漢文訓読: The Kanbun Layer | Japanese Grammar

For a thousand years, educated Japanese did not read Classical Chinese as Chinese. They read it as Japanese — the same characters, but silently re-ordered and re-inflected on the page by a system of tiny diacritics. That practice is 漢文訓読（かんぶんくんどく）, "reading 漢文 by Japanese gloss," and it is the single most important thing to understand about formal Japanese prose. Because Chinese is a verb-before-object language and Japanese is verb-final, 訓読 physically reverses the word order using 返り点（かえりてん）and adds Japanese endings with 送り仮名（おくりがな）. The by-product was a permanent new register — 漢文訓読体（かんぶんくんどくたい） — that grafted a bundle of Chinese-derived grammar onto Japanese, and that bundle still headlines law, editorials, speeches, and mottos today. This page is the map of the whole layer; the individual signatures each get their own page.

Two languages, opposite word order

Classical Chinese puts the verb before its object: 讀書 is literally "read book." Japanese, being verb-final, wants "book read." So a Japanese reader could not simply pronounce 漢文 top to bottom — the grammar came out backwards. 訓読 solves this by marking the character order with small signs and reading the characters out of their written sequence.

漢文 (written order)	Reordered to Japanese	書き下し文
讀書 (read → book)	book → read	書を読む
登山 (climb → mountain)	mountain → climb	山に登る
不入虎穴 (not → enter → tiger → den)	tiger den → enter → not	虎穴に入らず

書を読む。

sho o yomu

To read a book. (the 漢文讀書 has verb before object; 訓読 flips it to Japanese object-before-verb)

山に登る。

yama ni noboru

To climb a mountain. (登山 read as Japanese: the object 山 moves in front of the verb 登)

Notice the negation, too: Chinese 不 sits before the verb (不入, "not enter"), but Japanese negation is a suffix, so 訓読 sends it to the very end as ず — 入らず. Word order and negation both invert.

The reading marks: 返り点

The signs that tell you to jump backwards are the 返り点. Two do almost all the work, and their logic is worth internalizing because it recurs everywhere.

Mark	Name	What it tells you
レ	レ点（レてん）	Flip with the next character down: read the one below first, then come back up one.
一・二（・三）	一二点（いちにてん）	Jump back over two or more characters: read 一 first, then 二, then 三.
上・中・下	上下点（じょうげてん）	An outer tier, used when a 一二点 group is already nested inside.

A レ点 on a character means "the next one down comes first." In 讀書, a レ点 on 讀 says: read 書 first, then jump back up to 讀 — giving 書を読む.

A 一二点 handles bigger leaps. In 不入虎穴 ("do not enter the tiger's den"), Japanese wants 虎穴 (the object, two characters) read before 入 (the verb). So 入 gets 二 and 穴 gets 一: you read straight down through 虎 and 穴 (the 一 mark), then leap back up to 入 (the 二 mark), and the 不 surfaces as final ず. Together with the 送り仮名 (the Japanese okurigana written in small katakana to the lower-right), the whole thing resolves into a fully grammatical Japanese sentence — the 書き下し文（かきくだしぶん）, the "written-out" transcription in mixed kanji and kana.

虎穴に入らずんば虎子を得ず。

koketsu ni irazunba koji o ezu

If you don't enter the tiger's den, you won't get the tiger's cub — nothing ventured, nothing gained. (a live proverb straight out of 訓読; note the object 虎穴 read before the verb, and 不 landing as final ず)

💡

The umbrella term for all these glosses — 返り点 plus 送り仮名 plus punctuation — is 訓点（くんてん）. You will not add them yourself, but recognizing レ and 一・二 in a scan of an old text tells you instantly: "this is being read backwards into Japanese." That single realization is what turns a wall of characters into a parseable sentence.

The signatures of 漢文訓読体

Here is the payoff, and the reason this is a group of pages rather than one. The reordering machinery pulled a fixed set of Chinese function words into Japanese, each with a standard 訓読 reading. They travel together — where you see one, you tend to see the others — and their presence is the fingerprint of the kanbun register.

Function	Kanji	訓読 reading	Meaning
Comparison	如	〜ごとし	like / as / as follows
Causative	使・令	〜しむ	make / let (someone) do
Imminence	将・且	まさに〜んとす	about to
Correlative negation	未	未だ〜ず	not yet
Prohibitive	—	〜べからず	must not
Alternation	—	〜たり…たり	now… now…

Each is a page of its own: the comparison 〜ごとし, the causative 〜しむ, the correlative 未だ〜ず, and the prohibitive 〜べからず. Read one classical sentence built by 訓読 and you can watch several of them cluster at once:

天の将に大任を是の人に降さんとするや。

ten no masani tainin o kono hito ni kudasan to suru ya

When Heaven is about to lay a great charge on a person… (孟子; 将 read まさに…んとす, the imminence signature — 'is about to')

The character 将 is read twice here — once as the adverb まさに on the way down, once as the auxiliary んとす on the way back up. That double reading is a 再読文字（さいどくもじ）, and the whole family of them (未, 将, 当, 須…) is the correlative machinery covered on the 未だ〜ず page.

