The Heisig method for learning sinographs
« previous post | next post »
I Used to Know How to Write in Japanese:
Somehow, though, I can still read it
Marco Giancotti, Aether Mug (August 14, 2025)
During the last thirty to forty years, two of the most popular dictionaries for mastering sinographs were those of James Heisig and Rick Harbaugh. I was dubious about the efficacy of both and wished that my students wouldn't use them, but language learners flocked to these extremely popular dictionaries, thinking that they offered a magic trick for remembering the characters.
The latter relied on fallacious etymological "trees" and was written by an economist, and the former was based on brute memorization enhanced by magician's tricks and was written by a philosopher of religion. Both placed characters on a pedestal of visuality / iconicity without integrating them with spoken language.
I have already done a mini-review of Harbaugh's Chinese Characters and Culture: A Genealogy and Dictionary (New Haven: Yale Far Eastern Publications, 1998) on pp. 25-26 here: Reviews XI, Sino-Platonic Papers, 145 (August, 2004). The remainder of this post will consist of extracts of Giancotti's essay and the view of a distinguished Japanologist-linguist on Heisig's lexicographical methods.
I recently came across a short essay about kanji—Japanese logographic characters—by a certain James W. Heisig. His point is that learning kanji presents two obstacles: remembering what the shapes mean and remembering how they are pronounced. And it is a bad idea, claims Heisig, to try learning both at the same time. Japanese children learn the spoken language first, then they learn how to write it in elementary school; Chinese students of Japanese (who tend to be pretty good at it) have pre-existing knowledge of character meanings and forms from their mother tongue, so they only have to learn how to pronounce them. Therefore, a Western learner should first focus only on the meaning and writing of those couple of thousand common characters and, only after having mastered those, should move on to studying the pronunciations. Heisig professes simple divide and conquer.
That sounds plausible, but is it really an effective approach? How can you keep so many of those tangled squiggles in your head without even knowing how to say them out loud?
The answer is yes, it works. At least, it worked fantastically well for me. The first thing I did when I began learning the language in 2006 was opening Heisig's famous Remembering the Kanji Volume 1 and going through it, one kanji at a time, using the book's mnemonic techniques to commit to memory the meaning, construction, and stroke order of all 2042 characters in the book. No thought to pronunciations, words, grammar—just the way each character is written and understood. I filled several notebooks with handwritten characters as I practiced recalling them every day, and eleven months later, I had them all in my head. I could write and understand them all with little effort and, from that point on, learning how to pronounce and compose them into words and sentences felt like a breeze. Prof. Heisig has my eternal gratitude.
Time passes
But fast-forward two decades, and the situation has evolved in an interesting way that I would never have anticipated. I spent over thirteen years in Japan, and my Japanese has only gotten better. My friends and colleagues in this period have been mostly Japanese natives, as is my spouse. I use the language every day at home, I use it to read novels and send emails, to watch South Korean shows with Japanese subtitles, and to file my taxes. I use it more than my own native language, both in spoken and written form. And yet… I cannot handwrite most of those kanji any more.
Except for a few hundred simple and/or frequently recurring characters (like those in my home address), I just cannot recall how to draw them out with a pen. I haven't completely forgotten them, and I'm perfectly capable of reading and understanding them in the blink of an eye—it's just the act of turning the intended character into ink on paper that is often impossible for me.
I'm not alone in this "character amnesia," either. Whenever I tell a Japanese native about my lost ability, they all readily admit to having forgotten how to write many kanji, too. Apparently, this is a well-known phenomenon in Japan and in China. There is even a term for it, wahpro baka (ワープロ馬鹿), meaning "word-processor idiot," from the idea that spending too much time typing into Microsoft Word makes people's handwriting skills atrophy.
Is it only me, or is all of this surprisingly deep and fascinating?
I wrote previously about the beautiful dissociation of the Japanese language, where the way you write and the way you pronounce kanji are two separate worlds with no simple one-to-one correspondence between the two. What this handwriting forgetfulness shows is that there is an even deeper separation between how our brains process the act of reading and that of writing by hand.
