Why are Japanese still using kanji?
« previous post | next post »
The Koreans and Vietnamese got rid of them within the last century, even the Chinese — for more than a century — seriously considered abolishing the sinographs, and have simplified them until they are but a pale remnant of what they used to be. Moreover, after WWI, when — with the help of the American occupation — Japan had a real chance to switch to an alphabet, the Japanese, on the whole, still clung to the kanji. This is not to mention that the first great novel in an East Asian language, The Tale of Genji (before 1021 AD), which has a stature in Japan similar to that= of Shakespeare in the United Kingdom (Sonja Arntzen), was written by Lady Murasaki in the phonetic hiragana syllabary (aka "women's writing").
The fact that the Japanese still have not abandoned the archaic morphosyllabic / logographic script is a conundrum that has puzzled me since I first learned Chinese and Japanese more than half a century ago. Such a fundamental question about the history of East Asian writing is one that could scarcely escape the attention of rishika Julesy. Here is her video about this thorny matter, "Why Kanji Survived in Japan (But Not in Korea or Vietnam)" (22:25). I am confident that, as always, she will have something enlightening to say about this perplexing subject.
Well, this was the longest and arguably the most challenging video for Julesy to make, but, as usual (always?), she nails it. In the first part of her presentation, Julesy dives deep into the linguistic and practical side of the problem. In the second (relatively shorter) part of the presentation, she looks at the cultural and emotional aspects of the three countries' (Vietnam, Korea, and Japan) attachment to sinographs.
As usual, Julesy inevitably says something near the end that I disagree with. In this case, she faintly suggests that it might be nice if Korea and Vietnam welcomed back some of the sinographs so they could better understand the background of many common words in their languages. I don't think there's a snowball's chance in hell that will happen — unless China resubjugates Vietnam and Korea. In the current global order and economy, that doesn't seem very likely either.
One biggie that Julesy has yet to tackle, but is related to several of her previous investigations, including this one, is how English has factored in East Asian language development during the past century and more. Knowing Julesy's scientific bent well enough not to expect her to predict the future, I'd be happy enough just to hear her analysis of what has already happened with English in East Asia during the last century and more.
Selected readings
- "'The insane effort to preserve this ancient script'" (10/2/25) — with Julesy bibliography
- "Yet another sinographic stumbling block for Chinese modernization" (11/13/25)
- "Character amnesia and kanji attachment" (2/24/16)
- "The Heisig method for learning sinographs" (8/17/25)
- "Japanese survey on forgetting how to write kanji" (9/24/12) — with long bibliography
- "More katakana, fewer kanji" (4/4/16)
Laura Morland said,
December 5, 2025 @ 6:16 pm
And… she did!
Julesly published this video a little while ago, and so I don't recall all the details, but as I recall, one important factor is that Japanese has FAR fewer homopohones than does Vietnamese and Chinese.
Another is that neither hiragana nor katakana inserts spaces between words, and so kanji provides the visual aid of indicating new words.
Hope my memory is correct….
Scott P. said,
December 5, 2025 @ 6:47 pm
Laura,
Do you mean Japanese has MORE homophones?
Neil Kubler said,
December 5, 2025 @ 8:13 pm
If two Japanese people can understand each other when speaking (which they with few exceptions obviously can), it stands to reason that if they wrote more or less as they speak, then they could write everything in hiragana, katakana, or romanization and be understood. Kanji are not needed. The same logic applies to Chinese.
Victor Mair said,
December 5, 2025 @ 8:17 pm
Thank you for your penetrating percipience, Neil.
Jonathan Smith said,
December 5, 2025 @ 11:03 pm
Re: first two comments, the extent of homophony in a language (a word problem) is not the same as number of permissible syllables (a syllable problem). Being natural languages, all of Japanese, Korean, Chinese and Vietnamese have minimal/manageable homophony and homophony per se thus has zero bearing on the question of retention vs. retirement of Kanji.
Relatedly, it's been suggested in LL comments (see e.g. threads entitled Homographobia and Hmophonophobia) that in Japanese and/or Chinese, Kanji enable a written idiom wherein texts are not readily understandable to a listener. This is possible in theory but I very much doubt any such (serious! modern!) texts exist in real life; counterexamples welcome.
Chris Button said,
December 6, 2025 @ 12:08 am
The Tale of Genji is probably not the fairest comparison.
According to Frellesvig (2010), the lexical frequency of Sino-Japanese loanwords in the text is only 4.8%. The text frequency is higher at 12.6% but still paltry.
Tangentially, I just finished reading Gari Ledyard's amazing book The Korean Language Reform of 1446. His discussion about the role of 'phags-pa is truly fascinating. But it comes with a crucial caveat on page 437:
"Nothing would disturb me more, after this study is published, than to discover in a work on the history of writing a statement like the following: 'According to recent investigations, the Korean alphabet was derived from the Mongol 'phags-pa script…'."
And yet, according to Robert Ramsey's comment on a recent LLog thread, the proposal clashed with certain nativist views in Korea.
AG said,
December 6, 2025 @ 7:15 am
A selfish and non-linguistic comment: The Japanese tendency to not immediately abandon things when other nations would has been a source of immense joy throughout my life. Pilot makes cheap ballpoints, but they also still make possibly the world's best fountain pens, and I'm glad they do.
