How to use Free Software to learn Japanese, and more.

Kanji composition in relation to reading Japanese

February 23, 2021 — Bundou Makoto

As mentioned briefly in Learning Kanji, kanji are formed from radicals or from other kanji. There are six classes of kanji characters. Altogether they are called 六書.


These six fundamental principles govern how kanji are formed and used.

This includes:

  • Four ways that kanji can be formed: 象形指事会意形声,
  • Two ways in which kanji are used: 転注仮借.


First up on the list is 象形.

しょうけい 【象形】

This is essentially a pictogram, a simple visual representation of the object that the kanji is supposed to represent. For example, the kanji for tree 「木」 visually resembles a tree, the kanji for mountain 「山」 looks like a mountain, the kanji for sun 「日」looks like the Sun.

You might have been told that all kanji originated this way historically. However, in reality, only 12.4% of all kanji are 象形 characters.


Next on the list is 指事.

しじ 【指事】

In contrast to visual representations, these characters symbolize abstract concepts such as numbers. For example, . When you look at the kanji for two 「ニ」 or three 「三」, they consist of two lines or three lines respectively. Similarly, the kanji for up 「上」 and down 「下」 are representations of a line with something above or below it.

Given that one, two, up, and down are not physical objects but rather ideas or concepts, they serve as symbolic representations rather than visual depictions. These characters are relatively rare, they account for only about 0.5% of all kanji.

Compound ideographs

Moving on, we have 会意.

かいい 【会意】

This category involves merging several simple kanji characters. The meaning is a combination of the meanings of the two parts.

For example, combining the characters for mountain 「山」 and rock 「石」 creates the compound character for boulder 「岩」. Similarly, combining two characters for tree 「木」 results in the compound character for thicket「林」. 会意 characters constitute 24.6% of all characters.

Semantic + Phonetic compounds

The remaining category, 形声, constitute 62.5% of all kanji. And if you take all kanji, not just the 常用漢字, it is estimated to be closer to 97%.

けいせい 【形声】

形声 characters are comprised of two different parts. One part which tells you the vague meaning and one part which tells you the pronunciation.

For example, take the character for copper 「銅」. On the left side, we have (かねへん) which is the radical for metal, while on the right side, we have 「」(どう), which is the character which means same.

Understanding why mountain plus rock equals boulder is clear. But why does metal plus same = copper? In reality, it doesn't. The character for same is used in the character for copper only because the on-yomi of that character is どう. And the pronunciation of the word copper is also どう. The left side of the kanji indicates that the meaning has something to do with metal, while the right side indicates that the pronunciation is どう. The original meaning of the right half, same, has no correlation with copper. It is there for its pronunciation value.

This elucidates why most kanji may seem illogical if you solely focus on the meanings of the original radicals. Apparently, these principles would work perfectly in ancient China when kanji first emerged. If you were looking at a 形声 character, and you could isolate which part represented the pronunciation, you could be positive that you would know exactly how to pronounce that character. But because pronunciation changes over time, and that was 3,000 years ago, and we're learning Japanese, not Chinese, these principles don't work perfectly. Nevertheless, they remain relatively reliable if you're trying to guess the pronunciation of a character that you're seeing for the first time.

Besides, if you can notice these patterns as you're trying to memorize the readings of characters, it can make it a lot easier to have them stick in your memory. When you come across a new character, presume that it is one of the 形声 characters because chances are it probably is. Then, try to isolate the part of the character that represents the vague meaning. This shouldn't be too hard because there's a very limited number of these basic radicals like the one for "water" (), or the one for "person" (), or the one for "hand" (). Then it's pretty safe to assume that the rest of the character represents the pronunciation. If you can recall another character you know with the same component, and you know its on-yomi, it's pretty safe to assume it'll be the same.

For example, suppose you are already familiar with the word 構造, meaning structure. Then you encounter the new word 講演, and you've never seen that first kanji before. Notice that the left half of the character is (言偏), which is the radical which represents speech. Now you can infer that the rest of the character () represents its pronunciation. Then you remember that the right part of the character is the same as the right part of the character 構(こう) in 構造.

講 → 冓 → 構

Now you can safely conclude that this new character has some sort of meaning related to speech and has the pronunciation of こう.

How kanji are used

The other two components of the 六書, 転注 and 仮借, deal with how kanji can be used in different ways.

The first is known as 転注. It refers to the process of assigning a new meaning to a kanji character. A kanji character can evolve to convey a different meaning which is similar to its original meaning.


For instance, the character originally meant music. However, because when you listen to music you have a fun time, it also acquired the meaning of "fun". This explains why 音楽 and 楽しい use the same character.

The last principle is known as 仮借.


In ancient China, when a new concept emerged, they needed a new character for that concept. Sometimes, they would adopt an existing character with the same pronunciation as that new concept but a totally different meaning. The existing character would then acquire the new meaning.

For example, the character , which has the on-yomi of , was originally used to denote a kind of weapon. And then, completely independently of that, a first-person pronoun pronounced as was created.

Essentially, even though the first-person pronoun had absolutely no relation to the weapon that was pronounced as , people associated the first-person pronoun with the original weapon character because they were both pronounced .

When you look up a kanji in 広辞苑, it will provide you the etymology of the character. This information includes which of the four types it is. And for 形声 characters it explains which part represents what.

Tags: kanji