Why Did Japan Keep Chinese Characters?

In 1866, Maejima Hisoka proposed abolishing Chinese characters. His argument survives in a text crowded with them.^[1]

There was no necessary contradiction. He was addressing people already educated in the script he wanted future children to escape. His concern was the time it took to acquire that education. Years spent mastering difficult forms, he argued, could be spent learning things. A simpler way of writing would widen access to knowledge.

Maejima was asking a serious question, not making an eccentric complaint about homework. Japan possessed phonetic scripts capable of recording Japanese. Why continue teaching thousands of characters when a much smaller collection of signs could represent its sounds?

The answer is not that Japanese was impossible to write otherwise. It is that writing already did several jobs, for different readers, inside institutions with different needs. Across centuries, people had repeatedly made Chinese characters more usable in Japanese. When reformers called for their removal, they confronted the results of all those earlier changes.

The name kanji means Chinese characters. But the history of Japanese writing cannot be understood as a choice between writing Chinese and inventing something entirely separate.

Characters could be used for their sounds rather than their meanings. This is the principle associated with man'yōgana: writing Japanese sounds with selected Chinese graphs. More than one character could represent the same sound. In that use, a sign's contribution to a sentence might be phonetic even though it possessed a meaning in another context.^[2]

That distinction opened an important possibility. A writing system acquired through Chinese texts could also record words and grammatical forms belonging to Japanese. Borrowing the signs did not require borrowing every sentence into which they were placed.

The two kana scripts developed from this material. Hiragana emerged through cursive forms of whole characters; katakana used selected parts of characters. They were adaptations of the available writing, not alphabets delivered from a completely unrelated source. Linguist Mariko Kuno explains their different early associations: hiragana with Japanese literary composition, katakana with annotation and scholarly reading.^[3]

Their development did not assign kanji an expiry date. A new script could make one task easier while leaving another script useful for other tasks. Someone composing a Japanese poem and someone reading a Chinese Buddhist text were not solving precisely the same problem.

This is the first reason the familiar question can mislead. It imagines a replacement device arriving fully formed beside an obsolete one. What actually developed was a repertoire. Writers could draw upon characters, phonetic signs, and combinations of them. The important decisions concerned where each would work, and for whom.

One of the most consequential adaptations happened around a text rather than inside its main lines.

Kunten are reading marks added to Chinese writing. They can supply phonetic guidance, indicate grammatical elements, and show that words should be taken in a different order. The resulting practice, kundoku, makes it possible to read a Chinese text in a Japanese arrangement. It requires learned conventions, but it does not require every reader to produce a separate, fully rewritten translation.^[4]

The distinction becomes concrete in the Saidai-ji copy of the Golden Light Sutra. Its early Heian annotations preserve evidence of Japanese reading in the first half of the ninth century. The National Institute for Japanese Language and Linguistics treats these annotations as a source for historical Japanese, even though the scripture's principal text is Chinese.^[4]

The page consequently holds two kinds of order. The larger text retains its transmitted sequence; the smaller apparatus tells a trained reader how to approach it. An inherited work can remain recognizable while its route into another language changes.

That arrangement gave continued character learning a practical purpose. The reward was access to particular texts, together with the teaching accumulated around them. Learning the apparatus made a reader part of that interpretive community. It did not make the work effortless, but it helped make the effort reusable.

The small marks matter to the larger question. Kana could assist the continued use of kanji rather than compete with it. An innovation that reduced one obstacle to reading could strengthen the system it seemed, at first glance, capable of replacing.

Nor was annotation simply pronunciation written beside an unfamiliar sign. A modern typographical account of kanbun distinguishes the elements that direct reading, including marks that change the sequence in which the text is taken. The reader must carry out operations on the written arrangement.^[5] Keeping Chinese characters therefore did not mean that Japanese grammar had remained absent from the page. It could be present in the instructions for reading it.

Different uses of writing produced different-looking books.

