From Oracle Bones to Three Scripts: How Chinese Characters Diverged Across East Asia

hbaristr 7 min read

3,200 years of continuous deployment

The kanji you study tonight were carved into turtle shells under the late Shang dynasty. Anyang divination pits on one end, the postwar reform committees of Tokyo and Beijing on the other. Nothing else in daily production use has a stack trace that long.

Inscribed turtle plastron from the Shang dynasty, c. 1200 BCE, showing oracle bone characters and divination cracks
Tortoise plastron with divination inscription from the reign of King Wu Ding (c. 1200 BCE), National Museum of China. The cracks were produced by applying heat to the reverse; the diviner's verdict ("auspicious") sits at the bottom right. Source: Wikimedia Commons.

Six eras, six phase transitions

Chinese writing did not evolve gradually. It jumped — every jump driven by a new medium, a new ruler, or a clerk who needed to write faster.

Era Script Chinese Approximate Dates Medium / Driver
1 Oracle Bone 甲骨文 c. 1250–1050 BCE Ox scapulae, turtle plastrons; divination
2 Bronze 金文 c. 1100–256 BCE Cast bronze vessels; ritual, legal records
3 Seal 篆書 c. 770 BCE–220 CE Li Si's standardization under Qin Shi Huang (221 BCE)
4 Clerical 隸書 c. 220 BCE–220 CE Bamboo strips, brush and ink; bureaucratic speed
5 Regular 楷書 c. 200 CE–present Paper; the form transmitted to Japan, Korea, Vietnam
6 Simplified 简体字 1956–present PRC reform; also Japan's shinjitai (1946)

The decisive break was seal → clerical. Seal script (篆書) still held onto the curvy, pictographic body of the bronze inscriptions — you can almost see the animal in 馬. Then Qin clerks got tired of drawing curves on bamboo. They flattened the curves into angles and added the horizontal-sweeping 波磔 (bozhe) brush technique. That moment — the 隸變 libian, "clerical change" — is why modern characters look nothing like their pictographic ancestors. The abstraction is older than the Roman Empire. The Wikimedia Commons Ancient Chinese Characters Project maintains SVG files showing characters like 馬, , and across every stage.

Evolution chart of the Chinese character 虎 (tiger) across nine historical script stages from oracle bone to modern
虎 (tiger) across nine stages: early and late Shang oracle bone, four Western Zhou bronze inscriptions, Spring & Autumn bronze, Qin small seal, modern. The crouched cat is unmistakable on bone. By the seal stage it is gone. The libian did this. Source: Wikimedia Commons.

Crossing to Japan: sword, sutra, scholar

Chinese characters reached Japan via the Korean peninsula. The traditional account credits Wani (王仁), a scholar from the Baekje kingdom who — per the Kojiki (712 CE) — arrived around 404 CE with ten volumes of the Analects and one of the Thousand Character Classic. Wani only shows up in Japanese sources, so historians argue. The material record does not. The Inariyama Sword, pulled from a 5th-century burial mound in Saitama, carries 115 gold-inlaid Chinese characters dated to 471 CE. That is the earliest confirmed use of Chinese writing on Japanese soil.

By the 6th century, Buddhist sutras and diplomatic dispatches in Literary Chinese were court reading. But Japanese is polysyllabic and agglutinative — structurally the opposite of monosyllabic, isolating Chinese. A bad fit forces invention. That mismatch produced the most consequential script adaptation in East Asian history.

Man'yōgana: one script becomes three

By the Nara period (710–794), scribes were using certain kanji purely for their sound values, ignoring meaning entirely. The system — man'yōgana (万葉仮名) — drew on up to 970 kanji to spell ~90 Japanese morae. In the 9th century the system split in two: hiragana cursivized out of court correspondence (草書 whole-character forms); katakana was carved out of fragments by Buddhist monks annotating sutras in the margins. Two registers, one root. About 67% of modern kana pairs trace back to the same source kanji.

Mora Hiragana Source Katakana Source
a
i
u
ka
shi
te
ru

The full derivation chart maps all 46 modern morae to their man'yōgana sources.

Three reforms, three destinations

After World War II, three governments reformed the same inherited script on their own clocks. They arrived at three different places.

Japan — shinjitai (新字体, 1946). The Tōyō kanji list simplified 332 of 1,850 official characters. The most extreme cut: 廳 (25 strokes) collapsed to 庁 (5 strokes), shedding 20. But the reform was conservative — roughly 15% of the set was touched, and many of the "new" forms were pre-existing historical variants that already lived in calligraphic practice. The Jōyō kanji list reached 2,136 characters by 2010.

China — jiāntǐzì (简体字, 1956/1964). The PRC simplified 2,236 characters in two phases — roughly 64% of the 3,500 most common forms. The methods were more radical: component substitution, cursive adoption, phonetic replacement. A bigger cut, a sharper break with the past.

Traditional Chinese (繁體字) held on in Taiwan, Hong Kong, and Macau — untouched by either reform. Only ~30% of simplified Chinese forms match their shinjitai equivalents. Same problem, different solutions.

Meaning Traditional Trad. Strokes Japanese JP Strokes Simplified Simp. Strokes
country 11 8 8
study 16 8 8
dragon 16 10 5
drawing 14 7 8
iron 21 13 10
wide 15 5 广 3
old 17 5 5
doctor 18 7 7

Look at 竜 vs. 龙. Japan reached back to a Warring States variant and revived it. China cut strokes more aggressively. Both fled 龍. Neither landed in the same place.

Kokuji: characters Japan invented

Japan minted entirely new characters — kokuji (国字). An estimated 400+ exist; 9 made the Jōyō list. The component logic is the interesting part — it shows how fluent users think about character construction when they have the full machinery to play with.

Kokuji Reading Meaning Components Logic
tōge mountain pass + + mountain where you go up then down
hataraku to work 亻+ a person who moves
hatake dry field + a field cleared by burning
tsuji crossroads 辶 + a road meeting a cross
iwashi sardine 魚 + a weak fish (spoils quickly)
sakaki sacred tree + tree of the gods
sen gland (肉) + 泉 flesh + spring (secretion)
shitsuke discipline 身 + 美 making the body beautiful

Most kokuji carry only kun'yomi — there was no Chinese pronunciation to borrow. A few made the return trip: 腺 (gland) and 働 (work) were absorbed into Chinese and assigned Mandarin readings via the Guobiao standard. Two-way traffic, twelve centuries late.

The wider CJK picture

Japan was not alone. Korea ran on hanja for centuries until Sejong the Great promulgated hangul in 1443, and even coined its own small set of gukja (國字) — including 畓 (답, paddy field). South Korea keeps hanja around for disambiguation; North Korea cut them in 1949.

Vietnam went further. Chữ Nôm (字喃) extended Chinese characters with thousands of Vietnamese-coined forms — far past Japan's ~400 kokuji. The Nom Preservation Foundation catalogs over 25,000 chữ Nôm characters. Then in the early 20th century Vietnam abandoned Han characters entirely for the Latin-based quốc ngữ. The only major CJKV civilization to fully exit the logographic tradition.

Unicode's CJK Unified Ideographs now encode over 97,000 characters across all of it — a digital reconciliation of divergences that began on oracle bones 32 centuries ago. The bureaucrats finally meet in UTF-8.

References

Send feedback

Optional — only if you'd like a reply.