Indeed, neuroscience research has shown that reading activates visual-language pathways in the left hemisphere of the brain, from the occipitoparietal to the posterior temporal cortex. Writing kanji, however, is driven by our motor-planning and primary motor cortex, as well as a network in the posterior parietal cortex specialized in remembering the sequence of strokes necessary for the task.
In other words, what feels like a single, monolithic "literacy" ability is actually two distinct skills, each exercised in different instances and each capable of improving and decaying on its own. We all learn two ways to handle text, not one, although we usually learn them at the same time. Spend years typing on a phone with autocomplete, and your pen-focused neural network weakens.
Hold on. This explanation is quite convincing, but it doesn't solve the entire mystery. The thing is, I have aphantasia: I do not have, nor can I choose to conjure, images in my mind. On the surface, this atypical trait seems to explain quite well why I can draw a blank when asked to write the kanji for "plant" (植) from memory. I don't see the character in my mind, so it makes sense that I can't reproduce it on paper.
What confuses me is that other people can form images in their minds. Are all those with character amnesia also aphantasic? That can't be, given that aphantasics amount to less than 5% of the population, while a much larger number of people forget how to write (70% of teenage participants in a Chinese TV show were unable to write the word "toad"!).
How is it possible for you to "see" the text in your mind and not be able to replicate it with a pen? Even if the mental image is faint and fuzzy, surely you can sketch it out roughly at first, then refine it until it settles into its exact form? Apparently, that is not how mental images work, either.
Alphabet vs. logographs
Admittedly, I've never heard of someone forgetting how to write a letter from the Latin alphabet. Character amnesia is mostly a thing with logograph-based languages. The simplicity of letters and the much higher frequency of those symbols probably play a role—you encounter "R" in a text much more often than "plant." I wish there were more research on where exactly the line lies: how complex and how numerous do the symbols in a writing system have to be in order to create a character amnesia problem? The answer sounds important, because it would tell us about some fundamental limits to how the brain processes visual information. According to the Language Closet, researchers "found that character frequency, age of acquisition, spelling regularity, familiarity, stroke count, and imageability were significant predictors of character amnesia," but this result is partial and muddled by other, less relevant factors.
Back to the phantasia paradox: for some reason, mental images don't help significantly with writing. I think this hints at the incredible compression ability of the brain.
There is a widely accepted theory in cognitive science called fuzzy trace theory, which hypothesizes that there are two ways we encode (record) memories in the brain: verbatim traces and gist traces. The verbatim kind is what we typically think of as "good" memory—the ability to remember something in full detail, almost literally. These are relatively precise memories, but they are difficult to retrieve and easy to forget. They're not "sticky".
A gist trace, on the other hand, is the quick and very sticky transcription of only the salient parts of the experience—minus the details. They are the sublimation of sensory information into fuzzier, abstract "meaning"—a form of compression. Gist memories jump back at us more readily and are harder to forget, but they lack all the particulars of verbatim memories. And, crucially, forming gist traces doesn't depend on having verbatim traces.
When you read something, both verbatim and gist traces are recorded in your brain, but the latter leave a stronger and longer-lasting presence. In the case of 雨, your gist trace might simply read "rain / Chinese character made mostly of horizontal lines and drops". The exact (verbatim) shape may not even leave a mark in your mind until you've seen and studied it carefully several times over.
In practice, this means that you don't have a single, complete "source file" for a character or image stored anywhere in your brain. There is the abstract gist of it somewhere, with just enough features to allow you to recognize it when you see it again, but the exact details are nowhere to be found. The details may be scattered in other parts of your brain—for instance, in your motor memory networks, which allow you to write the character down—a different set of neurons that needs its own painstaking training. Reading a complex character, then, means "recognizing the gist," and writing it means "activating your memory of the precise movements needed to reproduce it." Two starkly different tasks.
