Kokuji and Ateji: Japan-Made and Phonetic Kanji | Japanese Grammar

Two comfortable assumptions about kanji are both wrong. The first is that every kanji was imported from China; the second is that every kanji carries a meaning you can read off its parts. Kokuji (国字, "national characters") break the first assumption — they were invented in Japan and often exist nowhere in Chinese. Ateji (当て字, "assigned characters") break the second — they are kanji borrowed purely for their sound, glued onto a word regardless of what the characters mean. Understanding these two categories is what separates a reader who trusts the writing system blindly from one who knows where its edges are.

Kokuji: characters born in Japan

When Japan adopted Chinese characters, it found there were native things and ideas with no Chinese character to write them. So Japanese scribes did the obvious thing: they built new characters out of existing components, following the same logic Chinese used. These home-grown characters are kokuji, also called wasei-kanji (和製漢字, "Japan-made kanji").

Because a kokuji was coined for a native Japanese word, it almost always has only a kun-reading (the native reading) and no on-reading (no Chinese-derived reading) — there was never a Chinese pronunciation to borrow. This is the single most useful diagnostic: a character that stubbornly refuses to give you an on-reading is very often a kokuji.

Kokuji	Reading	Meaning	Built from
峠	とうげ (tōge)	mountain pass	山 mountain + 上 up + 下 down
働	はたら-く (hatara-ku)	to work	亻 person + 動 move
込	こ-む (ko-mu)	to be crowded / put into	辶 movement + 入 enter
畑	はたけ (hatake)	dry field	火 fire + 田 rice paddy
辻	つじ (tsuji)	crossroads	辶 road + 十 cross
凪	なぎ (nagi)	a lull, calm sea	風 wind + 止 stop

Look at how transparent the construction is. 峠（とうげ）is a point where the path goes 上 (up) then 下 (down) on a 山 (mountain) — a pass. 畑（はたけ）is a 田 (paddy) cleared by 火 (fire), which is exactly what a dry field is, as opposed to the water-flooded 田. 凪（なぎ）is when the 風 (wind) 止 (stops). These little pictorial stories are how kokuji were designed, and they are genuinely helpful for remembering them.

この峠を越えると、海が見える。

kono tōge o koeru to, umi ga mieru

Once you get over this pass, you can see the sea.

父は近くの工場で働いている。

chichi wa chikaku no kōjō de hataraite iru

My father works at a nearby factory.

祖父は毎朝、畑に出て野菜を育てている。

sofu wa maiasa, hatake ni dete yasai o sodatete iru

Every morning my grandfather goes out to the field and grows vegetables.

💡

Because most kokuji only have a kun-reading, they tend not to form the tidy two-kanji Sino-Japanese compounds (jukugo) that on-readings produce. When you meet a lone kanji verb like 働く or 込む that reads with native okurigana, suspect a kokuji.

The two-way street: kokuji that Japan exported back

The usual story of kanji is one-directional: China → Japan. But two famous kokuji ran the other way. 働 ("to work"), a character Japan invented by putting 亻 (person) beside 動 (move), was later borrowed into written Chinese, where it appears in the word for "labor." 腺 (せん, sen, "gland") was coined in the Edo period by a Japanese scholar translating Dutch anatomy books, and it too was adopted into Chinese medical writing.

These two are worth remembering because they puncture the idea that the character trade only flowed one way. Japan was not just a passive receiver of Chinese writing — occasionally it manufactured a character good enough that China imported it.

労働時間が長すぎて、体をこわした。

rōdō jikan ga naga-sugite, karada o kowashita

My working hours were so long that I ruined my health.

Notice that in 労働（ろうどう, rōdō, "labor"), 働 is pronounced with the on-reading ドウ (dō) — an on-reading that Japan invented for it by analogy with 動 (also dō). So the "only kun-reading" rule has exceptions: a small number of kokuji, including 働 and 腺, picked up an on-reading over time. The rule is a strong tendency, not a law.

Ateji: kanji chosen for sound, not meaning

Ateji work on the opposite principle. Here the characters are picked for their sound, to spell out a word phonetically, and their meanings are simply ignored. This was Japan's way of writing foreign loanwords (and some native words) in kanji before katakana became the default tool for the job.

The classic example is 珈琲 (コーヒー, kōhī, "coffee"). The two characters mean, respectively, a woman's hairpin ornament and a string of pearls. Neither has the faintest connection to a hot drink. They were chosen in the Edo period solely because, read together, they sound like "kō-hī." If you tried to derive a meaning from "hairpin + pearls," you would get pure nonsense — and that is exactly the trap.

駅前の古い喫茶店の看板には「珈琲」と書いてある。

ekimae no furui kissaten no kanban ni wa 'kōhī' to kaite aru

The sign on the old coffee shop by the station reads 'coffee' (in kanji).

The ateji spelling 珈琲 is (literary / decorative) today — you meet it on old café signs, roasters' packaging, and menus that want a traditional feel, but in ordinary writing everyone uses katakana コーヒー.

