lines to communicate through a common intelligence.

For a collective reply to the criticism of Jean-François Billeter, see Oser construire, Pour François Jullien, with notable contributions from Philippe d'Iribarne, Jean Allouche, Jean-Marie Schaeffer, Wolfgang Kubin, Du Xiaozhen, Léon Vandermeersch, Bruno Latour, Paul Ricœur, and Alain Badiou.


  1. Brian Spooner said,

    March 26, 2017 @ 10:44 am

    Though with little knowledge of Chinese, I find much of great interest in this post because of a longstanding interest in the history of writing in various languages and scripts. In Iran I was always fascinated and puzzled by seeing Persians read Persian much faster than I could read English. I organised a series of meetings (with my colleague, Bill Hanaway) to try to understand this. Persian did not become a regularly printed language until the first half of the 20th century. Before that it had gone through a number of changes of writing style, which varied to some extent according the purpose (e.g. naskh, ta`liq, nasta`liq, shekasta). We finished up writing a book on how to read bad or careless handwriting (Reading Nasta'liq: Persian and Urdu Hands from 1500 to the Present). We didn't talk about "the mind's eye," but we were thinking in those terms. Our answer was that literate Persians (and Urdu-writers) think, see and act much less in terms of individual letters (than we do, and than we are taught to do as we learn to read and write). They see and write pen-strokes, each of which may contain 1, 2, three or even four letters, possibly (though rarely) more. The letters are written differently according to where they come in the pen-stroke and the pen-strokes are written differently according to what comes before or after them. Readers read pen-strokes (the number of which is very large), not letters. They read faster because they are scanning by series of pen-stokes. We also tend to read English by scanning words rather than reading letters, but the Persian reader is much more efficient and faster at then. The general approach and relationship to writing is different from ours. Its practice is different. And its result is different. There is much more to be said about this. But enough for this comment.

  2. B.Ma said,

    March 26, 2017 @ 11:42 am

    Thanks for reporting on this little experiment.

    I tried it on myself with similar results. I do manage to visualize the complete character in my mind, but I must first visualize it being written very rapidly, like one of those animated stroke order guides that online dictionaries sometimes have.

    So I would tend to agree that writing characters is a form of **procedural memory**, like brushing teeth or tying shoelaces. I also find a similar phenomenon with English spelling. If I try to imagine an English word in my mind I also need to visualize it being spelled out very quickly or at least scan from left to right along the word.

    This is despite Brian Spooner's comment that reading is about recognizing shapes, which I also agree with in both Chinese and European languages – yet I can't visualize the shape of an English word in my mind's eye just like (as Victor has demonstrated) I can't immediately visualize the shape of a Chinese character without thinking about the order in which to write it.

    I wonder if there have been any studies of Chinese people who have suffered some sort of brain injury. If they forget how to brush their teeth, but can still name objects, does this mean they become unable to write characters, but are still able to read them? Sort of like Broca's aphasia but for reading and writing?

  3. leoboiko said,

    March 26, 2017 @ 11:56 am

    In the Western calligraphic world, the word for the gestural, procedural basis of a character is ductus. Anthropologist Tim Ingold once compared writing to dancing: each character is a tiny sequence of dance steps with your hand (now streamlined to minimalistic, modernistic dance-steps with computer keyboards and cellphone touch-screens, but still made from procedural finger-dancing). This is true of all writing systems.

    We receive speech with our ears, and produce it with the vocal apparatus; but we receive writing with our eyes, and produce it with hand gestures. Nunberg once called writing, very felicitously IMO, an "application" of the principles of language. I wonder if it isn't, more specifically, an application of the capacity for sign language—hand gestures and sight.

  4. Bob Ladd said,

    March 26, 2017 @ 12:58 pm

    Procedural memory of this sort also applies to typing or keyboard entry in any alphabetic writing system; it's not unique to Chinese characters. Exactly as VHM says, you just sort of start and motor memory takes over. The effect is most obvious when you are trying to type a less frequent word that involves a sequence you type all the time. I find it very difficult to type the word glad with only one D, for example, because the L-A-D-D-space sequence is so well rehearsed in my own keyboard use. I also recall an occasion on which I had to type the word morphine a few times in an email and kept finding morpheme on the screen when I had finished.