This distinction applies to much more than East-Asian symbol recognition. In Reading Blood Meridian with Aphantasia, I described my experience reading a notoriously gruesome novel by Cormac McCarthy. I noted how the vivid and atrocious depictions in the book don't provoke the deep, visceral reactions that others talk about.
It feels important but remote, not something related to me personally. Kind of like observing the events from a flying bird's (metaphorical) eyes—more than far enough for objectivity, but keen enough to take it all in.
This sense of remoteness might imply that, due to aphantasia, I only get the gist of the written experience from the text, bypassing the verbatim kind. My image-less reading experience is almost pure abstraction, because there is very little visual verbatim information to be stored in the first place. I have countless powerful "memories" of that novel, but I would never be able to reconstruct even a single intense scene to any degree of fidelity.
This is also why I believe that language is a bottleneck for thought. Most of what you remember is nothing like an approximate copy of the things you experienced in real life—even in the specific case of text, memory is not even remotely like a paraphrase of previously read words. Many of our thoughts happen in a highly abstracted and distilled form, interacting and connecting with each other as a network that simply cannot be faithfully converted into a sequence of words, however long. The fact that people can fail even at something as basic as sketching a kanji or a vehicle they've seen hundreds of times before is just another example of the same phenomenon.
It turns out that the bottleneck is not only between different minds, but also between parts of the same mind. In my case, Prof. Heisig's divide-and-conquer approach worked well to create the perfect scaffolding to learn the Japanese language as a whole. But when the scaffolding was dismantled to reveal the completed structure, the ability to write by hand was thrown away with the rest of the junk.
Harry Lorayne (1926-2023), "The Yoda of Memory Training" and "The World's Foremost Memory-Training Specialist"
Selected readings
- "Character amnesia yet again: game (almost) over" (4/28/22) — with very long, essential bibliography
- "Character amnesia and the emergence of digraphia" (9/25/13)
- "Dumpling ingredients and character amnesia" (10/18/14)
- "Character amnesia and kanji attachment" (2/24/16)
- "'They're not learning how to write characters!'" (11/5/21)
- "Aphantasia — absence of the mind's eye" (3/24/17)
- "Japanese survey on forgetting how to write kanji" (9/24/12)
- "How to learn to read Chinese" (5/25/08)
- "How not to learn Chinese" (4/16/17)
- "How to learn Chinese and Japanese" (2/17/14)
- "The future of Chinese language learning is now" (4/5/14)
- "Learning to read and write Chinese" (7/11/16)
- "Learning languages is so much easier now" (8/18/17)
- "A revolution in Sinitic language conceptualization and learning" (1/7/21)
- "Characterless Sinitic" (9/1/21)
- "Characterless future" (3/9/18)
- "Sinological suffering" (3/31/17)
- "Chineasy? Not" (3/19/14)
- "Chineasy2" (8/14/14)
- "Chinese dictionary" — Wikipedia
Deficiencies
Victor H. Mair lists eight adverse features of traditional Chinese lexicography, some of which have continued up to the present day: (1) persistent confusion of spoken word with written graph; (2) lack of etymological science as opposed to the analysis of script; (3) absence of the concept of word; (4) ignoring the script's historical developments in the oracle bones and bronze inscriptions; (5) no precise, unambiguous, and convenient means for specifying pronunciations; (6) no standardized, user-friendly means for looking up words and graphs; (7) failure to distinguish linguistically between vernacular and literary registers, or between usages peculiar to different regions and times; and (8) open-endedness of the writing system, with current unabridged character dictionaries containing 60,000 to 85,000 graphs.