A fragment catalogued by the Kyoto National Museum makes the point at a very early date. Copied in 859 by a monk of Tōdai-ji, it contains a commentary on the Golden Light Sutra. The catalogue contrasts its compact, cursive writing with the more regular, spacious form customary in sutras copied as acts of religious merit. Here the immediate purpose was study.^[6]

The difference concerns the work a page was expected to do. A carefully presented offering and a copy used for examining an argument could share a religious subject without requiring the same visual treatment. “Buddhist writing” did not prescribe one universal reading experience.

Printed books continued to display variation. In a study of Edo-period printing, Hisada Yukio traces the use of regular-script kanji mixed with hiragana. His published abstract places examples in early eighteenth-century medical books, followed by Buddhist works and other genres. The familiar-looking mixture did not have to wait for modern movable type to become possible.^[7]

This matters because it moves change away from a single dramatic decision to abolish or preserve. A book could retain characters while changing their visual form, their relationship to kana, and the readership it could address. Reform could occur within a genre before becoming a general prescription.

It also warns against treating “easy writing” as one self-evident thing. A simplified sign might be easier to learn from scratch. A familiar form might be quicker for an experienced reader to recognize. A compact page might suit a trained reader following a technical discussion while giving a beginner very little help. These are different advantages, not rival measurements of the same one.

That tension would become central when elementary education entered the argument. A society cannot simply assume that the convenience of its accomplished readers is identical to the convenience of those still learning. But neither can it change the latter without deciding what to do with the books, teachers, and established readers already in existence.

Maejima's proposal distinguished characters from the words written with them. He did not require every expression of Chinese origin to disappear from spoken Japanese. A word could remain while its representation changed. In his examples, familiar expressions are rewritten phonetically.^[1]

This was an important separation. Abolishing a script and abolishing a language's accumulated vocabulary are very different projects. The first changes how words are recorded. The second would demand replacements for things people already say and understand.

He also anticipated the obvious objection: different words can have the same sound. The published text discusses hashi, which can refer to a bridge, chopsticks, or an edge. His answer was that context, grammatical arrangement, and the conventions of the replacement writing could distinguish meanings. Kanji were not the only imaginable means of doing so.^[1]

The argument does not settle every practical difficulty. It does prevent an easy dismissal. Japanese speakers already communicate without displaying characters as they speak. A phonetic writing system would need workable conventions, but homophones alone could not prove it impossible.

For an established reader, however, replacing a familiar distinction could remove information immediately available to the eye. The question was not whether context could ever supply that information; it was how much work readers would be asked to do, and what new habits they would have to learn. For a child, the balance could look different. The visual distinction was useful only after it had been acquired.

Maejima's proposal thus made visible a cost distributed unevenly across generations. The competent reader had already paid much of it. The beginner had not. An argument about national writing was also an argument about whose effort should count most heavily.

Other reformers offered different answers. Nanbu Yoshikazu advocated romanization in 1869, and Nishi Amane argued for writing Japanese with Western letters in 1874. Kana and the Roman alphabet were competing routes to a less character-dependent future, not one unanimously agreed replacement.^[8]

The choice of an alternative therefore opened further questions. Which sounds would it distinguish? How would words be separated? Which written conventions would schools teach? Removing characters would not remove the need for a standard. It would require constructing a different one.

Fukuzawa Yukichi approached the problem at the scale of an actual book. His Moji no Oshie, published in 1873, offered elementary instruction in characters and sentences. Its preface did not celebrate the burden of difficult writing. It argued for reducing it.^[9]

Fukuzawa proposed keeping the stock of generally necessary characters within bounds. Specialists might still require terms particular to their work. Beginners did not need every difficult alternative for an ordinary expression. Instead of waiting for complete abolition, writers could begin making choices that reduced the obstacle now.^[10]

That was more than a middle position between enthusiasm and rejection. It supplied a different unit of reform: the character selected for a lesson, rather than the entire writing system selected for a nation.