  5. Andreas Johansson said,

    March 26, 2017 @ 2:40 pm

    Is there any reason to assume that there's anything special with Chinese characters here? Latin letters are simpler, and more directly tied to phonology, but the basic mental processes for recognizing and reproducing them are presumably the same.

    Before reading the couple pieces here on aphantasia I would probably, if the issue was brought to my attention, have assumed that being able to visualize Latin letters was a prerequisite for writing them by hand (in a non-school setting where you can't just copy them from the blackboard).

  6. Ran Ari-Gur said,

    March 26, 2017 @ 4:34 pm

    Reading in the Brain, by Stanislas Dehaene, talks about the relationship of motor memory and visual memory to reading and writing. He points out that young children often accidentally write mirror images of the correct text, and argues that we mainly manage to distinguish forward writing from mirror-image by tapping into our motor memory skills. (In essence, he argues that our visual memory provides left-right symmetry, and that we have to learn to suppress that for text.) If so, then this supports Bob Ladd and Andreas Johansson's comments here, that this is not specific to Chinese text. (Dehaene's book is mostly about alphabetic scripts, though he does touch on Chinese.)

  7. Victor Mair said,

    March 26, 2017 @ 4:40 pm

    Writing words in alphabetical scripts and writing Chinese characters are very different processes for me. As described in detail above, characters are discrete, integral, square-shaped clumps / clusters of strokes that I have to control as a whole and power through from beginning to end to get right. The overall arrangement and composition of the strokes and components are extremely important. The individual strokes, and even the component groups of strokes, are not phonetically operative as I complete the writing of the character.

    When I spell words with letters, I can (and often do) stop to reflect and retry, all the while sounding out the letters and syllables in my mind in a linear fashion. When I write out words on the blackboard for my students, especially relatively long and complicated ones (e.g., "(a)estheticism", "otorhinolaryngological", "microscopy", "morphosyllabic", etc.), I will try to write it off in one fell swoop (often sounding it out in my mind as I go), but if it doesn't feel right, I will go back and sound it all out again carefully, syllable by syllable and even letter by letter; in really tough cases, I will step back from the board and look at the whole word once more, pronouncing it as I go from left to right, from the beginning to the end of the word.

    That is not at all how I write Chinese characters. When I miss a stroke or two of a character, I usually have to give up, in which case I'll probably write the syllable in Pinyin until I get a chance to look it up.

  8. AntC said,

    March 26, 2017 @ 5:21 pm

    @Victor I have often reiterated that writing Chinese characters is a highly neuro-muscular act.
    @Bob Ladd … typing or keyboard entry …motor memory … a sequence you type all the time

    Might I compare playing a complex piano piece from memory? (Say, a Beethoven sonata.) I do not visualise the notes on the page. I can hear the music in my mind's ear. I do remember "clumps/clusters of [notes] that I have to control as a whole". So if I get interrupted, I can't just restart where I left off, I have to go back to the start of a passage.

    And again you break down musical phrases into "sequences you [play] all the time", like a scale or arpeggio. So as Bob describes, playing 12-tone pieces needs you to 'unlearn' those sequences.

  9. Rubrick said,

    March 26, 2017 @ 6:04 pm

    I neither speak nor read Chinese, but as a child I did learn to write some characters of about the same order of complexity as Chinese characters — specifically cartoon characters. For example, I spent a great deal of time memorizing the sequence of strokes that would yield the face of Fred Flintstone. There's even a (very approximate) parallel to the sharing of sets of strokes among different characters. Nearly all Hanna-Barbera characters, for example, include the nose-plus-jowl-area "radical".

    I'd be curious to know (via fMRI or such) the extent to which such a feat is actually equivalent to learning to write a Chinese character.

  10. John Rohsenow said,

    March 27, 2017 @ 2:06 am

    @Victor I have often reiterated that writing Chinese characters is a highly neuro-muscular act.
    I have always insisted to my beginning students that learning to write Chinese characters is a muscle-memory act like tennis, and make them "write" the characters in the air with their fingers while I am writing on the board, at least at the beginning.
    I seem to be noticing more that Chinese friends here in the USA, often seem unable to remember how to write many characters when I ask them to, even though they continue to be able to read them. I take this to be due to lack of practice in writing, especially b/c they do most of what little "writing" in Chinese they do by using a western keyboard for entry even though the output is in characters.