- "The economics of Chinese character usage" (9/2/11) — inspired by Rick Harbaugh and Apollo Wu
- "How many more Chinese characters are needed?" (10/25/16)
- "The infinitude of Chinese characters" (10/9/20) — with an extremely lengthy bibliography
- "Cucurbits and junk characters" (3/30/24)
- "Sino-Semitica: of gourds, cassia, and hemp and Old Sinitic reconstructions" (2/1/20)
- "Korean words for 'bottle gourd" (8/30/23)
- "Jichang Lulu" (9/26/22) — for rumbling intestines; the name reminds me of "hulu"
- Victor H. Mair. "Southern Bottle-Gourd (hu-lu) Myths in China and Their Appropriation by Taoism." In Chung-kuo shen-hua yü ch'uan-shuo hsüeh-shu yen-t'ao-hui (Proceedings of the Conference on Chinese Myth and Legend). Han-hsüeh yen-chiu chung-hsin ts'ung-k'an (Center for Chinese Studies Research Series), No. 5. Vol. 1 of 2. Taipei: Han-hsüeh yen-chiu chung-hsin, 1996. Pp. 185-228.
- "Goblet word" (5/30/20)
[h.t. Stefan Krasowski; thanks to Jim Breen]
S Frankel said,
August 17, 2025 @ 9:15 pm
Just to reinforce a point made by several of the writers above: Heisig's method does not claim to teach Japanese. It merely lays a foundation. The point is that when the student does learn Japanese, the characters will already be familiar.
Victor Mair said,
August 18, 2025 @ 8:14 am
Marco Giancotti's post points out the vast gulf between active PRODUCTION and passive RECOGNITION of Chinese characters.
Pamela said,
August 18, 2025 @ 9:47 am
I haven't read Heisig or Harbaugh, though this has got me very interested. How does it all relate to the system in Creel, Classical Chinese by the Inductive Method? or, to Mateo Ricci's and other Renaissance types and their "memory palaces"? It all amounts to a phenomenology of Chinese characters as a paralinguistic phenomenon, something not unrelated to language as it is spoken, but a distinct phenomenon with its own neurological and esthetic dynamics. I wonder if we are recalling the experiences of generations of missionaries and historians in the US and Europe who learned to read classical Chinese very well, but had no competence (and no wish to be competent) in the spoken language? Is it related neurologically to religious iconography like the ankh, and the aim? Does it relate to the writing systems like Dongha, or some native american writings systems/
David Marjanović said,
August 18, 2025 @ 10:40 am
I have, on a few occasions, temporarily forgotten two or three handwritten forms in Latin (and Cyrillic, after I learned Russian Cyrillic handwriting much later) – always extremely rare ones (I remember having once forgotten capital Y, which hardly occurs in German).
By "handwritten" I mean more or less what's called "cursive" in the US, except it's the only form of handwriting I was ever taught.
There are people whose every visual memory has high enough resolution to make such replication possible. They're a subset (!) of those with so-called photographic memory – a truly tiny portion of the population.
This means most people find recognizing things (or faces, or the character 雨) much easier than replicating them from memory. "I know it when I see it", not generally when I don't.
David Marjanović said,
August 18, 2025 @ 10:43 am
…and the obvious failure mode is that I, for one, easily confuse similar-looking Chinese characters if I never learned to write them: my visual memory doesn't, by default, have enough resolution to reliably tell them apart.
Brian said,
August 18, 2025 @ 1:37 pm
> Admittedly, I've never heard of someone forgetting how to write a letter from the Latin alphabet.
A better analogy would be somone forgetting how to write a _word_, despite knowing perfectly well how to pronounce it. This does happen to English speakers/writers, to a majority in fact, such that there is no specific term for it: it is considered normal. The difference, I guess, is that for most English users it happens with a smaller percentage of their vocabulary, presumably due to their being a bit more of a connection between a word's pronunciation and its written form than in kanji (not always of course, but in the aggregate).
Jonathan Smith said,
August 18, 2025 @ 2:17 pm
Re: initially approaching characters "paralinguistically" to use Pamela's term, same for the alphabet in EFL pedagogy. The task is a couple orders of magnitude smaller of course…
In the case of this author, they seem to have gone to Japan and learned the language correctly accidentally; I doubt that initial exposure to Heisig was of much help in this endeavor.