The book's opening exercises make that unit visible. Large headwords introduce a person, a horse, going, and coming. Smaller examples recombine them: the person goes; the horse comes. Additional particles change how the pieces belong together. The page teaches a limited stock and then asks it to perform several tasks.^[11]

A child was not simply being presented with a miniature dictionary. The combinations mattered. Recognition of a sign had to lead toward understanding a statement. In the preface, Fukuzawa explicitly separates learning forms and sounds from grasping what a sentence means.^[10]

The teacher, too, had work to do. The instructions distinguish large lesson characters from the smaller connected text and explain how to introduce and reuse what has been learned. Control over the sequence reduces the number of unfamiliar things arriving at once. The retained characters become teachable through selection, repetition, and combination.^[11]

Nothing in the surviving book measures how successfully every child learned from it. Its significance lies in the educational design it actually offers. A critic of excessive character learning could make a practical case for teaching a restricted selection well. Reform did not have to wait until the last objection to abolition had been answered.

This helps explain why retention and criticism could coexist. The continued presence of kanji does not tell us that proposals to simplify writing achieved nothing. Some of their concerns could enter teaching even when their most sweeping solution did not.

The apparent simplicity of a kana alternative becomes more complicated when we ask which kana a school would teach.

Hiragana had long included multiple forms for the same sound. The variants now called hentaigana were not originally a separate collection of decorative oddities. They belonged to the working repertoire of handwriting and print. Some could also help mark distinctions within a running text.^[2]

In 1900, elementary-school regulations selected a much narrower set of forms, generally one for each sound. A study by Takada Tomokazu, Yada Tsutomu, and Saitō Tatsuya shows why this was not simply an order that created uniformity overnight. Variant forms had already been declining in print. The relevant legal restriction was subsequently removed in 1908, without undoing the change in ordinary practice.^[12]

The sequence separates three forces that can easily be confused: what printers were already doing, what schools were instructed to teach, and what habits persisted afterward. A rule could reinforce a developing convention; repeal need not restore the repertoire that readers had ceased to expect.

It also identifies a substantial simplification inside the retained system. A learner's task could become less varied even though kanji remained on the page. The alternatives were not limited to complete abolition or complete immobility.

Scripts also changed jobs. Kuno notes that prewar official documents commonly combined kanji with katakana, and that children once learned katakana before hiragana. The familiar modern allocation of the two phonetic scripts should not be projected unchanged into the past.^[3]

What looks natural after schooling is therefore partly an outcome of schooling. A child does not encounter “Japanese writing” as every form ever used. Teachers, publishers, and educational authorities select a particular route through it. Once widely learned, that route can make an altered system feel like the obvious one.

These cases do not reveal a single moment when Japan examined every alternative and chose kanji forever. They reveal people solving different problems with the materials already available.

A reader of a Chinese scripture could add instructions rather than replace its principal text. A printer could change the appearance of mixed writing. A reformer could distinguish the vocabulary he wanted to retain from the signs he wanted removed. An elementary author could reduce and sequence what a child had to learn. Schools could narrow kana variation without eliminating characters.

Together, these operations explain a form of continuity that is easy to mistake for refusal. The script remained useful partly because its use did not remain fixed. Changes around it could reduce particular difficulties while preserving access to familiar texts and accumulated expertise.

There were costs on both sides. Retention left learners with a substantial task. Replacement would have shifted work toward rewriting, retraining, and establishing new conventions. This is not proof that either arrangement was uniquely efficient. It explains why dissatisfaction did not point everyone toward the same remedy.

Nor does the script's Chinese ancestry decide the matter. By Maejima's time, it belonged to a long history of Japanese reading, teaching, and composition. Removing it would have altered practices developed within Japan as well as its relationship to inherited Chinese learning.

The most revealing answer to Maejima may therefore be Fukuzawa's small exercise. It neither denies the burden nor postpones every improvement until abolition. A few selected signs become several intelligible sentences. The immediate question is what the learner can do next.

Japan kept Chinese characters through a history in which people repeatedly changed the answer to that question. Their survival was not the survival of one untouched method. It was the continued making of routes into a difficult and adaptable written world.