  11. tangent said,

    March 27, 2017 @ 5:24 am

    Brian Spooner, that's something I'd never much thought about, that one language might be quicker than another to read the same content (defining "the same" is a a bit of a trick). But sure, why should it not be.

    We (some of us) have a tendency to take it as almost an axiom that languages are equivalent in what they can accomplish, but that's probably a stronger assumption around capacities than around speed.

    Anyone aware of formal measurements of reading speed like this? Or writing for that matter.

  12. Victor Mair said,

    March 27, 2017 @ 6:20 am


    Reading speed is one thing, but how much you absorb, understand, and retain is another.

  13. william holmes said,

    March 27, 2017 @ 11:19 am

    In the background here, I think, is the encounter between the venerable Chinese written language and the upstart of digitization. On which, two observations:
    — "Use it or lose it", viz.: with digitization, the need to write characters from memory is fading. Has anyone seen data on "writing literacy" levels among schoolchildren, or under-30s? (How essential is "writing literacy" on the 高考 (mainland university entrance exam)?
    — On the other hand, and perhaps counterintuitively, won't digitization extend the longevity of the character system beyond its otherwise normal span?

  14. John Roth said,

    March 27, 2017 @ 1:40 pm

    This reminds me of the ancient story of the fox and the caterpillar.

    One day, the fox was ambling along and noticed a caterpillar crawling along a leaf. He asked: "Hey, how do you coordinate all those legs?"

    The caterpillar answered: "I don't know?"

    "Well", the fox said, "which foot do you pick up first?"

    The caterpillar pondered it for a while. Then a bird swooped by. When the caterpillar noticed it, he tried to flee, but his feet got all tangled up and he couldn't move.

    The bird chirped: "Thanks for the snack."

  15. tangent said,

    March 27, 2017 @ 2:10 pm

    Sure, they're different, but it's not a vacuous question to ask: can one text be surface-read faster than another, even if storage into longer-term memory is limited.

  16. Victor Mair said,

    March 27, 2017 @ 5:40 pm

    From Robert Harrist, a specialist on Chinese art history, especially calligraphy:

    I was thinking about musical memory, which someone in fact mentions. Back in my music student days at Indiana, I could play fairly long oboe pieces by memory, but I never did so by visualizing the scores, and indeed, to write out a score (rather simple with the single line of the oboe part) I would have had to play a few measures, write them down, play some more, write, and so on tediously to the end of the piece. Perhaps some musicians at much higher levels of achievement can write out scores of memorized pieces without resorting to this procedure.

  17. Suzanne Kemmer said,

    March 27, 2017 @ 8:19 pm

    An interesting article about aphantasia:
    I once had a colleague who did not experience visual images. I asked him how he understood his dreams – did he not see things in his dreams? He said no, he didn't see anything, but his dreams had plots (strange ones as in any dreams) and he was aware of himself in the dream, moving and carrying out actions. He could also perceive spoken language as well as speak in the dream, and could 'hear' non-linguistic sounds like screaming, laughing, etc. So he could experience motor and auditory imagery, just not visual imagery (with the word 'imagery' here referring to internal simulation of perception from any of various perceptual modalities). I have heard from cognitive psychologists working on visual perception that probably 5% or more of the population lacks visual imagery; but they don't seem to suffer any particular disability from it. Maybe learning to read Chinese characters as an adult might be an exception!

  18. Dave Cragin said,

    March 27, 2017 @ 8:26 pm

    When I first started learning Chinese, I tried to use a paper dictionary to look up characters. This was virtually impossible. In millisecond it took for my eyes to move from the page to dictionary, I had already forgotten the character; I had no ability to visual the character.

    It was somewhat akin to rock climbing for me. Before learning rock climbing, I would look at a rocky cliff and see just the rocks. After doing some rock climbing, I would automatically look for hand-holds/foot-holds.

    Similarly, over time with Chinese, I was slowly better able to retain characters in my mind’s eye, albeit it, I’m much better at recognizing them in context when I type pinyin as opposed to reading them when I don’t know the context.

    @William holmes – Will digitation extend the longevity of the character system? I’d say “yes,” particularly because of the coupling with pinyin. Pinyin allows the use of a standard keyboard for typing characters. Pinyin works superbly in this way.

