Cangjie Decomposition Rules (取碼規則) Explained with Examples

The handful of rules that decide every Cangjie code, including head and body, single-unit characters, enclosures and keeping shapes whole, each shown with verified examples.

Step 1: find the parts in writing order

Every Cangjie code starts with a question: what are this character’s parts, in the order you write them? Chinese characters are written left to right, top to bottom and outside to inside, and Cangjie follows the same order.

In 明 AB日月, 日 comes before 月. In 思 WP田心, 田 sits on top of 心. When parts repeat, each is coded: 品 RRR口口口, 森 DDD木木木.

Step 2: decide whether it is single-unit or compound

A compound character (合體字) divides cleanly into two or more parts, side by side, stacked or enclosed. A single-unit character (獨體字) does not: its strokes cross or interlock so there is no clean division. 馬, 車, 東, 重 and 事 are single-unit; 明, 你 and 說 are compound.

Rule for compound characters: head and body

The first part of a compound character is its head (字首) and everything after it is its body (字身).

• The head keeps at most 2 codes: its first and last.
• The body keeps at most 3 codes: its first, second and last.

This is why the longest possible code is five letters. The rule shows most clearly when both parts are complex. The speech radical 言 is YMMR on its own, but as a head it keeps only YR. The body 兌 is CRHU and keeps CRU:

The same pattern is everywhere. 糸 (VIF) becomes VF as a head, and 巠 keeps its first, second and last codes, M V M, giving 經 VFMVM女火一女一. The body of 謝 YRHHI卜口竹竹戈 is 射 (HHDI), which keeps HHI. In 媽 VSQF女尸手火, the body 馬 (SQSF) keeps SQF.

When the head is a single radical, it contributes one code: 你 ONF人弓火, 打 QMN手一弓, 江 EM水一, 情 PQMB心手一月.

Top-bottom characters

The same head/body logic applies vertically. The top part is the head. In 想 DUP木山心, the head 相 (DBU) keeps first and last, D and U, and the body 心 adds P. In 意 YTAP卜廿日心, the first division is 立 over the rest: 立 gives YT and the body 日心 gives AP. 學 HBND竹月弓木 splits into its top (HB) and 子 (ND).

Rule for single-unit characters

Single-unit characters take their first, second, third and last radicals, up to four codes. Anything in between is dropped.

重 HJWG竹十田土 is written as a sequence of shapes that Cangjie reads as 竹 十 田 土. 馬 SQSF尸手尸火 keeps S Q S from the start and F (灬) from the end.

Enclosures: the frame comes first

For characters where one part surrounds another, the outside is the head. The enclosure 囗 is W in 國 WIRM田戈口一 and 圓 WRBC田口月金; the gate 門 keeps AN in 問 ANR日弓口 and 閱 ANCRU日弓金口山; the frame of 同 BMR月一口 is B.

The walk radicals 辶 and 廴 are coded first, even though they are written last. So 這 YYMR卜卜一口 is 辶 (Y) plus 言 (YMR), and 建 NKLQ弓大中手 is 廴 (NK) plus 聿 (LQ). 道 YTHU卜廿竹山 is Y plus 首 (THU).

Keep shapes whole

Choose the decomposition that uses whole, recognisable radicals rather than splitting a shape into loose strokes. 天 MK一大 is 一 + 大, not 二 + 人, because 大 is a radical in its own right. 夫 QO手人 is Q + O. When two splits are possible, the one with fewer, larger components is usually correct.

Checking your answers

These rules produce the right code for the vast majority of characters, but Cangjie also has conventions for particular shapes that take practice, such as 海 (EOWY) and 電 (MBWU). When your answer doesn’t match, look the character up and compare how the dictionary split it. The mismatch usually teaches you a new auxiliary shape. For the classic troublemakers, see difficult Cangjie characters.

Frequently asked questions

Related guides

← Learn Cangjie