Country names were another huge ateji field. 亜米利加 (アメリカ, Amerika, "America") is written with 亜 (a) + 米 (me) + 利 (ri) + 加 (ka). The 米 there means "rice" — but that meaning is irrelevant; it was chosen for the sound "me." As full spellings, these are (archaic) now. But here is the living twist:

新聞では日米関係やアジア情勢がよく取り上げられる。

shinbun de wa nichibei kankei ya ajia jōsei ga yoku toriagerareru

Newspapers often cover Japan–US relations and the situation in Asia.

The 米 of 亜米利加 survives as the standard one-character abbreviation for America: 米国 (べいこく, beikoku) and 日米 (にちべい, nichibei, "Japan–US"). The same happened across the map — 英 (England, from 英吉利), 仏 (France, from 仏蘭西), 独 (Germany, from 独逸), 露 (Russia, from 露西亜). Every one of those newspaper abbreviations is a fossil of a phonetic ateji. So ateji are not a dead museum piece: you read them in a headline every day and never notice.

今夜は近所の回転寿司に食べに行こう。

konya wa kinjo no kaitenzushi ni tabe ni ikō

Let's go eat at the neighborhood conveyor-belt sushi place tonight.

寿司 (すし, sushi) is an ateji too — 寿 means "longevity / congratulations" and 司 means "to administer." The dish was originally written 鮓 or 鮨 (both with the "fish" radical), but 寿司 was chosen partly for the auspicious ring of 寿, and it stuck. Unlike 珈琲, this ateji is fully standard modern spelling. (Note also that 寿司 voices to -zushi inside 回転寿司 — that is rendaku at work.)

💡

A rough sorting rule: if a kanji spelling of a loanword or everyday word feels too "fancy" for the thing it names (hairpins and pearls for coffee?), you are probably looking at ateji chosen for sound. Do not read the parts — read the whole word.

A note on the meaning-based cousin

To be complete: not every non-literal kanji spelling is phonetic. A separate class, jukujikun (熟字訓) or gikun, assigns a native reading to a group of kanji chosen for their meaning, not their sound — 田舎 (いなか, inaka, "countryside"), 大人 (おとな, otona, "adult"), 土産 (みやげ, miyage, "souvenir"). Here you can read a meaning off the characters, but you cannot build the reading from the individual characters' sounds. Ateji (sound-based) and jukujikun (meaning-based) are two mirror-image ways of breaking the normal one-character-one-reading link; this page is about the sound-based ateji, but it is worth knowing the meaning-based twin exists.

How this differs from English

English has nothing quite like kokuji, but it does have a loose analogue to ateji: rebus spelling, where symbols stand for sounds regardless of meaning — think of "gr8" for "great" or a license plate reading "IC U R 2 QT." In both, the reader is meant to hear the sound and ignore the literal symbols. The difference is that Japanese did this systematically, across the whole vocabulary of loanwords, for centuries, and left the results embedded in ordinary spelling.

Kokuji, meanwhile, expose something English speakers rarely think about: that a writing system can be added to by its users. English does not let you invent a new letter of the alphabet. Japanese, with its compositional character logic, did exactly that — coining 峠 and 働 the way an English speaker might coin a compound word.

Common mistakes

❌ 珈琲 = 「かんざしと真珠」の意味だと考える

Incorrect — reading 珈琲 as 'hairpin and pearls.' Ateji have no readable meaning; they spell the sound kōhī.

✅ 珈琲 = コーヒー（音を当てただけ）

kōhī (the characters just spell the sound)

Correct — 珈琲 is ateji for 'coffee'; ignore the characters' meanings.

❌ 峠の音読みを探す

Incorrect — hunting for an on-reading of 峠. As a kokuji it has only the kun-reading tōge.

✅ 峠 = とうげ（訓読みだけ）

tōge (kun-reading only)

Correct — 峠 is a Japan-made kokuji with no Chinese on-reading.

❌ すべての漢字は中国から来たと思い込む

Incorrect — assuming every kanji came from China. Kokuji like 働 and 腺 were invented in Japan.

✅ 働・腺・峠は日本製の国字

hataraku, sen, tōge wa nihon-sei no kokuji

Correct — 働, 腺, and 峠 are Japanese-made kokuji.

❌ 米国の「米」を「rice country」と訳す

Incorrect — reading 米国 as 'rice country.' The 米 is an ateji fragment of 亜米利加 (America).

✅ 米国 = アメリカ

beikoku = Amerika

Correct — 米国 means 'America'; the 米 is borrowed for its sound, not its meaning.

Key takeaways

Kokuji (国字) are kanji invented in Japan; they usually have only a kun-reading — the absence of an on-reading is the tell (峠, 働, 込, 畑, 辻).
A few kokuji (働, 腺) later gained an on-reading and were exported into Chinese — the borrowing ran both ways.
Ateji (当て字) spell a word by sound, ignoring the characters' meanings (珈琲 kōhī, 亜米利加 Amerika, 寿司 sushi). Never read meaning off the parts.
Ateji country-name abbreviations (米, 英, 仏, 独) are still alive in newspaper headlines today.
Most historical phonetic ateji have been replaced by katakana in everyday writing; the kanji forms are (literary / archaic), surviving on signs, packaging, and in set words.

Now practice Japanese

Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.

Start learning Japanese→