Re: the recent aphantasia hype, it's mostly pickmeism as the kids say — 1/3 of the population has not suddenly been found to be incapable of forming "mental imagery" (to use a term that may be part of the problem) in the typical human manner.
Scott P. said,
August 18, 2025 @ 5:39 pm
I feel some commonality with the way I understand foreign languages — in most cases, especially in a language I am not intimately familiar with, I will hear a sentence in that language, envisage the sentence as a written string of characters, then read that sentence in my head (and it is literally a process of reading, there are imagined letters and words that I have created), and then translate the written sentence.
I've become fluent enough in Spanish at least that I can sometimes short-circuit the process, but it still occurs more often than not. And the process is fluid enough that I can do this at normal conversational speed, it's not laborious in the slightest.
Does anyone else do this?
On a related note, it's quite common that after I've listened to a short speech, say, and translated it into English in my head as I go, that afterwards I am quite incapable of saying for certain whether the subject spoke English or Spanish when they were speaking.
Kenny Easwaran said,
August 18, 2025 @ 5:49 pm
The ampersand is arguably a character in the English language writing system, along with the 26 letters, 10 digits, and a few other punctuation marks. But it's a particularly complex one, that is written particularly infrequently. So a lot of people don't know how to write it. But I don't know if there are many people who have learned to write an ampersand at some point, but then forgotten how.
Josh R. said,
August 18, 2025 @ 6:42 pm
"Spelling amnesia" will be coming more and more for users of alphabet-based languages as the use of predictive text becomes more prevalent. Users of Sinograph-based languages have merely had a head start.
Chris Button said,
August 18, 2025 @ 8:43 pm
I can still read Japan's kana (both kinds) and Taiwan's bopomofo just fine. I pause on the rare occasion (basically never) when I have to write them.
Jerry Packard said,
August 19, 2025 @ 7:33 am
Great post! Where do I start?
First off, the writing of graphs is indeed a motor skill, but the degree of motor depends on how much motor was used in inputting them. In other words, if you learn kanji using a ‘gestalt form’ method, then writing them will not engage that big of a motor component, while if you input them with repetitive writing, then there will be a strong motor component. I can just barely visualize characters, but if I want to produce a character I will just set my hand free and put it comes. I do the same in English. When I am not sure how to spell a word I let my hand go and out it comes.
Q:How is it possible for you to "see" the text in your mind and not be able to replicate it with a pen?”
A: The same way it is possible to see a visual landscape or a bowl of fruit in your mind but not be able to produce it with your hand. Artists seem to be good at this, but it is difficult if not impossible for the rest of us.
I for one do forget how to write individual letters especially in Russian. For English my motor doesn’t help much for easily confused letters for some reason. So before I write a ‘b’ I often have to stop and visualize it or chances are I’ll write a ‘d’.
For Blood Meridian, I am only a weakly visual person but McCarthy’s descriptions definitely evoked stark visual images for me.
Re language as a bottleneck, there is a great deal of individual variation on that factor. Some folks will claim that they can do no thinking without language, and so they’d feel that language is a helper rather than a hindrance to their thought processes. Others feel that language restricts them, because they feel they have a hard time forming their thoughts within the strictures of language.
“reading activates visual-language pathways in the left hemisphere of the brain, from the occipitoparietal to the posterior temporal cortex”
That is true, but if rigorous motor is used in learning to read, then the frontal motor strip will be automatically be activated as well. Experiments using Chinese characters reveal that subjects performing completely oral tasks are influenced by the specifically written forms of characters.
Pamela said,
August 19, 2025 @ 9:19 am
Love Kenny's comment on the ampersand. I think that, indeed, must incorporate the same neurological paralinguistic issues. Acronyms might also be close, and of course it is now impossible to read social media without being able to recognize a rapid lengthening list of cliché acronyms. "pwned" originally nobody knew how to pronounce it (it looked like Welsh to me) and few knew what it meant. Most pronounce it "owned" and it just means that in a certain sense of defeating and humiliating somebody, so the perverse spelling is unneeded. What is the "p" about? Is it to avoid too-obvious references to the ideology of slavery? Is it to represent "p" as in "power"? It is also segregated from speech–even the most social media dipped podcasts avoid saying the word, even if they write it daily. I think because it is something that only has rhetorical effect when written, now when spoken.