    Also, even Microsoft Word will provide the pinyin for a character (i.e., you don't need to buy separate software – just standard word processing software). Technology has completely changed the learning experience — for the better.

  19. Quinn C said,

    March 27, 2017 @ 11:43 pm

    I'm not convinced that motor memory is essential for alphabetical reading – I was a fluent reader before I could write much by hand, and preferred a typewriter early on because my hands were clumsy. I also never made many spelling mistakes, which tells me that I wrote from visual memory of words rather than from sounds. I guess there are huge individual differences in this respect, with dyslexia being one possible outcome at the other end of the spectrum.

  20. Dan Lufkin said,

    March 28, 2017 @ 9:56 am

    I have a small benign brain tumor that occasionally interferes with my speech for periods of a few hours. I can read aloud and recite from memory perfectly well but spontaneous speech comes out garbled. A speech pathologist coached me to imagine what I want to say as printed on a wall and just to read it from the wall. This sounds bizarre but, with practice, it works very well. My speech is a bit slower than normal but I can stay in control and speak coherently.

    After reading all the above, I'm deeply thankful that my native language isn't Mandarin.

  21. doublev said,

    March 29, 2017 @ 2:31 am

    I can't image long English word too…

  22. Andy Wilson said,

    March 29, 2017 @ 2:45 am

    I use the Heisig method to learn Simplified Chinese Characters. It is brilliant. The book(s) help you construct a story for every character to help you remember how to write it. This method doesn't require you to do loads of practice, simply to remember the story, although I have written plenty of characters in sentences and I feel what you say about muslce memory also has some truth. For me, the story has dropped away for the characters I know well and I simply write it in a rush, letting my muscle memory do the work as you suggest.

  23. David Marjanović said,

    March 30, 2017 @ 4:59 am

    I just used a combination of procedural memory and visually imagining the next stroke to recall a six-stroke character I haven't written in years and read in weeks or months…

    I used procedural memory alone to confirm that otorhinolaryngological is (ideally) three strokes in my handwriting: I stared at the screen while actually moving my hand in the air under the desk.

    Words that I hear or want to say often appear black-on-white in my visual imagination. (Russian in Cyrillic; Mandarin in Pīnyīn.) Fun fact: my native dialect isn't written, and its sound system is different enough from Standard German that the spelling conventions for the latter often can't be ported to the former without generating misleading results. I end up visualizing shorter sequences, and less clearly.

    I don't have photographic memory. My visual imagination is just as imprecise: anything out of imagined focus is basically not there. People in my dreams don't necessarily have any particular face!

  24. David Marjanović said,

    March 30, 2017 @ 5:01 am

    Oops, forgot:

    I use the Heisig method to learn Simplified Chinese Characters. It is brilliant. The book(s) help you construct a story for every character to help you remember how to write it.

    Remembering hundreds or thousands of stories sounds even harder to me than remembering the characters more directly!

    (But of course this, too, is individual variation. I couldn't use Cicero's method to memorize a speech either.)

  25. william holmes said,

    March 30, 2017 @ 9:19 am

    Victor having inspired a very long chain with just one character (德), I wonder if he might outdo himself with another one or two, or three. For example, at a future dinner party, see how many people can visualize 道, or 順. Based on various comments there seems to be a range of learning/memory tools used more or less successfully for learning and retention — with success probably varying widely across characters. Query whether the serious calligrapher, when preparing to write 羲 (for example), can see it in the mind's eye.

  26. Brendan said,

    April 2, 2017 @ 3:38 am

    @Andy Wilson

    I also used the Heisig method, and really took to it. It took me about two years to learn around 3000 characters, and I have found anecdotally that my recall is much better for mid-frequency characters such as 擅,骄傲 etc., even compared with native speakers. I think this method is much better suited to people in non-Chinese environments, who can't keep up muscle memory practice.

    The only possible downside is a reduction in speed of writing (writing 我,你,时候 are much faster, as they are in my muscle memory). However, being "corrected" by a Masters educated Chinese speaker that I had written 冷 incorrectly (she thought it was with 今 as the phonetic component) cemented in my mind that the traditional way of learning characters would not work for learners coming from alphabetic written languages.

    On this, does anybody know of any research into the use of visual mnemonics in the SL Chinese classroom?

RSS feed for comments on this post