Philip Taylor said,
December 6, 2025 @ 7:25 am
"Pilot makes cheap ballpoints, but they also still make possibly the world's best fountain pens" — in which case, let us sincerely hope that Cardfactory's products are not exported to Japan :
Chris Button said,
December 6, 2025 @ 7:44 am
How much "iikae" (rephrasing) of "kango" (Sino-Japanese words) into "wago" (native Japanese words) is done when presenting a formal piece of writing orally for rapid comprehension?
The dedicated existence of NHK's "News Web Easy" suggests that this kango-wago relationship perhaps takes the definition of "easy words" in Japanese beyond just a children's/learner language definition in other languages.
Scott P. said,
December 6, 2025 @ 10:58 am
If two Japanese people can understand each other when speaking (which they with few exceptions obviously can), it stands to reason that if they wrote more or less as they speak, then they could write everything in hiragana, katakana, or romanization and be understood. Kanji are not needed. The same logic applies to Chinese.
In speaking, there are more contextual clues to disambiguate words than there is in a written text. I've also heard that gestures are frequently used, with some speakers even miming writing a particular character to clarify a homophone.
Peter Grubtal said,
December 6, 2025 @ 11:26 am
Josh R. 's comment in the previous thread:
https://languagelog.ldc.upenn.edu/nll/?p=71720
can't be beaten, in my opinion
Stephen Goranson said,
December 6, 2025 @ 11:54 am
Even I, largely ignorant on such, have seen Chinese, using a finger on a palm to convey the character.
JMGN said,
December 6, 2025 @ 12:05 pm
I feel the real question should be "why sould they not anymore?"
Jon Forrest said,
December 6, 2025 @ 12:27 pm
I managed to offend a very well educated native speaker of Chinese when I put forth the proposition that characters weren't linguistically necessary in Chinese. At the time I was careful to agree that the characters had major non-linguistic value but it wasn't enough.
David Morris said,
December 6, 2025 @ 4:02 pm
Korean high school students were learning hanja when I taught English there in 2008-9, and a quick search suggests that they still are.
David Marjanović said,
December 6, 2025 @ 4:08 pm
Direct link to the comment, and I agree.
David Morris said,
December 6, 2025 @ 4:19 pm
The most common use of Chinese/Japanese writing I saw in Korea is at train stations. The bigger signs have Korean, English, Chinese and Japanese.
Victor Mair said,
December 6, 2025 @ 4:32 pm
@David Marjanović:
Peter Grubtal already gave us the direct link to Josh R.'s comment.
I wrote my undergraduate thesis on Middle English, and I would not say that the English-speaking world uses 15th-century spelling conventions.
The logic of you three doesn't explain why Japanese partially cling to kanji while Koreans and Vietnamese abandoned them. Julesy's nuanced, reasoned, and documented explanation makes far more sense.
It's one thing for Josh R. to make his sensational (but hollow) claims, but for PG and DM then simply to repeat them is mòmíngqímiào 莫名其妙 ("baffling / inexplicable / strange / odd / unaccountable").
Philip Taylor said,
December 6, 2025 @ 4:41 pm
Well, let me try to offer a different perspective — why is Shinto the majority religion in Japan while Christianity is the majority religion in most (if not all) of the English-speaking world ?
anon said,
December 6, 2025 @ 4:50 pm
She forgot to mention Wa State in Myanmar they also use Chinese characters as one of the official writing system of the Wa language. They also made Mandarin one of the official languages of the state.
Stephen Goranson said,
December 6, 2025 @ 5:52 pm
That Julsey "forgot" Myanmar Wa usage might be a presumption, given the complexity of what she already did take on.
Victor Mair said,
December 6, 2025 @ 6:57 pm
Indeed!
Chris Button said,
December 6, 2025 @ 6:33 pm
I don't think this is correct.
Wa State straddles Myanmar and China, but neither Chinese nor Burmese scripts are used to write the language.
Chris Button said,
December 6, 2025 @ 6:43 pm
I also highly doubt this to be true–at least from the perspective of Wa State. But, unlike the script, I can't say that with confidence.
~flow said,
December 7, 2025 @ 6:15 am
> The fact that the Japanese still have not abandoned the archaic morphosyllabic / logographic script is a conundrum that has puzzled me since I first learned Chinese and Japanese more than half a century ago.
FWIW I've been dabbling in Ancient Egyptian recently and it's surprising how many principles of writing that we encounter the world over have already been present at this earliest stage—one of the earliest stages of writing that we know of, together with Mesopotamian cuneiform writing systems. They had pictographs, of course, but they also had highly conventionalized classifiers, not unlike what you see in Chinese where e.g. the Silk radical is used to write many morphemes that have ceased to be directly intertwined [hah] with silk as such. They also had phonetic writing, mostly in the form of signs that signify one to three consonants and where the pictorial content becomes irrelevant. As can be gleaned from the Rosetta stone in the names of Ptolemy and Cleopatra, they had developed ways to write vowels, too, although in a somewhat clumsy way, and mostly used only for foreign proper names and loans, but present already a thousand years before the Rosetta stone was engraved. Last but not least, there's a recurrence, a standardization in writing over the millennia that is astounding in its time depth. And it's also an enigmatic script, for example Egyptologists are still trying to figure out the proper reconstruction of the name Osiris which is, most of the time, written as an eye above a throne which tells you next to nothing how it was read.
