Chinese Character Decomposition — Split Hanzi into Radicals & Components Online
Split hanzi into radicals and components online: structure types, stroke counts, one-click restore. Free, all 6,763 GB2312 characters, runs in your browser.
About this tool
This free Chinese character decomposition tool splits hanzi into their radicals and components: paste a sentence and watch every character break apart into a stream of components; type a single character to see its structure type, radical and stroke count. Three split levels, one-click restore back to the original text, and all 6,763 GB2312 hanzi covered — everything runs locally in your browser. No sign-up, no uploads, works offline.
Decomposition (拆字, chaizi) is how Chinese characters are actually built: most hanzi are compound characters assembled from smaller parts. Understanding those parts is the fastest way for learners to make sense of a character instead of memorizing strokes — and for teachers to explain why a character looks the way it does.
How characters split
- 维 (wéi) is left-right: 纟 + 隹
- 想 (xiǎng) is top-bottom: 相 + 心
- 国 (guó) is a full enclosure: 囗 + 玉
- 湖 (hú) splits into 氵 + 胡, and 胡 splits again into 古 + 月
Every component is itself a writable hanzi or radical. Fun mode decomposes a whole text this way — the classic Chinese word game where readers reassemble the pieces in their head. Structure analysis mode shows each character's structure type, radical and stroke count one by one.
Radical, component, or root?
Three terms you will meet in dictionaries and textbooks:
| Term | Meaning | Example |
|---|---|---|
| Radical (部首) | The indexing component dictionaries file characters under; China's official table lists 201 | 湖 is filed under 氵, 想 under 心 |
| Component (偏旁) | The side-by-side parts of a compound character, now used for parts in any position | 湖 = 氵 + 胡 |
| Component root (字根) | The smallest reusable building blocks, split all the way down | 想 = 木 + 目 + 心 |
A character has exactly one radical but can have several components and roots. Classical dictionaries index by the 214 Kangxi radicals, so any Kangxi radical lookup starts from this field — the tool labels it for every character in analysis mode. The Wubi input method's famous "roots" come from this same idea.
Structure types at a glance
| Structure | IDS | Examples |
|---|---|---|
| Left-right | ⿰ | 好, 湖, 明 |
| Top-bottom | ⿱ | 字, 花, 笔 |
| Left-middle-right | ⿲ | 街, 班 |
| Top-middle-bottom | ⿳ | 品, 晶 |
| Full enclosure | ⿴ | 国, 困 |
| Surround from above | ⿵ | 问, 间 |
| Surround from upper-left | ⿸ | 病, 屋 |
| Surround from lower-right | ⿺ | 赶, 起 |
Analysis mode labels each character automatically. Gentle mode only splits left-right characters so the text stays readable; Full mode attacks every structure above.
Data coverage
Decomposition data is built from the open-source CHISE IDS database (cjkvi-ids): about 5,600 of the 6,763 GB2312 hanzi decompose into components; the rest are single-component characters (人, 大, 心) kept as-is. Radical and stroke data come from the Unicode Unihan database, with radicals shown in their simplified dictionary forms (湖 shows 氵, not 水). Everything is embedded in the page — restore works offline too.
Working with Chinese text? Also try our Simplified–Traditional converter, word counter and Chinese numbers to uppercase converter.
FAQ
What is chaizi? What are the small characters it produces?
Chaizi splits a compound character into its components — radicals, radicals' parts and roots. For example 维 (left-right) splits into 纟 and 隹; 想 (top-bottom) splits into 相 and 心. Each piece is still a hanzi or component that can be written on its own.
What is the difference between Gentle, Full and Partial?
Gentle only splits left-right characters, so the text stays roughly readable; Full splits every decomposable structure (top-bottom, enclosure…) for near-total "cipher text"; Partial splits only every third character, landing in between.
Can I turn the split result back into the original text?
Yes. Click Restore and the component stream is merged back into the original characters. You can also copy the result, paste it back later and restore — the reverse lookup is built into the page.
Are radicals, components and character roots the same thing?
Not exactly. A radical is the indexing component dictionaries use to file characters (201 in China's official Table of Han Radicals); components are the parts you get from splitting; character roots are the smallest reusable building blocks. The three overlap a lot in daily speech, so this tool simply shows structural components.
Which characters are covered? Is my text uploaded?
The built-in dataset covers all 6,763 GB2312 hanzi: 5,611 are decomposable, the rest are single-component characters kept as-is. Everything runs offline in your browser — nothing is uploaded, ever.