S Frankel said,
August 19, 2025 @ 11:49 am
@Pamela: The 'p' in pwned comes from the keyboard layout. P is next to O, so 'pwned' was a common enough misprint, at least among over-caffeinated hackers, that it became an in-joke and spread from the.
Jim Breen said,
August 19, 2025 @ 5:37 pm
I've never been a fan of Heisig's approach to learning and memorising kanji. I include the index numbers for kanji in his books in kanjidic, but I've resisted adding his mnemonics.
I know some people really like his approach – it seems it's popular in France, for example. My friend Yves Maniette has published French equivalents of his books.
Roscoe said,
August 19, 2025 @ 5:41 pm
As a Mitch Hedberg fan, I was hoping the subheading to the Giancotti post would be “I still do, but I used to, too.”
Michael Vnuk said,
August 19, 2025 @ 6:51 pm
Kenny referred to the ampersand as being complex and infrequently written. Reflecting on my own learning to write at school in 1963 in Australia, I (a native speaker of Australian English) don’t ever recall being taught to write the ampersand in that year or later. At some indeterminate point, I picked up how to form an ampersand and I occasionally include it in my writing.
Besides its infrequency and curly complexity, the ampersand has another property that makes it (slightly) harder to remember. We were taught to start letters at the top left, eg a downstroke for E or M and then do the other lines. Some letters didn’t start in the top left, eg A at top middle and G near top right, but all started somewhere at the top. The ampersand starts at the lower right, which might mess with the simple motor memory developed from the other letters. (The musical treble clef has a similar pattern, starting at the base and curling around. Once I had taught myself to do it, I would use it as an occasional doodle because it is a fun thing to do, as opposed to more boring letters like T or O.)
When writing carefully, I still start letters at the top, although I will start them elsewhere if I am in a hurry or the writing is less formal, eg starting A or M at the lower right.
I had a difficulty with writing the letter S after a period of not writing, probably the long holiday break between Grade 1 and Grade 2. I remembered to start the S in the top right, but I would curve around too generously so that I was then forced to scrunch up the base of the lower curve to avoid going below the baseline. I think I worked out the correct technique after a few days.
S starts near the top right, and beginning writers sometimes form the S backwards because they start in the top left, as with most other letters. In fact, a backwards S is often a cliche for representing a new or poor writer. However, it seems to me that backwards C and G are less common than backwards S in both real and fictional examples.
Ryan Chang-Hill said,
August 19, 2025 @ 8:48 pm
Also possibly relevant is artists trying to draw Pokemon from memory, such as this: https://www.reddit.com/r/pokemon/comments/1cs6byn/today_i_drew_all_151_gen_1_pok%C3%A9mon_from_memory_in/
For most of them, the basic gist of each Pokemon's important features are present, but the details are off. I imagine that most people look at Pokemon far less often than they read Latin alphabet or Chinese characters, and certainly don't put in deliberate practice into drawing the same Pokemon every day.
I personally have less issues with forgetting Chinese characters, but that might be because I try to study them on my flash cards every day, writing out each character with my finger. I also think that I have a more than average fascination with the shapes of the characters themselves. My issue is forgetting the readings, which happens much more often than forgetting character shape details, and that is probably because I need to spend more time studying them contextualized in actual Chinese/Japanese texts.
Michael Vnuk said,
August 20, 2025 @ 1:34 am
Correction for para 3: 'starting A or M at the lower left.'
Rodger C said,
August 20, 2025 @ 9:26 am
I've seen a lot of people reverse an ampersand, turning it into a capital S in the style of cursive we were once taught.
Philip Taylor said,
August 23, 2025 @ 2:36 am
Michael — could the fact that I invariably start capital As and Ms at lower-right be because I am left-handed ?