In all, I find some aspects of Ancient Egyptian culture and script surprisingly modern, despite its age.
FWIW some Korean texts definitely do profit from or even necessitate the use of additional Hanja, I've frequently seen them used in academic papers. Other than that, at least around the early 2000s the characters 南 and 北 would be frequently seen in newspaper headlines. And why not? These are easily learned symbols with 'obvious' (well, for the ones who learned them, but that goes for 100% of all writing) meanings and unambiguous readings. A few hundred of these would certainly go a long way to liven up most modern Korean texts that are rife with Sino-Korean loans. There's a reason these newspapers used to use Hanja which are, at this point, a 100% opt-in.
*Edit* so I went through some South Korean newspaper homepages, and here are some samples of what I found on https://www.chosun.com and https://www.donga.com ; as such, this is an arbitrary selection but it seems to tell me that some newspapers still think that some Hanja are approved by their audience not unlike they expect the occasional sprinkle of Latin letters, mostly in abbreviations like AI and F-15. Just imagine a Western newspaper using one or two Hanja / Kanji / Hanzi in a headline, and that *without* transcription:
* 조슈아 반, 판토자 누르고 '亞 남성 최초' UFC 챔피언
* 美 "한국 돈으로 원전부터 건설" 우리는 '감원전'한다니
* 中 전투기 '오키나와 도발' 日 전투기에 레이더 발사
* 李대통령-손정의 AI 협력
* 韓 ‘ARM 스쿨’ 이달내 윤곽… “반도체 설계역량 강화할 자산”
* “AI인재 연봉 프리미엄, 한국 6%-美는 25%”
* 中전투기 ‘의도적 도발’…日자위대 F-15에 2차례 레이더 쐈다
* 女 중고생, 주말엔 하루 7시간 스마트폰
* ‘현지 누나’ 파장…野 “V0의 국정농단” vs 與 “김건희와 달라”
* 전기차 격전지 되는 韓…외산 공세 속 방어책 찾는 현대차그룹[자동차팀의 비즈워…
* “유럽, 20년내 알아볼 수 없게 될것” 美 진단에 유럽 발칵
Victor Mair said,
December 7, 2025 @ 8:52 am
We have residue of Greek (esp. philosophy, science), Latin (lots and lots of it, esp. law, medicine), Italian (music, cooking), French (about 60% of everything), etc., etc.
Recently, also with the help of Julesy, we dissected the Chinese typewriter. In the past, with the assistance of my capable colleagues in the Penn Museum, I have dabbled in typewriting and typesetting Sumerian, Egyptian, Akkadian, and other ancient languages. If I find time, I will dive more deeply into them in the coming years.
TB said,
December 7, 2025 @ 4:09 pm
From my (very limited) exposure to Hittite, it seems that there is a clear parallel to Japanese mixed script in the way Akkadograms and Sumerograms are used to write many words, even though those words could hypothetically be written syllabically, and Hittite grammatical suffixes written syllabically sometimes follow words written as Akkadograms or Sumerograms. To my understanding, its development follows a period of writing in Akkadian similar to how the adoption of the Japanese script follows a period of writing in Chinese. I think Akkadian itself may provide a similar parallel wrt. Sumerian by virtue of containing Sumerograms, a system which Hittite inherited. I don't have any larger point to make here, I just think it's an interesting comparison!
Josh R. said,
December 7, 2025 @ 7:24 pm
This thread has been quite a journey for me.
*Sees post* "Ah, Julesy's video! This was good. A nice balanced view of the history of the issue."
*Sees Peter Grubtal's comment* "Oh, that's nice. That comment got no responses previously, so it's nice that it was appreciated."
*Sees David Marjanović's comment* "Oh, a direct link to the comment. That was considerate."
*Sees Victor Mair's comment* "Ack! Oof! That was surprisingly venomous! All warm feelings now gone…"
It should not need to be said (but I'll say it anyway), but the previous comment was of course not meant to be some "explanation" of why the Japanese stick with kanji. There's no one explanation to be had here. As Julesy points out, there are a variety of factors, historical and cultural, for why Japanese have not made the move to a purely phonetic orthography, be that roman letters or kanji-derived syllabary, at any number of possible inflection points. (The best opportunity was certainly the Meiji-era drive towards modernization, when many heretofore traditional practices were revised or abandoned, and there was widespread societal motivation to do so.)
My comment refers only to putative arguments that Japanese should abandon kanji *now*. Would it be more efficient and easier to learn and use? Probably! But so what? English would be more efficient and easier to learn and use if its spelling was modernized (Prof. Mair's disingenuous nitpicking notwithstanding), but there is hardly any serious movement to do so. Rather, the pedants are a stronger force against lay reforms such as "thru" or "ur". (Personally, I think we should just move to the objectively superior Shavian alphabet, but good luck getting that to happen!)
And therein lies the primary point of the previous post: to reframe the question in a way that an English speaker can relate to. The question of the utility of kanji in the Japanese language is so often posed from a position of incredulity, as if it's a question of Japanese recalcitrance rather than one of *heavy* sociological inertia. It's almost as if it's that meme: "Why don't the Japanese just use a purely phonetic system? Are they stupid?" But to the Japanese, it's not even a question. They write this way because they've been doing so for centuries, it gives them access to older texts, and they get to do fun things with it like use furigana for layered meanings. It's working fine for them, so why change?
