Chinese Character Etymology: The Six Scripts & 3,000 Years of Evolution

By the HanziWise team · Published 2026-09-26

The six scripts: how characters are constructed

Around 100 AD, the scholar Xu Shen compiled the first great Chinese dictionary, the Shuowen Jiezi, and sorted every character into six construction types — the six scripts (六书 liùshū). Four of them describe how characters are built; the last two describe how they are used. Once you know the four building methods, most characters you meet will make immediate sense.

1. Pictographs (象形 xiàngxíng) — drawing the thing

The oldest characters are simplified drawings. 日 was a circle with a dot — the sun. 月 is a crescent moon. 山 shows three peaks, 水 a flowing river with splashes, 人 a person seen from the side. Over centuries the drawings were straightened and squared off until the picture became nearly invisible — compare the oracle bone forms on those character pages with the modern shapes.

Pictographs feel like the "real" Chinese characters to beginners, but they are a small club: only about 4% of characters are true pictographs. Their importance is outsized, though, because they became the building blocks for everything else — nearly every component you will ever learn (木 tree, 口 mouth, 心 heart, 火 fire, 田 field, 目 eye, 马 horse, 鱼 fish, 鸟 bird) started as a pictograph.

2. Simple ideographs (指事 zhǐshì) — pointing at the idea

Some ideas cannot be drawn directly, so a mark points them out. 一 二 三 are tally marks. 上 (up) puts a stroke above a baseline; 下 (down) puts it below. 本 (root, origin) adds a mark to the base of 木 (tree) — the character literally points at the tree's root. This is the smallest category, but its members are extremely common words.

3. Compound ideographs (会意 huìyì) — ideas added to ideas

Combine two existing characters and their meanings fuse into a new one. 明 (bright) puts the sun 日 next to the moon 月 — the two brightest things in the sky. 林 (forest) doubles the tree 木. 好 (good) joins a woman 女 and a child 子. These characters reward a moment of imagination, and they are among the easiest to remember once someone tells you the story.

4. Phono-semantic compounds (形声 xíngshēng) — the engine of the whole system

Here is the pattern that produces about 80% of all Chinese characters: take one component for the meaning and one for the sound. 妈 (mā, mother) = 女 "woman" (meaning) + 马 mǎ (sound). 河 (hé, river) = 氵"water" + 可 kě. 想 (xiǎng, to think) = 心 "heart" + 相 xiāng. 草 (cǎo, grass) = 艹"grass" + 早 zǎo. The phonetic rarely matches perfectly after a thousand years of sound change, but it usually gets you into the right neighborhood — and it explains instantly why so many unrelated-looking characters share pieces.

This is why the radicals (the meaning components used for dictionary lookup) and the most common phonetics are the two highest-leverage things an intermediate learner can study. Each of our character pages breaks its character down exactly this way — formation type, meaning part, sound part — under "How this character works".

5 and 6. Loan characters and mutually explanatory characters (假借 / 转注)

The last two categories are usage patterns rather than construction methods. Loan characters (假借 jiǎjiè) reuse an existing character purely for its sound to write an unrelated word: 来 originally depicted wheat but was borrowed for the verb "to come", which sounded similar. Mutually explanatory characters (转注 zhuǎnzhù) are near-synonyms that shifted into each other's territory — a category even specialists debate. As a learner you can safely treat both as historical footnotes; the first four types do the real work.

From oracle bones to your pen: the evolution of the script

Oracle bone script (甲骨文) is where the written record begins. During the late Shang dynasty (c. 1250–1050 BC), royal diviners carved questions into turtle plastrons and ox shoulder blades, heated them until they cracked, and read the answers in the cracks. The script was forgotten for three millennia until 1899, when a Beijing scholar recognized ancient writing on "dragon bones" sold by pharmacists — traced back to Anyang, the last Shang capital. Around 4,500 distinct oracle bone graphs are known; roughly a third have been deciphered. Characters like 日, 月, 山, 水, 人 are still recognizable at a glance.

Bronze script (金文) followed, cast into ritual vessels during the Zhou dynasty — rounder, heavier, more ceremonial. Small seal script (小篆) arrived with the Qin unification in 221 BC: the first emperor standardized writing across the empire, ironing regional variants into a single elegant, uniform system. Seal script is the ancestor of every modern character, which is why the Shuowen Jiezi analyzed it — and why the small seal forms on our character pages look so systematically related to today's shapes.

The revolution came with clerical script (隶书) in the Han dynasty: scribes writing fast on bamboo strips flattened the seal script's curves into straight strokes and sharp angles — the moment Chinese writing stopped being drawing and became writing. Regular script (楷书), perfected by the Tang dynasty, fixed that into the balanced standard form still taught in schools today. Simplified characters (简体字), standardized in mainland China in the 1950s–60s, mostly adopted cursive and folk shorthand forms already centuries old.

You can see this whole journey on any of our character pages: the evolution timeline shows the oracle bone, bronze and small seal forms (where they survive) next to the modern regular script, with per-file license credits on the glyph attribution page. Start with 水, 心 or 马 — their 3,000-year transformations are among the most dramatic in the script.

Why this matters for learning

Etymology is not trivia; it is compression. A character you understand ("氵+ 可 = river, sounds like kě") costs a fraction of the memory of a character you memorize as a picture. Knowing the six scripts tells you which kind of understanding to look for: a picture, a pointer, a story, or a meaning-plus-sound pair. Combined with the radical index and daily writing practice, it is the difference between learning 3,000 characters and learning a few hundred components.

Frequently asked questions