Kanji order in compound words

I’m at the 20th level and have started to come across word pairs that include the same kanji but in different order, such as 事変 vs 変事, 栄光 vs 光栄, 論理 vs 理論.

There is not much difficulty remembering this vocabulary, but I wonder if there is a pattern to the order kanji appear in compound words in Japanese. For example, in some languages you can assume that in a compound word, the last root would conduct the main meaning of the word and the first would modify it (e.g., you can guess that sunflower is probably a flower that has some sun-like features and not vice versa). Is there something similar in Japanese, or is kanji order absolutely incidental?

1 Like

That’s essentially how it works in Japanese, yeah: things come before the things they describe, whether it’s kanji compounds or descriptions in sentences. Sometimes it’s not as obvious, especially with more abstract concepts, but it’s a good rule of thumb!

(Just for funsies, take 生物学者(せいぶつがくしゃ) as an example: 生 modifies 物 to give you living things, both modify 学 to tell you the kind of study, and it all modifies 者 to tell you what the person studies)


Since most of these come from Chinese, that’s the language that determined it, rather than Japanese.

There are like 5 or 6 different categories of how compounds were conceived, and yes, and they align with Chinese grammatical concepts of putting the components together.

@Jonapedia can probably explain it best.


Thanks for the tag. Just, um…

Do you have a source for this? Or do you remember where you saw it? Because it makes some sense when I think about it, but I don’t think I can come up with 5 categories instantly. I’ve just got ‘noun + noun’ and ‘verb + noun’ or ‘noun/adjective + verb’ in my head for now. I usually think of 六書 when I hear ‘5-6’ and ‘kanji’, but those are kanji creation principles (which I don’t know all by heart anyway). It would be interesting to read more about the categories if someone else has already done such analysis. :slight_smile:

So, the answer is, at least in my experience… yeah, usually that’s exactly how it works. What comes first modifies what comes after. Why? Because that’s how modification works in most cases in both Chinese and Japanese. Both languages place their relative clauses before nouns, adjectives before nouns, adverbs before verbs etc. Are there exceptions? Definitely, but I can’t think of one now excluding the fact that 会社 and 社会 used to be equivalent. In the case of Chinese, there are also cases of compounds that mean almost the same thing whose grammatical class changes depending on which kanji comes first e.g. 適合 means ‘to suit’ (verb), whereas 合適 means ‘suitable’. Why? I suspect it’s because 合 is more common as a verb than 適.

Here’s the brief version of 栄光 vs 光栄:

And the full breakdown I tried doing:

The irony of it all is this: my Chinese dictionary tells me that the two are equivalent in Chinese, just that one is more common in literary works. It also only lists 光栄 as a noun. However, online definitions are generally a bit looser, and usage examples, including one from the famous Ancient Chinese poet 李白, show that Chinese speakers also use 栄光 as a noun. I guess no one dictionary really has all the answers.

I’ve already covered 論理 and 理論 somewhere else on the forums:

事変 and 変事 are a new pair for me, though the first word seems to exist in Chinese. In Chinese, 事変 seems to have kept its literal meaning: it refers to changes (変) in matters (事) around oneself. However, it can also mean ‘incident’, because a major ‘change in events/matters’ is exactly what an incident is. 変事 apparently also exists in Chinese, but I think it’s a lot rarer. It tends to mean ‘incident’, just like 事変. To be honest, if you look both words up in a Japanese monolingual dictionary, you’ll notice that both have very similar first definitions. It’s just that 事変 has more specific definitions afterwards including ‘riot/unrest’ and ‘an outbreak of armed conflict without a declaration of war’. 変事 seems to be a more general word whose meaning is fairly literal: it’s a 変な事 – a strange matter/event, which is to say, an abnormal/extreme/extraordinary/unexpected occurrence.

I think one example of a Japanese compound in which the first kanji doesn’t clearly modify the second would be something like 子猫(こねこ). At the very least, it doesn’t seem to make much sense at first glance, because a kitten is obvious a 猫の子, and not the other way around. However, well… you could argue that it’s a ‘child cat’, even if that’s a little strange in English. The alternative is to remember that こ can also be written as 小 in compounds, which indicates that it tends to act like a sort of diminutive prefix, in which case the first element still does modify the second. I’ll just say that the habitual order in Chinese is 〜子, like in 女子 or 男子. 子 is often a diminutive suffix in that case in Chinese, but I’m not entirely sure why 女子 and 男子 tend to mean ‘woman’ and ‘man’ in Chinese… perhaps it’s because 子 is being taken in the ‘child=descendant’ sense, and everyone is someone’s child, so there’s no particular diminutive meaning: 子 probably just sticks around so there are two syllables to say instead of one, which is more natural in modern Chinese.


Kanken tests on 5 of them, and I just figured I should refrain from saying there are exactly 5 without certainty.

The 5 on Kanken are






But I imagined there could be some exceptional types that they don’t bother testing on.


OK, so I guess these are the two main sorts that come to mind for me (the second is what I meant when I said ‘verb + noun’). I’ll see if I can think of other compounds that don’t fall into these categories. It’s true that these seem to cover a lot of the common compounds though.

Thanks for the list anyway! I’ve never seen any attempt at classifying compounds like this. Hahaha.


Oh, I thought of one. Expressions that come from proverbs or stories, without any other connection between the concepts.

These are actually tested separately on Kanken pre-1/1

This is one kind of 故事成語


Under the 四字熟語 section, I imagine? I mean, I think there aren’t as many kanji pairs that come from stories or proverbs as there are four-character compounds. Then again… 矛盾 is probably a pretty famous one, and the kanji for さすが(流石)are another good example… well spotted!

The other category I was thinking of is the case of suffixes that have almost no meaning, e.g. 子 in 女子 and 男子 when they’re used as classifiers for things like schools, sports or clubs. However, this sort of diminutive/euphonic usage is much more common in Chinese than in Japanese. (E.g. there’s this Chinese manga I occasionally check on, and in one chapter, a woman calls dogs 狗子 when the word 狗 would suffice. It does have a bit of an affectionate, diminutive feel, but I also think it’s just an example of wanting to have two syllables to say instead of one, which is much more natural in Mandarin today, because almost all our words contain two hanzi instead of one, common basic words aside. It fits the established rhythm of the language a little better, I suppose?)

1 Like

There’s a general 故事成語 section on pre-1 and 1, separate from yojijukugo. Any type of 故事成語 that isn’t a yojijukugo can be tested there, from compounds to longer expressions.

1 Like

Sounds like an interesting section (and a difficult one if you don’t know enough expressions).

By the way, I just thought of another sort, but I’m not sure if it’s a special case of this category

because I’ve heard someone call these features of Chinese grammar ‘complements’ before: there are compounds like 追出 and 補完 in which the second kanji expresses the result, direction or extent of the action expressed by the first kanji. They’re probably not as common ‘verb + object/target’ compounds in Japanese, and they’re analogous to stuff like ていく、てくる and てくれる in Japanese in the sense that the second kanji is usually acting as an auxiliary verb. They’re ‘verb+verb’ compounds or ‘verb+adverb’ compounds if you ask me, whereas most compounds in that Kanken category are probably ‘verb+noun’ compounds.


I think I added “baby cat” as a synonym. To me it just makes sense as a diminutive prefix.


The kanji on the left are modifying the kanji on the right. Technically each kanji is modifying the one to it’s immediate right, in practice some patterns of kanji are recognized as words.

Thank you for this elaborate explanation as well as links! it did clear things up a little :blush:.

1 Like

The 5 on Kanken are

Thank you, this is certainly a useful reference and something I wanted to research but did not quite know how to look up!

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.