Victor Mair said,
December 8, 2025 @ 7:51 am
They are changing.
Chris Button said,
December 7, 2025 @ 9:11 pm
@ ~flow
It was interesting, albeit not surprising, to read in Gari Ledyard's book about how much resistance there was to the promulgation of Hangul in Korea.
It's that same resistance that I believe prevented China from adopting an "alphabet" of sorts many hundreds of years before.
As I've posted on LLog before, I believe the ganzhi represented that "alphabet" accordingly:
甲, 乙, 丙, 丁, 戊, 己, 庚, 辛, 壬, 癸, 子, 丑, 寅, 卯, 辰, 巳, 午, 未, 申, 酉, 戌, 亥
k, Ɂ, p, t, b, ɣ, ᵏl, s, n, q, ʦ, x, l, ʁ, d, ʣ, ŋ, m, ɬ, r, χ, g
The broadly similar structure of Egyptian hieroglyphs and Chinese oracle-bones is striking. And that knowledge would surely have helped Champollion back in the day with his decipherment efforts!
Without presumably the same resistance in China that occurred in Korea all those centuries later, historians today might perhaps be talking about the "ganzhi" alphabet as the Chinese counterpart to the Phoenician alphabet …
Philip Taylor said,
December 8, 2025 @ 7:40 am
Josh — as a fully-certified pedant, I have no problem at all with "through" being abbreviated to "thru’" (the trailing apostrophe being required, of course) but may I ask what is "ur" meant to convey ? At first sight I can imagine it being used to convey "you’re" ("you are") or "your" (possessive), but that would require that the context make it abundantly clear that it was not intended to denote "ur-" as in "original, earliest, or primitive" or the ISO language code for Urdu.
Andreas Johansson said,
December 8, 2025 @ 10:07 am
"ur" used to be ubiquitous for "your" and "you're" in informal online English. I suspect the omnipresence of autocorrect has reduced the incidence now.
I don't believe I've ever seen a case were it could possibly be confused with the prefix or the language code.
wgj said,
December 8, 2025 @ 10:58 am
In somewhat related news (may deserve its own post), the upcoming Unicode version 18 will add 11328 seal script characters – from when the First Emperor of Qin "unified the writing":
https://www.unicode.org/L2/L2025/25226.htm#185-C3
Bob Ladd said,
December 8, 2025 @ 1:24 pm
On the main question in this thread: Nobody has mentioned the "sunk cost" problem with inefficient technologies that require significant effort to master. Once you're literate in Japanese, you're literate in Japanese, and why should you throw away all those hours you spent becoming literate just to become literate all over again? The same kind of thinking (or rather, the same kind of instinctive reaction) helps explain why we still have the QWERTY keyboard, why the US clings to traditional English weights and measures, and, of course, why English orthography isn't going away any time soon.
On a specific issue, namely kanji/hanzi and homophones: I've heard it suggested – I'm not sure by whom – that both in English and (especially) in French the traditional orthography often has the effect of disambiguating homophones (English two/too/to, French cent/sans/sang/sent etc.), and that this is one of the pressures against spelling reform. Even languages with nice regular surface-y orthographies often use diacritics to distinguish common homophones, especially function words (e.g. Greek, Spanish, Italian). Does anyone know of any research on whether some languages have more homophones than others and/or on how different orthographies deal with troublesome homophones? The simple idea that "if you write more or less as you speak, you could write everything phonemically and be understood" is surely hard to reconcile with modern research on speech processing or even with any of Mark Liberman's demos of ordinary spontaneous pronunciations of ordinary words.
Bob Ladd said,
December 8, 2025 @ 1:34 pm
I didn't mean to suggest that Greek has a "nice regular surface-y orthography" – it's true that going from spelling to sound is almost completely regular, but sound to spelling certainly is not. However, Greek is like Spanish and Italian in using diacritics to distinguish several pairs of monosyllabic homophones, e.g. ή 'or' vs. η 'the (fem. nom.)
Jerry Packard said,
December 8, 2025 @ 4:06 pm
@Jon Forrest I managed to offend a very well educated native speaker of Chinese when I put forth the proposition that characters weren't linguistically necessary in Chinese. At the time I was careful to agree that the characters had major non-linguistic value
He was insulted because saying they aren’t linguistically necessary is not the same as saying they do not have linguistic value, which your statement implies. Chinese characters add a good deal of value to the process of reading, and implying that they don’t is likely to insult an educated native reader of Chinese
wgj said,
December 8, 2025 @ 11:06 pm
I've heard and read many historians marveling at what a tremendous invention writing has been for human civilizations, allowing people to "travel through time" and "communicate directly to other long dead or yet unborn". Linguists, on the other hand, have seldom expressed such poetic admiration – perhaps they're missing the forest for all the trees.