A worked line: 論語 opening

The most-quoted 漢文 of all is the first line of the 論語（ろんご, the Analects）. In the original it is 學而時習之. Read it top to bottom as Chinese and 習之 is "practice it" — verb, then object. 訓読 flips 習 and 之 with a single レ点:

学びて時に之を習ふ。

manabite toki ni kore o narau

To learn, and to practice it in due time. (the 論語 opening in 書き下し form; 習之 → 之を習ふ — object before verb, one レ点 flip)

Every 論語 saying is introduced by 子曰 — read 子（し）曰（いわ）く, "the Master said." That 曰く is one of the register's lexical fossils, and it is still alive:

子曰く、過ちて改めざる、是を過ちと謂う。

shi iwaku, ayamachite aratamezaru, kore o ayamachi to iu

The Master said: to err and not correct it — that is what is called an error. (論語; 子曰く heads the quotation, exactly as it does in every citation of the Analects)

いわく付きの物件で、なかなか買い手がつかない。

iwakutsuki no bukken de, nakanaka kaite ga tsukanai

It's a property with a troubled history, so it's slow to find a buyer. (いわく付き = 'with a story attached' — the very same 曰く, now an everyday word)

Another fossil is 蓋（けだ）し, the kanbun adverb for "surely, presumably," which still turns up in editorials and formal prose:

蓋し名言である。

kedashi meigen de aru

It is, indeed, a fine saying. (蓋し = 'truly, surely' — a 漢文訓読 fossil that survives in written commentary)

Where you actually meet it

漢文訓読体 is not confined to the Analects. Its cadence and its auxiliaries were the default for legal statutes, imperial edicts, academic Sino-Japanese prose, war memorials, company mottos, and ceremonial address — and its lexical debris (曰く, 蓋し, 敢えて, 〜に非ず, 〜べからず) is scattered across ordinary formal Japanese. Modern editorials still open speeches with a kanbun-style quotation; four-character mottos on a 座右の銘 are compressed 漢文; a 立入禁止 sign that reads 立入るべからず is pure 訓読 grammar.

💡

The single most useful habit: when the grammar of a formal passage stops decoding — an auxiliary you don't recognize, an order that feels reversed, a negation stranded at the end — switch to kanbun-parsing mode. Ask "is this 訓読?" The telltale auxiliaries (ごとし, しむ, んとす, 未だ…ず, べからず) cluster, so spotting one tells you to expect the rest, and to read the sentence as Chinese wearing Japanese okurigana.

Common mistakes

❌ 漢文はそのまま上から下へ日本語で読める。

kanbun wa sono mama ue kara shita e nihongo de yomeru

Wrong — assuming 漢文 reads straight top-to-bottom in Japanese. It does not: the verb precedes its object, so you must reorder.

✅ 漢文は返り点に従って語順を入れ替えて読む。

kanbun wa kaeriten ni shitagatte gojun o irekaete yomu

漢文 is read by reordering the words according to the 返り点. (the whole point of 訓読 is that Chinese VO becomes Japanese OV)

The commonest beginner error is treating 漢文 as if the characters were already in Japanese order. They are in Chinese order; the diacritics exist precisely because they are not.

❌ レ点は「ここで文が終わる」という印だ。

re-ten wa 'koko de bun ga owaru' to iu shirushi da

Wrong — reading レ点 as an end-of-sentence mark. It is a reordering mark, not punctuation.

✅ レ点は「下の字を先に読み、一字戻れ」という印だ。

re-ten wa 'shita no ji o saki ni yomi, ichiji modore' to iu shirushi da

レ点 means 'read the character below first, then go back up one.' (it directs word order; it never ends a sentence)

レ点 looks like the katakana レ, which tempts learners to read it as a letter. It is a return-mark: it tells you to flip two adjacent characters, nothing more.

❌ 「子曰く」の「曰く」は「言う」の名詞形で「言葉」という意味だ。

'shi iwaku' no 'iwaku' wa 'iu' no meishikei de 'kotoba' to iu imi da

Wrong — glossing 曰く as a noun 'words.' Here it is a verb, 'said.'

✅ 「子曰く」は「先生が言うことには」で、曰く＝「言う」。

'shi iwaku' wa 'sensei ga iu koto ni wa' de, iwaku wa 'iu'

子曰く means 'as for what the Master says' — 曰く is the verb 'says/said.' (the everyday いわく付き preserves the same verb)

曰く is genuinely a verb of speech in 訓読. The modern noun-like いわく ("a hidden story, a reason") is a later development from that verb — but in the Analects it means "said."

❌ 訓読体の助動詞はそれぞれ無関係にばらばらに現れる。

kundokutai no jodōshi wa sorezore mukankei ni barabara ni arawareru

Wrong — assuming ごとし, しむ, 未だ…ず appear at random and unrelated. They are one register's toolkit and cluster together.

✅ ごとし・しむ・未だ〜ずは訓読体の一族で、まとまって現れる。

gotoshi, shimu, imada-zu wa kundokutai no ichizoku de, matomatte arawareru

ごとし, しむ, and 未だ〜ず are one family of the kanbun register and tend to appear together. (recognizing one primes you for the rest)

Key takeaways

漢文訓読 is the practice of reading Classical Chinese as Japanese by reordering it — Chinese verb-before-object becomes Japanese verb-final — using 返り点 and 送り仮名, producing the 書き下し文.
レ点 flips two adjacent characters (read the one below first); 一二点 jumps back over two or more; the 送り仮名 supply the Japanese endings, and Chinese 不 lands as final ず.
The by-product is a whole register, 漢文訓読体, with a recognizable cluster of auxiliaries: 〜ごとし, 〜しむ, まさに〜んとす, 未だ〜ず, 〜べからず, 〜たり…たり.
Lexical fossils survive everywhere: 子曰く, いわく付き, 蓋し, plus every kanbun-style motto and 立入るべからず sign.
The practical skill is mode-switching: when formal grammar resists modern parsing, read it as 訓読 — Chinese syntax wearing Japanese okurigana — and the cluster of signatures decodes it.

Now practice Japanese

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

Start learning Japanese→