I found that many (most?) Chinese characters look like they're composed of more primitive ones, like:

做 = 亻十口夂

你 = 亻尔

吃 = 卩乙

and so on.

I this a real phenomenon or just my European imagination?

If it is not real then why? If it is real, then what is the name of this phenomenon? I know there are "radicals", but I'm not sure all components I see belong to the set of radicals.

  • 1
    A comprehensive title which deals with this is Qiu Xigui’s Chinese Writing. It takes a lengthy book to explain. – dROOOze Jun 18 '18 at 17:22
  • @droooze just "yes" or "no" and if "no", then "why" – Dims Jun 18 '18 at 17:24
  • 1
    It’s not a generic yes or no, it depends on the character you’re looking at. So it’s sometimes a yes and sometimes a no and sometimes an illusion due to millenia of script evolution. – dROOOze Jun 18 '18 at 17:27
  • 1
  • 1
    Moreover, they all do, and here's a huge repo of decompositions github.com/amake/cjk-decomp – Vitaly Osipov Jun 19 '18 at 4:37

recent studies called them components (部件). there're 1300+ basic component:



There are multiple ways to decompose Chinese characters:

Decomposed into the different... Approximate
English analogue
historical character combinations etymology 有 ← 𠂇 + ⺼
components orthography (morphemes) 有 ← 𠂇 + 月
components orthography (letters) 有 ← (一 + 丿) + (二 + 冂)
strokes strokes 有 ← (一 + 丿) + (一 + 一 + 丨 + 𠃌) etc

The "etymology" and "orthography" are often but not always the same, since writing of characters (especially shorthand of combined ones) has evolved a lot over the centuries, the components that make up a modern character may be completely unrelated etymologically to the original characters combined to create it.

E.g. while 犬 looks like 大 plus one stroke, they actually derive from independent symbols which (when simplified) have a similar appearance:

enter image description here enter image description here

This is somewhat analogous to how e.g. woodchuck in English looks like it comes from wood + chuck, but is actually from the Algonquian word wuchak.

  • Characters are not made up of radicals, in any sense of the word. You cannot decompose even the most basic of characters, like 可 or 年, into radicals. – dROOOze Jan 21 at 11:10
  • @dROOOze is 可 not made of the following radicals, though? 口⼀⼅ – 月-- Jan 21 at 13:12
  • I mean, if you want to look at it that way, then 年 is made up of the radicals 丿一一丨一丨. “Radical” decomposition or “orthographic decomposition” is overly complicated and offers very little. – dROOOze Jan 21 at 17:20

It is true, here is an example

This character is composed of 氵 and 目,which represent water and eye respectively, and the character means tear!

In more traditional writings, this would become more obvious. In fact, if you look at ancient writings, you will find out that Chinese origins from actual drawing of stuff around us.


Naïve visual decomposition does not work for a lot of the time, so I’ll just demonstrate some examples.

Firstly, your own:

  • 做 is comprised of 作 and 攵. 乍 was graphically corrupted into 古, but that’s not it’s original form. 作 is in turn decomposable into 亻 and 乍.

  • 你 is decomposable into 亻and 尔.

  • 吃 is decomposable into 口 and 乞. 乞 is not decomposable; the way you did it is equivalent to “decomposing” the Roman letter “d” into “c” and “l”.

Next, some non-decomposable characters that look like they’re decomposable:

  • 龍, an entire picture of a dragon

  • 能, an entire picture of a bear

  • 魚, an entire picture of a fish

  • 它, an entire picture of a snake

  • 气, a picture of streaks of clouds in the sky

Finally, some decomposable characters which look like they’re non-decomposable, or have unexpected decompositions:

  • 之, decomposable into 止 and 一

  • 戍, decomposable into 人 and 戈

  • 喪, decomposable into 桑, 2x口, and 亾

  • Why do you say, that 能 is not decomposable, when it is obviously constists of ㄙ月匕匕 which are known and frequent parts? This is the question! How "non-decomposability" can be justified in such situations? Character genesis is not the same as their shape! Even if 作 historically transformed to 古 it doesn't make them the same character! I agree, that decomposition of d into c and l is strange, but decomposition of i into stick and dot is not, and even decomposition of w into v v is not strange it is even called "double v". – Dims Jun 19 '18 at 6:02
  • @Dims well, anyone's free to decompose characters in any way they like. Why stop at the decomposition that you suggested? We can even go down further and say that everything including ㄙ月匕古 is decomposable into one of 20 or so strokes. I don't see the usefulness of this kind of decomposition though, but you're free to use whatever you feel is most helpful for your purposes. – dROOOze Jun 19 '18 at 6:15
  • The question is not about my freedom, but about problem domain. Of course you can decompose into strokes or even pixels, but parts I mention are better because they are complex constructions repeating in different characters many times. – Dims Jun 19 '18 at 6:21
  • @Dims If you've already defined the problem domain (which seems like graphical decomposition) then there's no further question to be asked. You can write a simple algorithm to just detect graphically connecting regions of a character and process thousands of character images to retrieve a rapid decomposition that way. To understand how the characters work as part of Chinese language usage, however, does not overlap with such a decomposition method, and that would be my preferred problem domain. – dROOOze Jun 19 '18 at 6:26
  • @Dims I think there was someone on here that sells books that, decomposing in the method you're suggesting (see the OP of this "question"). – dROOOze Jun 19 '18 at 6:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.