If we consider writing as an instrument not just for immediate consumption of information, but its preservation and long-term storage, the perspectives changes quite dramatically. As noted in my comment above, the upcoming Unicode standard is adding 11K seal script characters from the Qin Dynasty (3rd century BCE). What does that have to do with our topic at hand? Well …
One of the most prominent pieces of writing we have from the Qin period is the Mount Tai stele, author by Emperor Qin upon his visit to the holy mountain, and penned (or more precisely, brushed) by his prime minister Li Si. The most astonishing thing about this relic, in my opinion, is that an average Chinese person today, without any specialist traning, can read a significant portion of it – something like half of the characters, and extract attain some (albeit incomplete) understanding of the text.
And thus we arrive at the core quality of sinographic writing that has been neglected from most of the many, many discussions on this subject on this blog: How many other writing systems have allowed an literate, but otherwise untrained person to read texts that are more than two millennia old? How about one millennium? Or even half a millennium?
Sinographic writing is harder to learn than alphabetic writing – that's hard to argue with. How much harder? Different people would quantify it differently, but I would say maybe 30% harder. It should also be noted that throughout Chinese history, the literacy rate was consistently among (if not straight up) the highest in the world – so the increase in learning difficulties, on a societal scale, is not an overwelming burden empirically.
But imagine you go back in time to an ancient king, or priest, or master scholar, who was about to introduce a writing system to their people, and give them two choices: an alphabetic writing that is comparatively easy to learn, but whose readability degrades rapidly after a few centuries; or an ideographic writing that is 30% harder to learn, but has a readability-halflife of several millennia. I'm quite confident that a person who gets to decide on such matters is more likely to choose the latter over the former.
Mike Ryan said,
December 9, 2025 @ 10:09 pm
Mr. Neil Kubler wrote: "If two Japanese people can understand each other when speaking (which they with few exceptions obviously can), it stands to reason that if they wrote more or less as they speak, then they could write everything in hiragana, katakana, or romanization and be understood. Kanji are not needed. The same logic applies to Chinese."
It certainly "stands to reason" that Kanji is not needed. But if you transcribed a simple newspaper article into Hiragana and had Japanese people read it, every one of them would stumble badly.
Scott P. said,
December 9, 2025 @ 11:11 pm
Different people would quantify it differently, but I would say maybe 30% harder
I think you're missing at least one zero.
The difference between Spanish and French orthography is closer to 30%.
Mike Ryan said,
December 9, 2025 @ 11:14 pm
Prof Mair replied: "They are changing."
Oh really? Got any there, Prof. other than a three word proclamation?
Mike Ryan said,
December 10, 2025 @ 2:04 am
Oh dear, I left out a key word in my contestation of Prof Mair. Here it is again;
Prof Mair replied: "They are changing."
Oh really? Got any proof there, Prof. other than a three word proclamation?
Victor Mair said,
December 10, 2025 @ 10:50 am
For references, see this lengthy comment by me below.
Gokul Madhavan said,
December 10, 2025 @ 4:10 am
@wgj: I have to concur with Scott P. that you’re underestimating the difficulty of learning the Chinese script by at least one order of magnitude, if not two. Furthermore, I don’t think an “ancient king, or priest, or master scholar” would have shared the same ideas about the tradeoff between script difficulty and half-life. At the risk of over-generalisation: kings may want to carve their words into stone; priests may wish to preserve the form of the language even if the meaning recedes into the background; while scholars might prefer that their ideas survive the ravages time even if they be expressed in a wholly different tongue. What’s more, your classification ignores the needs of yet another person: the bookkeeper, who needs comprehensible ledgers but is unconcerned with preserving things on the scale of millennia.
Gokul Madhavan said,
December 10, 2025 @ 4:16 am
Ah I have a misplaced HTML tag that ended up turning the entire message bold when all I wanted to do was to highlight a couple of names. My apologies to all.
I also have an off-topic question for Prof. Mair about the word “rishika” which he uses for Julesy. Is this meant to be a feminine version of Sanskrit ṛṣi? I tripped on it as I definitely wasn’t expecting to find it in this context (though of course the ṛṣis of old were fascinated by language), and initially read it as “rikisha” (perhaps due to some subconscious Japanese filtering mechanism?).
wgj said,
December 10, 2025 @ 6:04 am
I think some of you are mistaking the pains you yourself have suffered learning to read and write Chinese for a universal experience, forgetting that historically speaking, any non-native speaker is – by design – an atypical learner. Until seeing large-sampled, well-done studies that indicate otherwise, I cannot fathom that Chinese writing is twice as hard (or even harder) to learn for a school-age native speaker than English writing – either today or a thousand years ago. As I already argued, the literacy rate in ancient China vs. rest of the world certainly doesn't support the hypothesis of a dramatic differential in learnability.
Jarek Weckwerth said,
December 10, 2025 @ 6:29 am
@wgj some of you are mistaking the pains you yourself have suffered learning to read and write Chinese — It is also a possibility that you are projecting the experience of a person already familiar with the script and finding it usable.
Also, English is absolutely not the best example of an alphabetic writing system.
But even in English, at what age (after stating reading instruction) does a typical child become capable of writing down an arbitrary hitherto-unknown new word? Let us say "palatal" or "interplanetary" or "tibia"? (Which then makes them capable of looking it up in a dictionary or online.) I would guess that his happens within a year from day 0. It very certainly does in more reasonable languages. How long does it take for a Chinese child to learn the corresponding characters?
wgj said,
December 10, 2025 @ 7:25 am
During most of the history of writing, there was no such thing as a dictionary. You want to know how a new word is correctly written, or what its meaning is, you ask your teacher. That is absolutely the same for all writing systems.
English may or may not be a good example for an alphabetic writing system, but it's certainly not a terrible one. The idea that in a "typical" alphabetic system, one symbol (or combination) corresponds to one sound, is like thinking a typical wheel should roll on forever because friction is not significant enough to be included into consideration. For instance, the semitic scripts are among the oldest alphabetic writings in the world, but they never got around to include (all) vowels after all those millennia? By the mere fact that English is the most widely used language in the world today, it is – statistically speaking – representative of alphabetic writing.
Jerry Packard said,
December 10, 2025 @ 9:37 am
While the following does not exactly target some of the questions asked, it does provide a bit of insight into the process. (I hope the formatting survives)
Child Development Volume 74, Issue 1 pp. 27-47
Properties of School Chinese: Implications for Learning to Read
Hua Shu, Xi Chen, Richard C. Anderson, Ningning Wu, Yue Xuan
Abstract
The properties of the 2,570 Chinese characters explicitly taught in Chinese elementary schools were systematically investigated, including types of characters, visual complexity, spatial structure, phonetic regularity and consistency, semantic transparency, independent and bound components, and phonetic and semantic families. Among the findings are that the visual complexity, phonetic regularity, and semantic transparency of the Chinese characters taught in elementary school increase from the early grades to the later grades: Characters introduced in the 1st or 2nd grade typically contain fewer strokes, but are less likely to be regular or transparent, than characters introduced in the 5th or 6th grade. The inverse relation holds when characters are stratified by frequency. Low–frequency characters tend to be visually complex, phonetically regular, and semantically transparent whereas high–frequency characters tend to be the opposite. Combined with other findings, the analysis suggests that written Chinese has a logic that children can understand and use.
Grade 1 2 3 4 5 6 Total
# of characters 436 709 541 358 323 203 2570
% of total 17 28 21 14 13 7 100
Victor Mair said,
December 10, 2025 @ 10:30 am
"What good are kanji?" (10/21/25) — based on a video presentation by "That Japanese Man Yuta" that has these titles: "Why Do Japanese Still Use Kanji? Complicated Writing System…", "Why Do Japanese Use Kanji?", 漢字何故？"
"Character amnesia and kanji attachment" (2/24/16)
"Japanese survey on forgetting how to write kanji" (9/24/12)
"More katakana, fewer kanji" (4/4/16)
"Kana, not kanji, for names" (1/3/21)
"Striving to revive the flagging sinographic cosmopolis" (4/26/25)
"The mathematics of kana vs. kanji usage over time (1879-1968)" (4/28/25) — includes quantitative graphs and incisive comments by J. Marshall Unger; must reading
These, and scores of other Language Log posts written during the past two decades, not to mention hundreds and thousands of other publications, document — with data analysis and linguistic evidence — how radically the Japanese writing system (kanji, hiragana, katakana, rōmaji) has been changing during the past century and more.
Chris Button said,
December 10, 2025 @ 10:39 am
I expect that all non-native/fluent speakers who have tried to learn to read and write in Japanese and in Mandarin would agree on the following: Japanese is way harder.
That doesn't necessarily mean kanji don't work well in Japan. But it does necessarily mean foreigners have a tough time becoming literate.
wgj said,
December 10, 2025 @ 11:06 am
@Chris: Interesting, because anecdotally from my own circle of friends and acquaintances, most Chinese who've learned Japanese and Japanese who've learned Chinese would both say the opposite: That Japanese is easier to learn than Chinese as a foreign language, at least to a level that allows one to live and perhaps even work in the other language environment. Things may start to change when going even higher into the domain of (historical) fine literature.
wgj said,
December 10, 2025 @ 11:29 am
@Gokul: Interesting you brought up the bookkeeper. In my initial reading of your comment, I read it as "the person who maintains the books", aka the librarian – but of course you mean "the person who maintains the financial records", aka the accountant.
In the many books, articles, documentaries and podcasts I've read, watched and listened to on the topic of the invention of writing in ancient mesopotamia, the narrator usually emphasize – often with some amazement – that the first use of writing was for accounting purposes, and not some higher, more spiritual tasks. And every time I read and heard that, I had to think to myself: "But not so in ancient China."
Because all the (mainstream scientific) evidence we have show that Chinese was invented first for the purpose of divination, aka communicating with the spiritual realm, on grave matters of state. So right from the start, Chinese writing, just like so many things in China and East Asia more generally, has a top-down objective, whereas Western writing (from China's POV) is more bottom-up.
wgj said,
December 10, 2025 @ 11:40 am
Although the accountant did come on top at a later time in Chinese history, and invented the clerical script that is basically how Chinese has been written ever since (regular script is a mere aesthetical variation on the clerical script, and cursive script has always been reserved for calligraphers and other high intellectuals). Of course, accounting was only part of what the clerks did in ancient Chinese bureaucracy, but it's enough to be comparable.
Jarek Weckwerth said,
December 10, 2025 @ 1:54 pm
@wgj That was a rather curt reply, but I don't think it was a reply to what I had written.
During most of the history of writing, there was no such thing as a dictionary. — The discussion is about the current state and future of kanji. What the situation was in the past is irrelevant. And the existence of dictionaries, at least until recently, was a major boon.
You want to know how a new word is correctly written, or what its meaning is, you ask your teacher. That is absolutely the same for all writing systems. — You mean in the somewhat distant past, right? The present tense is unwarranted.
The idea that in a "typical" alphabetic system, one symbol (or combination) corresponds to one sound, is like thinking a typical wheel should roll on forever because friction is not significant enough to be included into consideration. — You are arguing with yourself here. I never said that that was the case, and a student of linguistics learns that it is not the case around the time of the second lecture in phonology.
My point was quite specifically, and explicitly, that after a short time of instruction in the general principles of a reasonable alphabetic system a user can write most things down with reasonable accuracy. Your one-symbol-per-sound requirement is a straw man and isn't part of that scenario.
That, and perhaps the number of symbols, is what is relevant to the question of whether Chinese writing is twice as hard (or even harder) to learn for a school-age native speaker than English writing.
wgj said,
December 10, 2025 @ 3:10 pm
If your argument is that we're talking about the reasons for using sinographs today, and the past is irrelevant in that question, then I couldn't object more strongly. Pretty much everyone agrees that the main reason for using them today is path dependency, even though the exact reasons for this path dependency are debatable and may vary from one societal group to another. In other words, people keep using sinographs because the (perceived) cost of switching to something else is higher than the benefits. And this cost include emotional attachment, as well as more rational factors. I don't think anybody today, if tasked with inventing a (practical) new writing system, would go for ideographs.
The past matters tremendously. I'd go as far as to suggest that since writing shapes language, and language shapes thought, a writing system creates (or at least tries to create) the culture in which it thrives, the way an organism changes its environment to fit its needs. Like trees in a rain forest creates their own rain clouds, the sinographic writing system has made the Chinese and the Japanese culturally addicted (either metaphorically or literally, everyone can decide for themselves ) to those characters.
Of course, this leads us back to the question of why the Vietnamese and the Koreans, who were (and to some degree, still are) culturally closer to the sinosphere than the Japanese, have been able to shake off this addiction. Here I want to propose a new idea, while fully aware that it could be (or at least be perceived as) culturally chauvinistic and offensive:
Is it possible that the degree of addiction to sinographs (or any already established writing system) is proportional to the width and depth of literary tradition of a society? For this hypothesis to work, one must demonstrate (or assume) that China and Japan have a larger (and more civilizationally significant) body of literature (this of course, is a potentially offensive proposition). To put it even more bluntly, Vietnam and Korea could switch more easily because they had less to loose, in terms of historical writings? Whereas China and Japan have more (or give more value to preserve) writings accumulated from the past?
Bob Ladd said,
December 10, 2025 @ 3:30 pm
I think Jarek Weckwerth has the right approach to trying to quantify the difficulty of learning to read in different systems: how many hours a week, and over how many years of school, do children spend mastering the skills of reading and writing? I don't know about Chinese schooling in any detail, but within the European languages there are clear differences between languages. In languages like Italian, once you get the alphabetic principle, you basically know how to read and write, and much less time is devoted to reading in school than in places where the kids are acquiring literacy in e.g. English or French. (Italian doesn't really even have a word for "spelling", and they regularly use the English word if they need to talk about how a particular word is written.)
Perhaps someone more knowledgeable on this thread could give us an idea of what proportion of Chinese schooling is devoted to learning how to read and write.
Chris Button said,
December 10, 2025 @ 4:33 pm
@ wgj
I was actually solely referring to the difficulty of learning the Japanese writing system over the Chinese (for Mandarin) writing system. I wasn't referring to the overall difficulties of learning Japanese or Mandarin.
Jarek Weckwerth said,
December 10, 2025 @ 5:31 pm
@wgj If your argument is that we're talking about the reasons for using sinographs today, and the past is irrelevant in that question, then I couldn't object more strongly — No, that is not my argument at all. As I pointed out above, I was reacting to your claim that Chinese writing isn't particularly more difficult to learn than English writing. For young people today.
I do agree with most of what you say about what the reasons are for sinographs being still around (albeit I would counsel caution; Josh R. has been told off in no uncertain terms for holding similar views). But that is a different matter.
And the fact that Korea and Vietnam have smaller amounts of text to preserve (and may not have been as fixated on the written tradition) is of course part of the equation.
@ Bob Ladd: I'm almost sure there has been hard quantification of the effort invested into writing instruction in terms of hours; Prof. Mair will have probably cited it in the many posts on the topic. (BTW, for the sake of clarity: I'm not saying that all that effort is pointless or judging it in any way; I'm simply saying there has to be much more of it.)
Jonathan Smith said,
December 10, 2025 @ 8:31 pm
Re: "more time/effort to learn to read Chinese than e.g. English" — surely, but not so easily quantified. If just reading, I suspect wgj is right that the task is not substantially harder for children than e.g. English. And as I have pointed out here before, the *nature of the system* (where one symbol = one syllable) is observed-ably much *more* not less intuitive to beginning readers.
And of course there's a sense in which reality chucks theory out the window: your kid will actually learn to read in school in China. Contrast e.g. the U.S., where you'd damn well better teach them yourself when they're 2. It is (again observed-ably) true, of course, that a 2-year-old reading star in English can do more than the same in Chinese… but the stars aren't the "big back end" of the problem.
Re: "Japanese harder to read than Chinese" — no, much easier. Japanese gives beginning readers entire sentence structures at a glance, not to mention almost all word divisions. Plus there is a practical and (relatively speaking) small limit to Kanji in Japanese as they relate to everyday reading materials. Note these are *good* things about Japanese; people tend to take this the wrong way.
Re: why change in Korea/Vietnam… Julesy's guess still sounds right-est — keener sense of a "Chinese" cultural yoke to be thrown off — but who knows.
Chris Button said,
December 10, 2025 @ 9:45 pm
@ wgj
I should clarify that I don't mean Japanese is necessarily harder to understand when reading, it is just harder to know how to correctly pronounce it if you aren't a native/fluent speaker.
And it's not always just the choice between various Sino-Japanese "on-yomi" options that is challenging. For example 歯医者 haisha "dentist" uses a kun-yomi plus two on-yomi, while its synonym 歯科医 shikai goes with on-yomi for all three.
Jonathan Smith said,
December 10, 2025 @ 11:10 pm
Re: Jerry Packard's data — interesting, and yes I've noted in discussions here that the higher-frequency characters of e.g. written Mandarin do *not* generally have good/any phonetic cues. Luckily this is irrelevant to kids, who remember stuff well… probably lots of times at literally one exposure.
This does however mean that when they run into an unfamiliar character, Chinese-reading kids cannot reliably "sound out" in order to (1) positively identify a word they already know or (2) learn a new word from scratch. Both (1) and (2) via "sounding out" are of course more possible for e.g. English-reading kids.
BUT after a Chinese-reading kid knows oh 1500 characters, which is oh second grade, they can do (1) and (2) EXTREMELY accurately (better by far than the English case) on words written with characters they know… which is tens of thousands a.k.a. most of them. This is itself a kind of syllable-by-syllable "sounding out" — more akin really to the letter-by-letter "sounding out" of English than is single-character guesswork (where yes attempted "spelling pronunciations" can get pretty crazy.)
And whatever the language, kids aren't learning most new words literally while reading.
That is, the bulk of reading for native-speaker learners involves identifying words they already know. (Protip apply this to your foreign language learning.) So re: "歯医者" "歯科医" etc. (with parallel mapping complexities appearing in practically every writing system), the key without which nothing else matters is knowing the word "dentist."
Chris Button said,
December 10, 2025 @ 11:15 pm
Literally, it's sort of like …
歯医者: tooth (ha 歯) doctor (isha 医者)
versus
歯科医: dental (shika 歯科) doc (i 医)
Mike Ryan said,
December 10, 2025 @ 11:25 pm
Vietnam and Korean may have cast of Kanji not because of a calculating, logical reason based on the type of language they use. As was the case in Korean, it may have been that some powerful leader or group just disfavored Kanji. Or any other accident of history. So, the disparity of Kanji use may not have its roots in the charecteristics of the actual language.
The video by the young Chinese woman was informative but she seems to suggest that Japan is less a part of the Sinosphere (or was) than Vietnam and Korea and so it would be easier for them to slough off Kanji. But the Japamese feel extremely attached to Kanji in spite of not feeling much cultural affinity towards China. Upon death people are given Buddhist names which are in Kanji, for example. Another example is the long tradition of Japanese caligraphy–still taught in elementary schools. Most would be loathe to part with Kanji.
Jerry Packard said,
December 11, 2025 @ 9:24 am
@Jonathan
Yes, your observations and musings are correct. As to whether kids remember stuff well – in Chinese you get about the same distribution of dyslexic readers as you do anywhere else, and the same distribution of phonetic/sight readers. One of the more interesting facts is that in English, a subject’s phonological awareness is more predictive of reading disability while in Chinese, a subject’s morphological awareness is the stronger predictor. On second thought, one might expect this.
Scott P. said,
December 11, 2025 @ 2:37 pm
Jerry,
Those seem like two of the weaker reasons to preserve kanji. Calligraphy can persist even without widespread literacy in kanji, just as it persists in Western languages that have lost the use of cursive lettering. And surely Buddhist names could be written in katakana?
Philip Taylor said,
December 12, 2025 @ 5:57 am
This is quite possibly a naïve question, Scott, but which Western languages have "lost the use of cursive lettering" ? I ask because although my own (left-handed) script is not cursive, I was virtually unique amonst my classmates in being unable to use the cursive letterforms (or "joined-up writing", as we then knew it).
Chris Button said,
December 12, 2025 @ 6:32 am
I suppose another point is that "furigana" pronunciation aids aren't exclusively for kids.
Personally I love kanji. The main reason I started studying Chinese was because I found the kanji used in Japanese so interesting.
But my personal problem is that I have learned too many languages badly rather than one really well. I suppose that's really helpful when pondering broad questions about linguistics on Language Log, but it isn't so great for day-to-day practical purposes.
Jerry Packard said,
December 13, 2025 @ 9:39 am
@Scott P
Yes, that is certainly true.
KIRINPUTRA said,
December 17, 2025 @ 4:17 am
Pretty shallow video. Not cool how she (too) refers to kanji using Mandarin readings as The Ultimate Readings in all contexts, esp. since she knows Korean.